1 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
7 (decode_addr_const): Update accordingly.
9 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
10 Alan Hayward <alan.hayward@arm.com>
11 David Sherwood <david.sherwood@arm.com>
13 * tree.h (bit_field_size, bit_field_offset): New functions.
14 * hsa-gen.c (gen_hsa_addr): Use them.
15 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
16 (simplify_vector_constructor): Likewise.
17 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
18 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
19 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
20 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
23 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
24 Alan Hayward <alan.hayward@arm.com>
25 David Sherwood <david.sherwood@arm.com>
27 * expr.h (emit_group_load, emit_group_load_into_temps)
28 (emit_group_store): Take the size as a poly_int64 rather than an int.
29 * expr.c (emit_group_load_1, emit_group_load): Likewise.
30 (emit_group_load_into_temp, emit_group_store): Likewise.
32 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
33 Alan Hayward <alan.hayward@arm.com>
34 David Sherwood <david.sherwood@arm.com>
36 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
37 an unsigned int to a poly_uint64.
38 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
39 sizes as poly_uint64s rather than unsigned ints.
40 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
43 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
44 Alan Hayward <alan.hayward@arm.com>
45 David Sherwood <david.sherwood@arm.com>
47 * lra-constraints.c (emit_inc): Change inc_amount from an int
50 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
51 Alan Hayward <alan.hayward@arm.com>
52 David Sherwood <david.sherwood@arm.com>
54 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
56 (add_stack_var, stack_var_cmp, partition_stack_vars)
57 (dump_stack_var_partition): Update accordingly.
58 (alloc_stack_frame_space): Take the size as a poly_int64 rather
60 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
61 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
62 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
63 rather than a HOST_WIDE_INT.
65 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
66 Alan Hayward <alan.hayward@arm.com>
67 David Sherwood <david.sherwood@arm.com>
69 * target.def (return_pops_args): Treat both the input and output
70 sizes as poly_int64s rather than HOST_WIDE_INTS.
71 * targhooks.h (default_return_pops_args): Update accordingly.
72 * targhooks.c (default_return_pops_args): Likewise.
73 * doc/tm.texi: Regenerate.
74 * emit-rtl.h (incoming_args): Change pops_args, size and
75 outgoing_args_size from int to poly_int64_pod.
76 * function.h (expr_status): Change x_pending_stack_adjust and
77 x_stack_pointer_delta from int to poly_int64.
78 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
79 (ARGS_SIZE_RTX): Update accordingly.
80 * calls.c (highest_outgoing_arg_in_use): Change from int to
82 (stack_usage_watermark, stored_args_watermark): New variables.
83 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
84 (emit_call_1): Change the stack_size and rounded_stack_size
85 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
87 (save_fixed_argument_area): Check stack_usage_watermark.
88 (initialize_argument_information): Change old_pending_adj from
89 a HOST_WIDE_INT * to a poly_int64_pod *.
90 (compute_argument_block_size): Return the size as a poly_int64
92 (finalize_must_preallocate): Track polynomial argument sizes.
93 (compute_argument_addresses): Likewise.
94 (internal_arg_pointer_based_exp): Track polynomial offsets.
95 (mem_overlaps_already_clobbered_arg_p): Rename to...
96 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
97 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
98 Check stored_args_used_watermark.
99 (load_register_parameters): Update accordingly.
100 (check_sibcall_argument_overlap_1): Likewise.
101 (combine_pending_stack_adjustment_and_call): Take the unadjusted
102 args size as a poly_int64 rather than an int. Return a bool
103 indicating whether the optimization was possible and return
104 the new adjustment by reference.
105 (check_sibcall_argument_overlap): Track polynomail argument sizes.
106 Update stored_args_watermark.
107 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
108 (expand_call): Likewise. Maintain stack_usage_watermark and
109 stored_args_watermark. Update calls to
110 combine_pending_stack_adjustment_and_call.
111 (emit_library_call_value_1): Handle polynomial argument sizes.
112 Call stack_region_maybe_used_p and mark_stack_region_used.
113 Maintain stack_usage_watermark.
114 (store_one_arg): Likewise. Update call to
115 mem_overlaps_already_clobbered_arg_p.
116 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
118 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
119 * config/microblaze/microblaze.c (microblaze_function_prologue):
121 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
122 TARGET_RETURN_POPS_ARGS interface.
123 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
125 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
126 * config/i386/i386.c (ix86_return_pops_args): Update for new
127 TARGET_RETURN_POPS_ARGS interface.
128 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
129 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
130 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
131 TARGET_RETURN_POPS_ARGS interface.
132 * config/vax/vax.c (vax_return_pops_args): Likewise.
133 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
134 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
135 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
136 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
137 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
138 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
139 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
140 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
141 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
142 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
143 and x_stack_pointer_delta from int to poly_int64.
144 * dojump.c (do_pending_stack_adjust): Update accordingly.
145 * explow.c (allocate_dynamic_stack_space): Handle polynomial
146 stack_pointer_deltas.
147 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
148 (pad_to_arg_alignment): Track polynomial offsets.
149 (assign_parm_find_stack_rtl): Likewise.
150 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
151 * toplev.c (output_stack_usage): Update reference to
152 current_function_pushed_stack_size.
154 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
155 Alan Hayward <alan.hayward@arm.com>
156 David Sherwood <david.sherwood@arm.com>
158 * function.c (in_arg_offset, var_offset, dynamic_offset)
159 (out_arg_offset, cfa_offset): Change from int to poly_int64.
160 (instantiate_new_reg): Return the new offset as a poly_int64_pod
161 rather than a HOST_WIDE_INT.
162 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
163 (instantiate_virtual_regs_in_insn): Likewise.
165 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
166 Alan Hayward <alan.hayward@arm.com>
167 David Sherwood <david.sherwood@arm.com>
169 * rtl.h (get_args_size, add_args_size_note): New functions.
170 (find_args_size_adjust): Return a poly_int64 rather than a
172 (fixup_args_size_notes): Likewise. Make the same change to the
173 end_args_size parameter.
174 * rtlanal.c (get_args_size, add_args_size_note): New functions.
175 * builtins.c (expand_builtin_trap): Use add_args_size_note.
176 * calls.c (emit_call_1): Likewise.
177 * explow.c (adjust_stack_1): Likewise.
178 * cfgcleanup.c (old_insns_match_p): Update use of
179 find_args_size_adjust.
180 * combine.c (distribute_notes): Track polynomial arg sizes.
181 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
182 end_true_args_size, beg_delay_args_size and end_delay_args_size
183 from HOST_WIDE_INT to poly_int64.
184 (add_cfi_args_size): Take the args_size as a poly_int64 rather
185 than a HOST_WIDE_INT.
186 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
187 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
188 polynomial arg sizes.
189 * emit-rtl.c (try_split): Use get_args_size.
190 * recog.c (peep2_attempt): Likewise.
191 * reload1.c (reload_as_needed): Likewise.
192 * expr.c (find_args_size_adjust): Return the adjustment as a
193 poly_int64 rather than a HOST_WIDE_INT.
194 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
195 to a poly_int64 and change the return type in the same way.
196 (emit_single_push_insn): Track polynomial arg sizes.
198 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
199 Alan Hayward <alan.hayward@arm.com>
200 David Sherwood <david.sherwood@arm.com>
202 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
203 from HOST_WIDE_INT to poly_int64.
204 * expr.c (push_block, emit_push_insn): Likewise.
206 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
207 Alan Hayward <alan.hayward@arm.com>
208 David Sherwood <david.sherwood@arm.com>
210 * function.h (frame_space): Change start and length from HOST_WIDE_INT
212 (get_frame_size): Return the size as a poly_int64 rather than a
214 (frame_offset_overflow): Take the offset as a poly_int64 rather
215 than a HOST_WIDE_INT.
216 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
217 (assign_stack_temp): Likewise for the size.
218 * function.c (get_frame_size): Return a poly_int64 rather than
220 (frame_offset_overflow): Take the offset as a poly_int64 rather
221 than a HOST_WIDE_INT.
222 (try_fit_stack_local): Take the start, length and size as poly_int64s
223 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
224 rather than a HOST_WIDE_INT.
225 (add_frame_space): Take the start and end as poly_int64s rather than
227 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
228 (assign_stack_temp): Likewise for the size.
229 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
231 (find_temp_slot_from_address): Handle polynomial offsets.
232 (combine_temp_slots): Likewise.
233 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
235 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
236 poly_int64 rather than a HOST_WIDE_INT.
237 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
238 than a HOST_WIDE_INT.
239 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
240 polynomial frame offsets.
241 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
242 as a poly_int64 rather than an int.
243 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
244 * config/v850/v850-protos.h (compute_frame_size): Likewise.
245 * config/v850/v850.c (compute_frame_size): Likewise.
246 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
247 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
248 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
249 * config/pa/pa.c (pa_compute_frame_size): Likewise.
250 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
251 rather than a HOST_WIDE_INT.
252 * explow.c (get_dynamic_stack_base): Likewise.
253 * final.c (final_start_function): Use the constant lower bound
254 of the frame size for -Wframe-larger-than.
255 * ira.c (do_reload): Adjust for new get_frame_size return type.
256 * lra.c (lra): Likewise.
257 * reload1.c (reload): Likewise.
258 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
259 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
260 * rtlanal.c (get_initial_register_offset): Return the offset as
261 a poly_int64 rather than a HOST_WIDE_INT.
263 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
264 Alan Hayward <alan.hayward@arm.com>
265 David Sherwood <david.sherwood@arm.com>
267 * reload1.c (elim_table): Change initial_offset, offset and
268 previous_offset from HOST_WIDE_INT to poly_int64_pod.
269 (offsets_at): Change the target array's element type from
270 HOST_WIDE_INT to poly_int64_pod.
271 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
272 (elimination_costs_in_insn, update_eliminable_offsets)
273 (verify_initial_elim_offsets, set_offsets_for_label)
274 (init_eliminable_invariants): Update after above changes.
276 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
277 Alan Hayward <alan.hayward@arm.com>
278 David Sherwood <david.sherwood@arm.com>
280 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
281 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
282 (decomposition): Change start and end from HOST_WIDE_INT
284 (decompose, immune_p): Update accordingly.
285 (find_inc_amount): Return a poly_int64 rather than an int.
286 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
289 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
290 Alan Hayward <alan.hayward@arm.com>
291 David Sherwood <david.sherwood@arm.com>
293 * tree.h (get_inner_reference): Return the bitsize and bitpos
294 as poly_int64_pods rather than HOST_WIDE_INT.
295 * fold-const.h (ptr_difference_const): Return the pointer difference
296 as a poly_int64_pod rather than a HOST_WIDE_INT.
297 * expr.c (get_inner_reference): Return the bitsize and bitpos
298 as poly_int64_pods rather than HOST_WIDE_INT.
299 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
301 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
302 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
303 (optimize_bit_field_compare): Update call to get_inner_reference.
304 (decode_field_reference): Likewise.
305 (fold_unary_loc): Track polynomial offsets and sizes.
306 (split_address_to_core_and_offset): Return the bitpos as a
307 poly_int64_pod rather than a HOST_WIDE_INT.
308 (ptr_difference_const): Likewise for the pointer difference.
309 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
310 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
311 * dbxout.c (dbxout_expand_expr): Likewise.
312 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
313 (loc_list_from_tree_1, fortran_common): Likewise.
314 * gimple-laddress.c (pass_laddress::execute): Likewise.
315 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
316 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
317 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
318 * tree-affine.c (tree_to_aff_combination): Likewise.
319 (get_inner_reference_aff): Likewise.
320 * tree-data-ref.c (split_constant_offset_1): Likewise.
321 (dr_analyze_innermost): Likewise.
322 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
323 * tree-sra.c (ipa_sra_check_caller): Likewise.
324 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
325 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
326 (instrument_bool_enum_load, instrument_object_size): Likewise.
327 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
328 to get_inner_reference.
329 * hsa-gen.c (gen_hsa_addr): Likewise.
330 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
331 * tsan.c (instrument_expr): Likewise.
332 * match.pd: Update call to ptr_difference_const.
334 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
335 Alan Hayward <alan.hayward@arm.com>
336 David Sherwood <david.sherwood@arm.com>
338 * fold-const.c (fold_comparison): Track sizes and offsets as
339 poly_int64s rather than HOST_WIDE_INTs when folding address
342 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
343 Alan Hayward <alan.hayward@arm.com>
344 David Sherwood <david.sherwood@arm.com>
346 * expr.h (get_bit_range): Return the bitstart and bitend as
347 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
348 as a poly_int64 rather than a HOST_WIDE_INT.
349 * expr.c (get_bit_range): Likewise.
350 (expand_assignment): Update call accordingly.
351 * fold-const.c (optimize_bit_field_compare): Likewise.
353 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
354 Alan Hayward <alan.hayward@arm.com>
355 David Sherwood <david.sherwood@arm.com>
357 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
358 (aarch64_legitimate_address_p): Use it instead of an rtx code,
359 as an optional final parameter.
360 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
361 (aarch64_legitimate_address_p): Likewise.
362 (aarch64_print_address_internal): Take an aarch64_addr_query_type
363 instead of an rtx code.
364 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
365 (aarch64_legitimate_address_hook_p): Likewise.
366 (aarch64_print_ldpstp_address): Likewise.
367 (aarch64_print_operand_address): Likewise.
368 (aarch64_address_cost): Likewise.
369 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
370 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
371 (aarch64_mem_pair_lanes_operand): Likewise.
373 2017-12-20 Richard Biener <rguenther@suse.de>
375 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
376 update_stmt_if_modified.
378 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
380 PR tree-optimization/83491
381 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
382 before walking uses. Improve coding style and comments.
384 2017-12-20 Tom de Vries <tom@codesourcery.com>
386 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
388 2017-12-20 Jakub Jelinek <jakub@redhat.com>
391 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
393 (pass_ipa_free_fn_summary): Add small_p private data member,
394 initialize to false in the ctor.
395 (pass_ipa_free_fn_summary::clone,
396 pass_ipa_free_fn_summary::set_pass_param,
397 pass_ipa_free_fn_summary::gate): New methods.
398 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
399 | TODO_dump_symtab if small_p.
400 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
401 entry and add another instance of the pass with false parm after
403 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
404 ipa_free_fn_summary here.
406 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
408 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
410 2017-12-20 Martin Sebor <msebor@redhat.com>
413 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
414 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
415 (expand_builtin_strncmp): Same.
417 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
420 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
421 allowed debug stmts before labels.
422 (expand_gimple_basic_block): Likewise.
423 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
424 * gimple-iterator.h (gsi_after_labels): Likewise.
425 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
426 rename reused variable, and simplify using gsi_move_before.
427 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
428 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
429 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
430 (gimple_verify_flow_info, gimple_block_label): Likewise.
431 (make_blocks): Move debug markers after adjacent labels.
432 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
433 allowed debug insns outside blocks.
434 * df-scan.c (df_insn_delete): Likewise.
435 * lra-constraints.c (update_ebb_live_info): Likewise.
436 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
437 (vt_initialize, delete_vta_debug_insns): Likewise.
438 (reemit_marker_as_note): Drop BB parm. Adjust callers.
440 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
441 Alan Hayward <alan.hayward@arm.com>
442 David Sherwood <david.sherwood@arm.com>
444 * poly-int-types.h (round_down_to_byte_boundary): New macro.
445 (round_up_to_byte_boundary): Likewise.
446 * expr.h (get_bit_range): Add temporary shim.
447 * gimple-ssa-store-merging.c (store_operand_info): Change the
448 bitsize, bitpos, bitregion_start and bitregion_end fields from
449 unsigned HOST_WIDE_INT to poly_uint64.
450 (merged_store_group): Likewise load_align_base.
451 (compatible_load_p, compatible_load_p): Update accordingly.
452 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
453 (split_group, imm_store_chain_info::output_merged_store): Likewise.
454 (mem_valid_for_store_merging): Return the bitsize, bitpos,
455 bitregion_start and bitregion_end as poly_uint64s rather than
456 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
457 (handled_load): Take the bitsize, bitpos,
458 bitregion_start and bitregion_end as poly_uint64s rather than
459 unsigned HOST_WIDE_INTs.
460 (pass_store_merging::process_store): Update call to
461 mem_valid_for_store_merging.
463 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
464 Alan Hayward <alan.hayward@arm.com>
465 David Sherwood <david.sherwood@arm.com>
467 * builtins.c (get_object_alignment_2): Track polynomial offsets
468 and sizes. Update the alignment handling.
470 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
471 Alan Hayward <alan.hayward@arm.com>
472 David Sherwood <david.sherwood@arm.com>
474 * tree.h (get_inner_reference): Add a version that returns the
475 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
476 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
477 the case in which bitpos is not associated with the first byte.
479 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
480 Alan Hayward <alan.hayward@arm.com>
481 David Sherwood <david.sherwood@arm.com>
483 * tree-affine.h (get_inner_reference_aff): Return the size as a
485 * tree-affine.c (get_inner_reference_aff): Likewise.
486 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
487 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
489 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
490 Alan Hayward <alan.hayward@arm.com>
491 David Sherwood <david.sherwood@arm.com>
493 * fold-const.c (pointer_may_wrap_p): Take the offset as a
494 HOST_WIDE_INT rather than a poly_int64.
496 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
497 Alan Hayward <alan.hayward@arm.com>
498 David Sherwood <david.sherwood@arm.com>
500 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
501 HOST_WIDE_INT to poly_int64_pod.
502 (perform_symbolic_merge): Update accordingly.
503 (bswap_replace): Likewise.
505 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
506 Alan Hayward <alan.hayward@arm.com>
507 David Sherwood <david.sherwood@arm.com>
509 * tree-affine.h (aff_tree::offset): Change from widest_int
511 (wide_int_ext_for_comb): Delete.
512 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
513 constants as poly_widest_int rather than widest_int.
514 (aff_combination_constant_multiple_p): Return the multiplier
515 as a poly_widest_int.
516 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
518 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
519 version static and add an overload for poly_widest_int.
520 (aff_combination_const, aff_combination_add_cst)
521 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
522 the constants as poly_widest_int rather than widest_int.
523 (tree_to_aff_combination): Generalize INTEGER_CST case to
525 (aff_combination_to_tree): Track offsets as poly_widest_ints.
526 (aff_combination_add_product, aff_combination_mult): Handle
528 (aff_combination_constant_multiple_p): Return the multiplier
529 as a poly_widest_int.
530 * tree-predcom.c (determine_offset): Return the offset as a
532 (split_data_refs_to_components, suitable_component_p): Update
534 (valid_initializer_p): Update call to
535 aff_combination_constant_multiple_p.
536 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
537 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
538 as a poly_int64 rather than a HOST_WIDE_INT.
539 (get_address_cost): Handle polynomial offsets.
540 (iv_elimination_compare_lt): Likewise.
541 (rewrite_use_nonlinear_expr): Likewise.
543 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
544 Alan Hayward <alan.hayward@arm.com>
545 David Sherwood <david.sherwood@arm.com>
547 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
548 as a poly_int64_pod rather than a HOST_WIDE_INT.
549 (get_addr_base_and_unit_offset): Likewise.
550 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
551 (get_addr_base_and_unit_offset): Likewise.
552 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
553 to poly_int64 in example.
554 * fold-const.c (fold_binary_loc): Update call to
555 get_addr_base_and_unit_offset.
556 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
557 (maybe_canonicalize_mem_ref_addr): Likewise.
558 (gimple_fold_stmt_to_constant_1): Likewise.
559 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
561 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
562 * match.pd: Likewise.
563 * omp-low.c (lower_omp_target): Likewise.
564 * tree-sra.c (build_ref_for_offset): Likewise.
565 (build_debug_ref_for_model): Likewise.
566 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
567 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
568 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
569 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
570 (constant_pointer_difference): Likewise.
571 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
572 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
573 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
574 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
575 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
576 (set_ssa_val_to): Likewise.
577 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
578 (maybe_diag_stxncpy_trunc): Likewise.
579 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
580 * tree.c (build_simple_mem_ref_loc): Likewise.
581 (array_at_struct_end_p): Likewise.
583 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
584 Alan Hayward <alan.hayward@arm.com>
585 David Sherwood <david.sherwood@arm.com>
587 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
588 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
589 (get_ref_base_and_extent_hwi): Declare.
590 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
591 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
592 (get_ref_base_and_extent_hwi): New function.
593 * cfgexpand.c (expand_debug_expr): Update call to
594 get_ref_base_and_extent.
595 * dwarf2out.c (add_var_loc_to_decl): Likewise.
596 * gimple-fold.c (get_base_constructor): Return the offset as a
597 poly_int64_pod rather than a HOST_WIDE_INT.
598 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
599 * ipa-polymorphic-call.c
600 (ipa_polymorphic_call_context::set_by_invariant)
601 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
602 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
603 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
604 rather than get_ref_base_and_extent.
605 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
606 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
607 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
609 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
610 call to get_ref_base_and_extent.
611 * tree-sra.c (create_access, get_access_for_expr): Likewise.
612 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
613 (stmt_kills_ref_p): Likewise.
614 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
615 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
617 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
618 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
619 when calling native_encode_expr.
620 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
621 call to get_ref_base_and_extent.
622 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
623 get_ref_base_and_extent.
624 * var-tracking.c (track_expr_p): Likewise.
626 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
627 Alan Hayward <alan.hayward@arm.com>
628 David Sherwood <david.sherwood@arm.com>
630 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
631 HOST_WIDE_INT to poly_int64_pod.
632 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
633 polynomail parameter offsets.
635 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
636 Alan Hayward <alan.hayward@arm.com>
637 David Sherwood <david.sherwood@arm.com>
639 * gengtype.c (main): Handle poly_int64_pod.
640 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
641 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
642 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
643 from HOST_WIDE_INT to poly_int64_pod.
644 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
645 (copy_cfa): New function.
646 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
647 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
648 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
649 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
650 (initial_return_save): Treat offsets as poly_ints.
651 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
652 offset is nonconstant.
653 (reg_save): Take the offset as a poly_int64. Fall back to
654 DW_CFA_expression for nonconstant offsets.
655 (queue_reg_save): Take the offset as a poly_int64.
656 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
658 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
659 Alan Hayward <alan.hayward@arm.com>
660 David Sherwood <david.sherwood@arm.com>
662 * rtl.h (operand_subword, operand_subword_force): Take the offset
663 as a poly_uint64 an unsigned int.
664 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
666 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
667 Alan Hayward <alan.hayward@arm.com>
668 David Sherwood <david.sherwood@arm.com>
670 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
671 'p' format code. Use INT_LIST rather than SUBREG as the example of
672 a code with an XINT and an XEXP. Remove the implication that
673 accessing an rtx field using XINT is expected to work.
674 * rtl.def (SUBREG): Change format from "ei" to "ep".
675 * rtl.h (rtunion::rt_subreg): New field.
676 (XCSUBREG): New macro.
677 (SUBREG_BYTE): Use it.
678 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
679 Update constructor accordingly.
680 (subreg_shape::operator ==): Update accordingly.
681 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
682 than an unsigned int.
683 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
684 a poly_uint64 rather than an unsigned int.
685 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
686 than an unsigned int.
687 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
688 (subreg_size_highpart_offset): Return a poly_uint64 rather than
689 an unsigned int. Take the sizes as poly_uint64s.
690 (subreg_offset_from_lsb): Return a poly_uint64 rather than
691 an unsigned int. Take the shift as a poly_uint64 rather than
693 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
694 as a poly_uint64 rather than an unsigned int.
695 (simplify_subreg_regno): Likewise.
696 (byte_lowpart_offset): Return the memory offset as a poly_int64
698 (subreg_memory_offset): Likewise. Take the subreg offset as a
699 poly_uint64 rather than an unsigned int.
700 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
701 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
702 poly_uint64 rather than an unsigned int.
703 * rtl.c (rtx_format): Describe 'p' in comment.
704 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
705 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
706 offset as a poly_uint64 rather than an unsigned int.
707 (byte_lowpart_offset): Return the memory offset as a poly_int64
709 (subreg_memory_offset): Likewise. Take the subreg offset as a
710 poly_uint64 rather than an unsigned int.
711 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
712 mode sizes as poly_uint64s rather than unsigned ints. Return a
713 poly_uint64 rather than an unsigned int.
714 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
715 (copy_insn_1): Handle 'p'.
716 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
717 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
718 an unsigned int. Return the shift in the same way.
719 (subreg_lsb): Return the shift as a poly_uint64 rather than an
721 (subreg_size_offset_from_lsb): Take the sizes and shift as
722 poly_uint64s rather than unsigned ints. Return the offset as
724 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
725 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
727 * rtlhash.c (add_rtx): Handle 'p'.
728 * genemit.c (gen_exp): Likewise.
729 * gengenrtl.c (type_from_format, gendef): Likewise.
730 * gensupport.c (subst_pattern_match, get_alternatives_number)
731 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
732 (subst_dup): Likewise.
733 * gengtype.c (adjust_field_rtx_def): Likewise.
734 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
735 (match_pattern_2): Likewise.
736 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
737 (rtx_test::subreg_field): New function.
738 (operator ==, safe_to_hoist_p, transition_parameter_type)
739 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
740 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
741 * genpeep.c (match_rtx): Likewise.
742 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
743 (rtx_writer::print_rtx_operand): Handle 'p'.
744 (print_value): Handle SUBREG.
745 * read-rtl.c (apply_int_iterator): Likewise.
746 (rtx_reader::read_rtx_operand): Handle 'p'.
747 * alias.c (rtx_equal_for_memref_p): Likewise.
748 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
749 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
751 * calls.c (expand_call): Likewise.
752 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
753 (make_extraction, gen_lowpart_for_combine): Likewise.
754 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
756 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
757 rather than an unsigned int. Treat subreg offsets as poly_ints.
758 (exp_equiv_p): Handle 'p'.
759 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
760 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
761 * dse.c (find_shift_sequence): Likewise.
762 * dwarf2out.c (rtl_for_decl_location): Likewise.
763 * expmed.c (extract_low_bits): Likewise.
764 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
765 (expand_expr_real_2): Likewise.
766 * final.c (alter_subreg): Likewise.
767 (leaf_renumber_regs_insn): Handle 'p'.
768 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
769 Treat subreg offsets as poly_ints.
770 * fwprop.c (forward_propagate_and_simplify): Likewise.
771 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
772 * ira.c (get_subreg_tracking_sizes): Likewise.
773 * ira-conflicts.c (go_through_subreg): Likewise.
774 * ira-lives.c (process_single_reg_class_operands): Likewise.
775 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
776 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
777 as a poly_uint64 rather than an unsigned int.
778 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
779 subreg offsets as poly_ints.
780 * lra-constraints.c (operands_match_p): Handle 'p'.
781 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
782 * lra-spills.c (assign_mem_slot): Likewise.
783 * postreload.c (move2add_valid_value_p): Likewise.
784 * recog.c (general_operand, indirect_operand): Likewise.
785 * regcprop.c (copy_value, maybe_mode_change): Likewise.
786 (copyprop_hardreg_forward_1): Likewise.
787 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
788 (record_subregs_of_mode): Likewise.
789 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
790 * reload.c (operands_match_p): Handle 'p'.
791 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
792 * reload1.c (alter_reg, choose_reload_regs): Likewise.
793 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
794 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
795 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
796 (simplify_const_poly_int_tests<N>::run): Likewise.
797 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
798 a poly_uint64 rather than an unsigned int.
799 * valtrack.c (debug_lowpart_subreg): Likewise.
800 * var-tracking.c (var_lowpart): Likewise.
801 (loc_cmp): Handle 'p'.
803 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
804 Alan Hayward <alan.hayward@arm.com>
805 David Sherwood <david.sherwood@arm.com>
807 * ira.c (get_subreg_tracking_sizes): New function.
808 (init_live_subregs): Take an integer size rather than a register.
809 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
810 to init_live_subregs.
812 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
813 Alan Hayward <alan.hayward@arm.com>
814 David Sherwood <david.sherwood@arm.com>
816 * expr.c (store_constructor_field): Change bitsize from a
817 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
818 HOST_WIDE_INT to a poly_int64.
819 (store_constructor): Change size from a HOST_WIDE_INT to
821 (store_field): Likewise bitsize and bitpos.
823 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
824 Alan Hayward <alan.hayward@arm.com>
825 David Sherwood <david.sherwood@arm.com>
827 * expmed.h (store_bit_field): Change bitregion_start and
828 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
829 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
830 (store_bit_field_1, store_integral_bit_field, store_bit_field)
831 (store_fixed_bit_field, store_split_bit_field): Likewise.
832 * expr.c (store_constructor_field, store_field): Likewise.
833 (optimize_bitfield_assignment_op): Likewise. Make the same change
834 to bitsize and bitpos.
835 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
836 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
837 same change in the constructor arguments.
838 (get_best_mode): Change bitregion_start and bitregion_end from
839 unsigned HOST_WIDE_INT to poly_uint64.
840 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
841 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
843 (bit_field_mode_iterator::next_mode): Update for new types
844 of m_bitregion_start and m_bitregion_end.
845 (get_best_mode): Change bitregion_start and bitregion_end from
846 unsigned HOST_WIDE_INT to poly_uint64.
848 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
849 Alan Hayward <alan.hayward@arm.com>
850 David Sherwood <david.sherwood@arm.com>
852 * rtl.h (simplify_gen_subreg): Add a temporary overload that
853 accepts poly_uint64 offsets.
854 * expmed.h (extract_bit_field): Take bitsize and bitnum as
855 poly_uint64s rather than unsigned HOST_WIDE_INTs.
856 * expmed.c (lowpart_bit_field_p): Likewise.
857 (extract_bit_field_as_subreg): New function, split out from...
858 (extract_bit_field_1): ...here. Take bitsize and bitnum as
859 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
860 extractions, check that BITSIZE matches the size of the extracted
861 value and that BITNUM is an exact multiple of that size.
862 If all else fails, try forcing the value into memory if
863 BITNUM is variable, and adjusting the address so that the
864 offset is constant. Split the part that can only handle constant
865 bitsize and bitnum out into...
866 (extract_integral_bit_field): ...this new function.
867 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
868 rather than unsigned HOST_WIDE_INTs.
870 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
871 Alan Hayward <alan.hayward@arm.com>
872 David Sherwood <david.sherwood@arm.com>
874 * expmed.h (store_bit_field): Take bitsize and bitnum as
875 poly_uint64s rather than unsigned HOST_WIDE_INTs.
876 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
877 that returns the byte size.
878 (store_bit_field_1): Take bitsize and bitnum as
879 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
880 to simple_mem_bitfield_p. Split the part that can only handle
881 constant bitsize and bitnum out into...
882 (store_integral_bit_field): ...this new function.
883 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
884 than unsigned HOST_WIDE_INTs.
885 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
887 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
888 Alan Hayward <alan.hayward@arm.com>
889 David Sherwood <david.sherwood@arm.com>
891 * lra-int.h (lra_reg): Change offset from int to poly_int64.
892 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
894 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
895 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
896 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
897 offset as a poly_int64 rather than an int.
898 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
899 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
900 * lra-constraints.c (equiv_address_substitution): Track offsets
902 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
903 (curr_insn_transform): Handle the new form of sp_offset.
904 * lra-eliminations.c (lra_elim_table): Change previous_offset
905 and offset from HOST_WIDE_INT to poly_int64.
906 (print_elim_table, update_reg_eliminate): Update accordingly.
907 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
908 (get_elimination): Update accordingly.
909 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
910 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
911 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
912 poly_int64 offsets generally.
913 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
914 (mark_not_eliminable, init_elimination): Update accordingly.
915 (remove_reg_equal_offset_note): Return a bool and pass the new
916 offset back by pointer as a poly_int64.
917 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
918 rather than a HOST_WIDE_INT.
919 (do_remat): Track offsets poly_int64s.
920 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
922 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
923 Alan Hayward <alan.hayward@arm.com>
924 David Sherwood <david.sherwood@arm.com>
926 * rtl.h (mem_attrs): Add a default constructor. Change size and
927 offset from HOST_WIDE_INT to poly_int64.
928 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
929 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
930 (widen_memory_access): Take the sizes and offsets as poly_int64s
931 rather than HOST_WIDE_INTs.
932 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
933 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
935 (adjust_offset_for_component_ref): Change the offset from a
936 HOST_WIDE_INT to a poly_int64.
937 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
938 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
939 * dce.c (find_call_stack_args): Likewise.
940 * dse.c (record_store): Likewise.
941 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
942 * print-rtl.c (rtx_writer::print_rtx): Likewise.
943 * read-rtl-function.c (test_loading_mem): Likewise.
944 * rtlanal.c (may_trap_p_1): Likewise.
945 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
946 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
947 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
948 (mem_attrs::mem_attrs): New function.
949 (set_mem_attributes_minus_bitpos): Change bitpos from a
950 HOST_WIDE_INT to poly_int64.
951 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
952 (clear_mem_offset, clear_mem_size, change_address)
953 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
954 initialize mem_attrs.
955 (set_mem_offset, set_mem_size, adjust_address_1)
956 (adjust_automodify_address_1, offset_address, widen_memory_access):
957 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
959 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
960 Alan Hayward <alan.hayward@arm.com>
961 David Sherwood <david.sherwood@arm.com>
963 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
964 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
965 rather than 0 to represent an unknown size. Assert that the size
966 is known when the mode isn't BLKmode.
967 (may_trap_p_1): Use -1 for unknown sizes.
968 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
970 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
971 Alan Hayward <alan.hayward@arm.com>
972 David Sherwood <david.sherwood@arm.com>
974 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
975 to poly_int64. Update commentary for positions_needed.large.
976 (read_info_type): Change offset and width from HOST_WIDE_INT
978 (set_usage_bits): Likewise.
979 (canon_address): Return the offset as a poly_int64 rather than
980 a HOST_WIDE_INT. Use strip_offset_and_add.
981 (set_all_positions_unneeded, any_positions_needed_p): Use
982 positions_needed.large to track stores with non-constant widths.
983 (all_positions_needed_p): Likewise. Take the offset and width
984 as poly_int64s rather than ints. Assert that rhs is nonnull.
985 (record_store): Cope with non-constant offsets and widths.
986 Nullify the rhs of an earlier store if we can't tell which bytes
988 (find_shift_sequence): Take the access_size and shift as poly_int64s
990 (get_stored_val): Take the read_offset and read_width as poly_int64s
991 rather than HOST_WIDE_INTs.
992 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
993 non-constant offsets and widths.
995 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
996 Alan Hayward <alan.hayward@arm.com>
997 David Sherwood <david.sherwood@arm.com>
999 * inchash.h (inchash::hash::add_poly_int): New function.
1000 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1001 Use poly_int64 rather than HOST_WIDE_INT.
1002 (ao_ref::max_size_known_p): New function.
1003 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1004 rather than HOST_WIDE_INT.
1005 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1006 to temporaries until its interface is adjusted to match.
1007 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1008 (aliasing_component_refs_p, decl_refs_may_alias_p)
1009 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1010 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1011 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1013 * alias.c (ao_ref_from_mem): Likewise.
1014 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1015 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1016 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1017 (maybe_trim_complex_store, maybe_trim_constructor_store)
1018 (live_bytes_read, dse_classify_store): Likewise.
1019 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1020 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1021 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1022 (vn_reference_lookup_3): Likewise.
1023 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1025 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1026 Alan Hayward <alan.hayward@arm.com>
1027 David Sherwood <david.sherwood@arm.com>
1029 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1030 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1031 instead of ranges_overlap_p.
1033 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1034 Alan Hayward <alan.hayward@arm.com>
1035 David Sherwood <david.sherwood@arm.com>
1037 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1038 sizes as poly_int64s rather than HOST_WIDE_INTs.
1040 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1041 Alan Hayward <alan.hayward@arm.com>
1042 David Sherwood <david.sherwood@arm.com>
1044 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1045 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1046 * gimple-fold.c (fold_ctor_reference): Likewise.
1048 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1049 Alan Hayward <alan.hayward@arm.com>
1050 David Sherwood <david.sherwood@arm.com>
1052 * target.def (dwarf_poly_indeterminate_value): New hook.
1053 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1054 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1055 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1056 * doc/tm.texi: Regenerate.
1057 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1058 offset as a poly_int64.
1059 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1060 offset as a poly_int64.
1061 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1062 Take the offset as a poly_int64.
1063 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1064 (frame_pointer_fb_offset): Change to a poly_int64.
1065 (int_loc_descriptor): Take the offset as a poly_int64. Use
1066 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1067 (based_loc_descr): Take the offset as a poly_int64.
1068 Use strip_offset_and_add to handle (plus X (const)).
1069 Use new_reg_loc_descr instead of an open-coded version of the
1070 previous implementation.
1071 (mem_loc_descriptor): Handle CONST_POLY_INT.
1072 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1073 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1075 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1076 Alan Hayward <alan.hayward@arm.com>
1077 David Sherwood <david.sherwood@arm.com>
1079 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1080 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1081 * inchash.h (inchash::hash::add_poly_hwi): New function.
1082 * gengtype.c (main): Register poly_int64.
1083 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1084 offset as a poly_int.
1085 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1086 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1087 offset as a poly_int64.
1088 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1089 * print-rtl.c (print_poly_int): New function.
1090 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1092 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1094 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1095 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1096 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1097 Enforce track_offset_p for parts of a PARALLEL.
1098 (vt_add_function_parameter): Use const_offset for the final
1099 offset to track. Use get_tracked_reg_offset for the parts
1102 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1103 Alan Hayward <alan.hayward@arm.com>
1104 David Sherwood <david.sherwood@arm.com>
1106 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1107 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1108 * doc/tm.texi: Regenerate.
1109 * hooks.h (hook_bool_uint_uint_true): Delete.
1110 (hook_bool_puint64_puint64_true): Declare.
1111 * hooks.c (hook_bool_uint_uint_true): Delete.
1112 (hook_bool_puint64_puint64_true): New function.
1113 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1114 instead of unsigned ints.
1115 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1116 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1118 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1119 Alan Hayward <alan.hayward@arm.com>
1120 David Sherwood <david.sherwood@arm.com>
1122 * optabs.h (expand_operand): Add an int_value field.
1123 (create_expand_operand): Add an int_value parameter and use it
1124 to initialize the new expand_operand field.
1125 (create_integer_operand): Replace with a declaration of a function
1126 that accepts poly_int64s. Move the implementation to...
1127 * optabs.c (create_integer_operand): ...here.
1128 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1129 the mode preserves the value of int_value, instead of calling
1130 const_int_operand on the rtx. Use gen_int_mode to generate
1133 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1134 Alan Hayward <alan.hayward@arm.com>
1135 David Sherwood <david.sherwood@arm.com>
1137 * dumpfile.h (dump_dec): Declare.
1138 * dumpfile.c (dump_dec): New function.
1139 * pretty-print.h (pp_wide_integer): Turn into a function and
1140 declare a poly_int version.
1141 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1143 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1144 Alan Hayward <alan.hayward@arm.com>
1145 David Sherwood <david.sherwood@arm.com>
1147 * doc/generic.texi (POLY_INT_CST): Document.
1148 * tree.def (POLY_INT_CST): New tree code.
1149 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1150 * tree-core.h (tree_poly_int_cst): New struct.
1151 (tree_node): Add a poly_int_cst field.
1152 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1153 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1154 instead of a wide_int_ref.
1155 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1157 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1158 instead of an unsigned HOST_WIDE_INT.
1159 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1160 (ptrdiff_tree_p): Declare.
1161 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1162 extern inline implementations if the target doesn't use POLY_INT_CST.
1163 (poly_int_tree_p): New function.
1164 (wi::unextended_tree): New class.
1165 (wi::int_traits <unextended_tree>): New override.
1166 (wi::extended_tree): Add a default constructor.
1167 (wi::extended_tree::get_tree): New function.
1168 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1169 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1170 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1171 (wi::tree_to_poly_wide_ref): New typedefs.
1172 (wi::ints_for): Provide overloads for extended_tree and
1174 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1175 (wi::to_wide): New functions.
1176 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1177 * tree.c (poly_int_cst_hasher): New struct.
1178 (poly_int_cst_hash_table): New variable.
1179 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1180 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1182 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1183 (init_ttree): Initialize poly_int_cst_hash_table.
1184 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1185 a poly_int64 instead of a HOST_WIDE_INT.
1186 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1187 instead of an unsigned HOST_WIDE_INT.
1188 (wide_int_to_tree): Rename to...
1189 (wide_int_to_tree_1): ...this.
1190 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1191 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1192 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1193 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1194 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1195 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1197 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1198 (streamer_read_tree_body): Likewise.
1199 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1200 (streamer_write_tree_body): Likewise.
1201 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1202 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1203 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1204 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1205 * gimple-expr.h (is_gimple_constant): Likewise.
1206 * gimplify.c (maybe_with_size_expr): Likewise.
1207 * print-tree.c (print_node): Likewise.
1208 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1209 * tree-pretty-print.c (dump_generic_node): Likewise.
1210 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1211 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1212 * tree-vrp.c (compare_values_warnv): Likewise.
1213 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1214 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1215 (force_expr_to_var_cost): Likewise.
1216 * tree-ssa-loop.c (for_each_index): Likewise.
1217 * fold-const.h (build_invariant_address, size_int_kind): Take a
1218 poly_int64 instead of a HOST_WIDE_INT.
1219 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1220 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1222 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1223 (int_const_binop_2): New function, split out from...
1224 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1225 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1226 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1227 * gimple-ssa-strength-reduction.c (slsr_process_add)
1228 (slsr_process_mul): Check for INTEGER_CSTs before using them
1230 * stor-layout.c (bits_from_bytes): New function.
1231 (bit_from_pos): Use it.
1232 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1233 by BITS_PER_UNIT to get the TYPE_SIZE.
1234 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1235 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1237 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1238 Alan Hayward <alan.hayward@arm.com>
1239 David Sherwood <david.sherwood@arm.com>
1241 * doc/rtl.texi (const_poly_int): Document. Also document the
1242 rtl sharing behavior.
1243 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1244 * rtl.h (const_poly_int_def): New struct.
1245 (rtx_def::u): Add a cpi field.
1246 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1247 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1248 (wi::rtx_to_poly_wide_ref): New typedef
1249 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1250 (poly_int_rtx_p): New functions.
1251 (trunc_int_for_mode): Declare a poly_int64 version.
1252 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1253 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1255 (strip_offset): Declare.
1256 (strip_offset_and_add): New function.
1257 * rtl.def (CONST_POLY_INT): New rtx code.
1258 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1259 (shared_const_p): Use poly_int_rtx_p.
1260 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1262 (gen_int_shift_amount): Likewise.
1263 * emit-rtl.c (const_poly_int_hasher): New class.
1264 (const_poly_int_htab): New variable.
1265 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1266 (const_poly_int_hasher::hash): New function.
1267 (const_poly_int_hasher::equal): Likewise.
1268 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1269 (immed_wide_int_const): Rename to...
1270 (immed_wide_int_const_1): ...this and make static.
1271 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1272 instead of a wide_int_ref.
1273 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1274 (gen_lowpart_common): Handle CONST_POLY_INT.
1275 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1276 * cselib.c (cselib_hash_rtx): Likewise.
1277 * dwarf2out.c (const_ok_for_output_1): Likewise.
1278 * expr.c (convert_modes): Likewise.
1279 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1280 * rtlhash.c (add_rtx): Likewise.
1281 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1282 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1283 Handle existing CONST_POLY_INT rtxes.
1284 * expmed.h (expand_shift): Take a poly_int64 instead of a
1286 * expmed.c (expand_shift): Likewise.
1287 * rtlanal.c (strip_offset): New function.
1288 (commutative_operand_precedence): Give CONST_POLY_INT the same
1289 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1291 * rtl-tests.c (const_poly_int_tests): New struct.
1292 (rtl_tests_c_tests): Use it.
1293 * simplify-rtx.c (simplify_const_unary_operation): Handle
1295 (simplify_const_binary_operation): Likewise.
1296 (simplify_binary_operation_1): Fold additions of symbolic constants
1297 and CONST_POLY_INTs.
1298 (simplify_subreg): Handle extensions and truncations of
1300 (simplify_const_poly_int_tests): New struct.
1301 (simplify_rtx_c_tests): Use it.
1302 * wide-int.h (storage_ref): Add default constructor.
1303 (wide_int_ref_storage): Likewise.
1304 (trailing_wide_ints): Use GTY((user)).
1305 (trailing_wide_ints::operator[]): Add a const version.
1306 (trailing_wide_ints::get_precision): New function.
1307 (trailing_wide_ints::extra_size): Likewise.
1309 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1310 Alan Hayward <alan.hayward@arm.com>
1311 David Sherwood <david.sherwood@arm.com>
1313 * emit-rtl.h (gen_int_shift_amount): Declare.
1314 * emit-rtl.c (gen_int_shift_amount): New function.
1315 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1317 * calls.c (shift_return_value): Likewise.
1318 * cse.c (fold_rtx): Likewise.
1319 * dse.c (find_shift_sequence): Likewise.
1320 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1321 (expand_shift, expand_smod_pow2): Likewise.
1322 * lower-subreg.c (shift_cost): Likewise.
1323 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1324 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1325 (expand_vec_perm_var): Likewise.
1326 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1327 (simplify_binary_operation_1): Likewise.
1328 * combine.c (try_combine, find_split_point, force_int_to_mode)
1329 (simplify_shift_const_1, simplify_shift_const): Likewise.
1330 (change_zero_ext): Likewise. Use simplify_gen_binary.
1332 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1334 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1336 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1338 * doc/invoke.texi (ARM Options): Document accepted extension options
1339 for -march=armv8.3-a.
1341 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1344 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1345 or --with-float=softfp, set the default CPU to arm10e.
1347 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1349 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1350 * config/visium/predicates.md (const_shift_operand): Likewise.
1351 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1352 (visium_legitimize_reload_address): Likewise.
1354 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1357 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1359 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1362 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1364 2017-12-20 Julia Koval <julia.koval@intel.com>
1366 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1367 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1368 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1369 * config.gcc: Include vpclmulqdqintrin.h.
1370 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1371 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1373 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1374 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1375 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1376 * config/i386/i386.c (isa2_opts): Add -mcx16.
1377 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1378 (ix86_option_override_internal): Move mcx16 to flags2.
1379 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1380 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1381 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1382 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1383 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1384 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1385 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1386 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1387 * doc/invoke.texi: Add -mvpclmulqdq.
1389 2017-12-20 Tom de Vries <tom@codesourcery.com>
1392 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
1393 * calls.c (rtx_for_static_chain): ... here. New function.
1394 * calls.h (rtx_for_static_chain): Declare.
1395 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1396 instead of targetm.calls.static_chain.
1397 * df-scan.c (df_get_entry_block_def_set): Same.
1399 2017-12-19 Tom de Vries <tom@codesourcery.com>
1401 PR tree-optimization/83493
1402 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1405 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1407 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1408 inform with hardcoded english plural handling.
1410 2017-12-18 Jeff Law <law@redhat.com>
1412 PR tree-optimization/83477
1413 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1414 a non-virtual PHI, always push a new range.
1416 2017-12-19 Martin Sebor <msebor@redhat.com>
1419 * builtins.c (compute_objsize): Handle non-constant offsets.
1421 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1423 PR tree-optimization/83444
1424 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1425 character load case, if get_stridx on MEM_REF's operand doesn't
1426 look usable, retry with get_addr_stridx.
1428 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1431 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1432 (variable_tracking_main_1): Keep markers even when VTA fails.
1435 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1436 even if there are markers before it.
1437 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1439 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1441 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1442 typical order conditions.
1443 * sel-sched.c (extract_new_fences_from): Likewise.
1444 * config/visium/constraints.md (J, K, L): Likewise.
1445 * config/visium/predicates.md (const_shift_operand): Likewise.
1446 * config/visium/visium.c (visium_legitimize_address,
1447 visium_legitimize_reload_address): Likewise.
1448 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1449 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1450 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1451 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1452 SET_DUMP_DETAIL): Likewise.
1453 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1454 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1455 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1456 avr_set_core_architecture, avr_set_current_function,
1457 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1458 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1459 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1460 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1461 avr_map_decompose, avr_fold_builtin): Likewise.
1462 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1463 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1464 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1465 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1466 * config/m32c/m32c.c (m32c_conditional_register_usage,
1467 m32c_address_cost): Likewise.
1468 * config/m32c/predicates.md (shiftcount_operand,
1469 longshiftcount_operand): Likewise.
1470 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1471 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1472 can_use_cdx_ldstw): Likewise.
1473 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1474 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1476 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1477 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1478 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1479 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1480 * config/vax/vax.c (adjacent_operands_p): Likewise.
1481 * config/ft32/constraints.md (L, b, KA): Likewise.
1482 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1484 * cfgexpand.c (expand_stack_alignment): Likewise.
1485 * gcse.c (insert_expr_in_table): Likewise.
1486 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1487 * cgraphunit.c (cgraph_node::expand): Likewise.
1488 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1489 * emit-rtl.c (add_insn): Likewise.
1490 * input.c (dump_location_info): Likewise.
1491 * passes.c (NEXT_PASS): Likewise.
1492 * read-rtl-function.c (parse_note_insn_name,
1493 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1495 * sched-rgn.c (sched_rgn_init): Likewise.
1496 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1497 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1498 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
1499 * explow.c (eliminate_constant_term): Likewise.
1500 * final.c (leaf_renumber_regs_insn): Likewise.
1501 * cfgrtl.c (print_rtl_with_bb): Likewise.
1502 * genhooks.c (emit_init_macros): Likewise.
1503 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1504 * tree-data-ref.c (conflict_fn): Likewise.
1505 * selftest.c (assert_streq): Likewise.
1506 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1507 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1508 fold_binary_loc, multiple_of_p): Likewise.
1509 * reload.c (push_reload, find_equiv_reg): Likewise.
1510 * et-forest.c (et_nca, et_below): Likewise.
1511 * dbxout.c (dbxout_symbol_location): Likewise.
1512 * reorg.c (relax_delay_slots): Likewise.
1513 * dojump.c (do_compare_rtx_and_jump): Likewise.
1514 * gengtype-parse.c (type): Likewise.
1515 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1516 simplify_const_relational_operation): Likewise.
1517 * reload1.c (do_output_reload): Likewise.
1518 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1519 * gengtype.c (type_for_name): Likewise.
1520 * gimple-ssa-sprintf.c (format_directive): Likewise.
1522 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1525 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1526 accessing it. Adjust comment.
1528 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1531 * predict.c (zero_one_minusone): New function.
1532 (apply_return_prediction): Avoid return prediction for functions
1533 returning only -1, 0 and 1 values, unless they only return -1 and 0
1536 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1538 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1539 fall-through warning.
1540 (arc_compute_frame_size): Remove unused variables.
1541 (arc_print_operand): Fix fprintif format.
1542 (arc_can_follow_jump): Clean fall-through warning.
1544 2017-12-19 Marek Polacek <polacek@redhat.com>
1547 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1550 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1552 * config/arc/arc.c (overriderregs): New variable.
1553 (arc_override_options): Track fixed/call saved/call options.
1554 (arc_conditional_register_usage): Check against overrideregs
1555 variable whenever we change register properties.
1557 2017-12-19 Nathan Sidwell <nathan@acm.org>
1559 * opts.c (finish_options): Don't prefix dump_base_name if it
1560 already contains directories.
1562 2017-12-19 Martin Liska <mliska@suse.cz>
1564 PR rtl-optimization/82675
1565 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1566 more element in sbitmap.
1568 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1570 * gimplify.c (gimplify_expr): Use error_operand_p.
1572 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1575 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1578 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1580 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1581 the jump is to a label.
1583 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1585 PR tree-optimization/83444
1586 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1591 * ipa-inline.c (flatten_remove_node_hook): New function.
1592 (ipa_inline): Keep only nodes with flatten attribute at the end of
1593 the array in the order from ipa_reverse_postorder, only walk that
1594 portion of array for flattening, if there is more than one such
1595 node, temporarily register a removal hook and ignore removed nodes.
1597 PR tree-optimization/80631
1598 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1599 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1600 IFN_REDUC_MAX or IFN_REDUC_MIN.
1602 2017-12-18 Jeff Law <law@redhat.com>
1604 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1605 record symbolic equivalences from backedges in the CFG.
1608 2017-11-19 Jeff Law <law@redhat.com>
1610 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1611 of degenerates resulting from ignoring an edge.
1613 2017-12-18 Martin Sebor <msebor@redhat.com>
1616 PR tree-optimization/78450
1617 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1618 (handle_builtin_strlen): Call it.
1620 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1622 PR rtl-optimization/83424
1623 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1625 2017-12-18 Marek Polacek <polacek@redhat.com>
1628 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1629 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1632 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1634 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1635 (umaddsidi4, umaddsidi_split): Likewise.
1637 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1639 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1642 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1645 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1647 2017-12-18 Richard Biener <rguenther@suse.de>
1649 PR tree-optimization/81877
1650 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1651 (outermost_indep_loop): Adjust.
1652 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1653 (can_sm_ref_p): Adjust.
1655 2017-12-18 Richard Biener <rguenther@suse.de>
1658 * tree.c (array_at_struct_end_p): Return true if the underlying
1659 object has space for at least one element in excess of what
1660 the array domain specifies.
1662 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1664 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1665 example to avoid overfull hbox.
1666 * doc/invoke.texi (Option Summary): Add missing @gol.
1667 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1670 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1671 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1673 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1674 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1675 (AArch64 Options): Document -mverbose-cost-dump.
1676 (ARM Options): Likewise, plus -mflip-thumb.
1678 2017-12-17 Martin Sebor <msebor@redhat.com>
1681 * gimple-ssa-warn-restrict.c
1682 (builtin_memref::offset_out_of_bounds): Correct the handling of
1685 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1687 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1689 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1691 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1693 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1695 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1698 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1700 * doc/invoke.texi: Fix some typos.
1702 2017-12-16 Martin Sebor <msebor@redhat.com>
1704 PR tree-optimization/78918
1705 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1706 * builtins.c (check_sizes): Rename...
1707 (check_access): ...to this. Rename function arguments for clarity.
1708 (check_memop_sizes): Adjust names.
1709 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1710 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1711 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1712 (check_strncat_sizes, expand_builtin_strncat): Same.
1713 (expand_builtin_strncpy, expand_builtin_memset): Same.
1714 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1715 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1716 (maybe_emit_sprintf_chk_warning): Same.
1717 (expand_builtin_strcpy): Adjust.
1718 (expand_builtin_stpcpy): Same.
1719 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1720 in pointer-checking forms of memcpy, memmove, and mempcpy.
1721 (gcall_to_tree_minimal, max_object_size): Define new functions.
1722 * builtins.h (max_object_size): Declare.
1723 * calls.c (alloc_max_size): Call max_object_size instead of
1724 hardcoding ssizetype limit.
1725 (get_size_range): Handle new argument.
1726 * calls.h (get_size_range): Add a new argument.
1727 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1728 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1729 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1731 (gimple_fold_builtin_memory_chk): Same.
1732 (gimple_fold_builtin_stxcpy_chk): New function.
1733 * gimple-ssa-warn-restrict.c: New source.
1734 * gimple-ssa-warn-restrict.h: New header.
1735 * gimple.c (gimple_build_call_from_tree): Propagate location.
1736 * passes.def (pass_warn_restrict): Add new pass.
1737 * tree-pass.h (make_pass_warn_restrict): Declare.
1738 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1740 (handle_builtin_strcat): Same.
1741 (strlen_optimize_stmt): Rename...
1742 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1743 stpncpy, strncpy, and their checking forms.
1745 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1747 PR rtl-optimization/82849
1748 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1749 and get_max_loop_iterations_int.
1751 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1752 Alan Hayward <alan.hayward@arm.com>
1753 David Sherwood <david.sherwood@arm.com>
1755 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1756 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1757 size as a poly_uint64.
1758 (mode_for_vector, mode_for_int_vector): Take the number of vector
1759 elements as a poly_uint64.
1760 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1761 size as a poly_uint64.
1762 (mode_for_vector, mode_for_int_vector): Take the number of vector
1763 elements as a poly_uint64.
1765 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1766 Alan Hayward <alan.hayward@arm.com>
1767 David Sherwood <david.sherwood@arm.com>
1769 * machmode.h (MACRO_MODE): New macro.
1770 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1771 * targhooks.c (default_libcall_value, default_secondary_reload)
1772 (default_memory_move_cost, default_register_move_cost)
1773 (default_class_max_nregs): Likewise.
1775 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1776 Alan Hayward <alan.hayward@arm.com>
1777 David Sherwood <david.sherwood@arm.com>
1779 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1780 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1782 * genautomata.c (main): Likewise.
1783 * genconditions.c (write_header): Likewise.
1784 * genemit.c (main): Likewise.
1785 * genextract.c (print_header): Likewise.
1786 * genopinit.c (main): Likewise.
1787 * genoutput.c (output_prologue): Likewise.
1788 * genpeep.c (main): Likewise.
1789 * genpreds.c (write_insn_preds_c): Likewise.
1790 * genrecog.c (writer_header): Likewise.
1791 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1792 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1793 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
1794 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
1795 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
1796 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
1797 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
1798 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
1799 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
1800 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
1801 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
1802 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
1803 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
1804 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
1805 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
1806 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
1807 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
1808 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
1809 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
1810 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
1811 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
1812 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
1813 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
1814 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
1815 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
1816 * config/darwin.c (IN_TARGET_CODE): Likewise.
1817 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
1818 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
1819 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
1820 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
1821 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
1822 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
1823 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
1824 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
1825 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
1826 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
1827 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
1828 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
1829 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
1830 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
1831 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
1832 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
1833 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
1834 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
1835 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
1836 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
1837 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
1838 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
1839 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
1840 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
1841 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
1842 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
1843 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
1844 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
1845 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
1846 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
1847 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
1848 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
1849 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
1850 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
1851 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
1852 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
1853 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
1854 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
1855 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
1856 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
1857 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
1858 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
1859 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
1860 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
1861 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
1862 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
1863 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
1864 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
1865 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
1866 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
1867 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
1868 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
1869 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
1870 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
1871 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
1872 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
1873 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
1874 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
1875 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
1876 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
1877 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1878 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
1879 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
1880 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
1881 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
1882 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
1883 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
1884 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
1885 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
1886 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
1887 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
1888 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1889 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
1890 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
1891 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
1892 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
1893 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
1894 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
1895 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
1896 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
1897 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
1898 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
1899 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
1900 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
1901 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
1902 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
1903 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
1904 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
1905 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
1906 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
1907 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
1908 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
1909 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
1910 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
1911 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
1912 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
1913 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
1914 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
1915 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
1916 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
1917 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
1918 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
1919 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
1920 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
1921 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
1922 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
1924 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1925 Alan Hayward <alan.hayward@arm.com>
1926 David Sherwood <david.sherwood@arm.com>
1928 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
1931 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1932 Alan Hayward <alan.hayward@arm.com>
1933 David Sherwood <david.sherwood@arm.com>
1935 * doc/generic.texi (VEC_SERIES_EXPR): Document.
1936 * doc/md.texi (vec_series@var{m}): Document.
1937 * tree.def (VEC_SERIES_EXPR): New tree code.
1938 * tree.h (build_vec_series): Declare.
1939 * tree.c (build_vec_series): New function.
1940 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
1941 * tree-pretty-print.c (dump_generic_node): Likewise.
1942 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
1943 * tree-inline.c (estimate_operator_cost): Likewise.
1944 * expr.c (expand_expr_real_2): Likewise.
1945 * optabs-tree.c (optab_for_tree_code): Likewise.
1946 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
1947 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
1948 * expmed.c (make_tree): Handle VEC_SERIES.
1949 * optabs.def (vec_series_optab): New optab.
1950 * optabs.h (expand_vec_series_expr): Declare.
1951 * optabs.c (expand_vec_series_expr): New function.
1952 * tree-vect-generic.c (expand_vector_operations_1): Check that
1953 the operands also have vector type.
1955 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1956 Alan Hayward <alan.hawyard@arm.com>
1957 David Sherwood <david.sherwood@arm.com>
1959 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
1960 (VEC_COND_EXPR): Add missing @tindex.
1961 * doc/md.texi (vec_duplicate@var{m}): Document.
1962 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
1963 * tree.c (build_vector_from_val): Add stubbed-out handling of
1964 variable-length vectors, using VEC_DUPLICATE_EXPR.
1965 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
1966 * cfgexpand.c (expand_debug_expr): Likewise.
1967 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1968 * tree-inline.c (estimate_operator_cost): Likewise.
1969 * tree-pretty-print.c (dump_generic_node): Likewise.
1970 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
1971 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
1972 (test_vec_duplicate_folding): New function.
1973 (fold_const_c_tests): Call it.
1974 * optabs.def (vec_duplicate_optab): New optab.
1975 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
1976 * optabs.h (expand_vector_broadcast): Declare.
1977 * optabs.c (expand_vector_broadcast): Make non-static. Try using
1978 vec_duplicate_optab.
1979 * expr.c (store_constructor): Try using vec_duplicate_optab for
1981 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
1983 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
1986 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
1987 div/mod latencies a bit.
1989 2017-12-15 Jeff Law <law@redhat.com>
1991 PR tree-optimization/36550
1992 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
1993 (mark_threaded_blocks): Rewrite code to avoid block copying when
1994 optimizing for size. Don't pessimize blocks which will be
1995 copied, but all the statements will be dead.
1997 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
1999 PR tree-optimization/81165
2000 * tree-ssa-threadupdate.c (uses_in_bb): New.
2001 (estimate_threading_killed_stmts): New.
2002 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2003 * tree-ssa-threadedge.c
2004 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2007 2017-12-15 Jeff Law <law@redhat.com>
2009 PR tree-optimization/83410
2010 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2011 threads when parallelizing loops.
2013 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2015 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2017 * config/alpha/alpha.c (vms_attribute_table): Swap
2018 affects_type_identity and handler fields, adjust comments.
2019 * config/mips/mips.c (mips_attribute_table): Likewise.
2020 * config/visium/visium.c (visium_attribute_table): Likewise.
2021 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2022 * config/microblaze/microblaze.c (microblaze_attribute_table):
2024 * config/spu/spu.c (spu_attribute_table): Likewise.
2025 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2026 * config/arc/arc.c (arc_attribute_table): Likewise.
2027 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2028 * config/v850/v850.c (v850_handle_interrupt_attribute,
2029 v850_handle_data_area_attribute): Formatting fixes.
2030 (v850_attribute_table): Swap affects_type_identity and handler
2031 fields, adjust comments.
2032 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2033 * config/arm/arm.c (arm_attribute_table): Likewise.
2034 * config/avr/avr.c (avr_attribute_table): Likewise.
2035 * config/s390/s390.c (s390_attribute_table): Likewise.
2036 * config/sh/sh.c (sh_attribute_table): Likewise.
2037 * config/i386/i386.c (ix86_handle_cconv_attribute,
2038 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2039 (ix86_attribute_table): Swap affects_type_identity and handler
2040 fields, adjust comments.
2041 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2042 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2043 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2044 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2045 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2046 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2047 * config/rx/rx.c (rx_attribute_table): Likewise.
2048 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2049 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2050 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2051 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2052 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2053 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2054 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2055 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2056 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2057 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2058 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2060 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2062 (register_scoped_attributes, decl_attributes): Formatting fixes.
2064 PR tree-optimization/83269
2065 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2066 subtraction in arg0's type if type is signed and arg0 is unsigned.
2070 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2071 convert? on @0 instead of convert. Check type of @1, not @0.
2072 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2073 convert? on @0 instead of convert. Check type of @1, not @0.
2074 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2075 only optimize if either both @1 and @2 types are narrower
2076 precision, or both are wider or equal precision, and in the former
2077 case only if both have undefined overflow.
2079 2017-12-15 Richard Biener <rguenther@suse.de>
2082 * internal-fn.def (IFN_NOP): Add.
2083 * internal-fn.c (expand_NOP): Do nothing.
2084 * lto-streamer-in.c (input_function): Instead of removing
2085 sanitizer calls replace them with IFN_NOP calls.
2087 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2088 Alan Hayward <alan.hayward@arm.com>
2089 David Sherwood <david.sherwood@arm.com>
2091 * dse.c (store_info, read_info_type): Replace begin and end with
2093 (print_range): New function.
2094 (set_all_positions_unneeded, any_positions_needed_p)
2095 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2097 (record_store): Likewise. Optimize the case in which all positions
2099 (get_stored_val): Replace read_begin and read_end with read_offset
2101 (replace_read): Update call accordingly.
2103 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2105 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2106 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2107 (loop_cand::loop_cand): Initialize above members.
2108 (loop_cand::supported_operations): Delete.
2109 (loop_cand::can_interchange_p): Inline above function.
2110 (loop_cand::classify_simple_reduction): Record number of constant
2111 initialized simple reductions.
2112 (should_interchange_loops): New parameters. Check stmt cost of loops
2114 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2115 Update call to should_interchange_loops.
2116 (should_interchange_loop_nest): Update call to
2117 should_interchange_loops.
2119 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2122 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2123 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2124 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2126 2017-12-15 Julia Koval <julia.koval@intel.com>
2128 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2129 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
2130 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2131 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2132 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2134 2017-12-15 Julia Koval <julia.koval@intel.com>
2136 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2137 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2138 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2139 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2140 _mm_aesenc_epi128): New intrinsics.
2142 2017-12-15 Julia Koval <julia.koval@intel.com>
2144 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2145 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
2146 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2147 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2148 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2150 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2152 * gimple-ssa-strength-reduction.c (analyze_increments):
2153 Distinguish replacement costs for constant strides from those for
2156 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2158 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2160 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2162 * doc/invoke.texi: Document -Wcast-function-type.
2163 * recog.h (stored_funcptr): Change signature.
2164 * tree-dump.c (dump_node): Avoid warning.
2165 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2167 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2170 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2173 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2175 PR tree-optimization/83312
2176 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2177 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2178 "val" param, and to cope with arbitrary basic blocks.
2179 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2180 handle NULL_TREE for "val", dropping "bb" param.
2181 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2182 drop "bb" param. Handle NULL_TREE for "val".
2183 (find_case_label_for_value): Make "switch_stmt" param const.
2184 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2186 (vrp_prop::check_all_array_refs): Reimplement as...
2187 (check_array_bounds_dom_walker::before_dom_children): ...this new
2188 vfunc. Replace linear search through BB block list, excluding
2189 those with non-executable in-edges via dominator walk.
2191 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2193 * config/arm/arm.opt (mverbose-cost-dump): New option.
2194 * config/arm/arm.c (arm_rtx_costs): Use it.
2196 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2199 * reload1.c (emit_input_reload_insns): Skip debug markers.
2201 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2203 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2207 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2208 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2209 * config/sh/sh.c: Skip debug insns besides notes.
2210 * config/sh/sh.md: Likewise.
2211 * config/sh/sh_treg_combine.cc: Likewise.
2212 * config/sh/sync.md: Likewise.
2214 2017-12-14 Tom de Vries <tom@codesourcery.com>
2216 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2217 Add item for weak_undefined.
2219 2017-12-14 Richard Biener <rguenther@suse.de>
2221 PR tree-optimization/67842
2222 * sese.h (bb_in_region): Remove #if 0'ed code.
2224 2017-12-14 Richard Biener <rguenther@suse.de>
2226 PR tree-optimization/83326
2227 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2228 may_be_zero parameter and handle it by not marking the first
2229 peeled copy as not exiting the loop.
2230 (try_peel_loop): Likewise.
2231 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2232 to handle the case of constant or zero iterations and perform
2233 loop header copying on-the-fly.
2235 2017-12-14 Richard Biener <rguenther@suse.de>
2237 PR tree-optimization/83418
2238 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2239 Instead of asserting we don't get unfolded comparisons deal with
2242 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2245 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2246 basic blocks. Assert debug bind insns don't appear outside of bbs,
2247 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2250 PR tree-optimization/83198
2251 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2252 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2253 value if arg is a REAL_CST with incompatible type.
2255 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2256 Bin Cheng <bin.cheng@arm.com>
2259 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2261 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2264 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2266 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2267 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2269 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2270 Alan Hayward <alan.hayward@arm.com>
2271 David Sherwood <david.sherwood@arm.com>
2273 * poly-int.h: New file.
2274 * poly-int-types.h: Likewise.
2275 * coretypes.h: Include them.
2276 (POLY_INT_CONVERSION): Define.
2277 * target.def (estimated_poly_value): New hook.
2278 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2279 * doc/tm.texi: Regenerate.
2280 * doc/poly-int.texi: New file.
2281 * doc/gccint.texi: Include it.
2282 * doc/rtl.texi: Describe restrictions on subreg modes.
2283 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2284 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2285 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2286 * targhooks.h (default_estimated_poly_value): Declare.
2287 * targhooks.c (default_estimated_poly_value): New function.
2288 * target.h (estimated_poly_value): Likewise.
2289 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2290 (wi::unary_traits): Delete.
2291 (wi::binary_traits::signed_shift_result_type): Define for
2292 offset_int << HOST_WIDE_INT, etc.
2293 (generic_wide_int::operator <<=): Define for all types and use
2294 wi::lshift instead of <<.
2295 (wi::hwi_with_prec): Add a default constructor.
2296 (wi::ints_for): New class.
2297 (operator <<): Define for all wide-int types.
2298 (operator /): New function.
2299 (operator %): Likewise.
2300 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2301 (ASSERT_MAYBE_NE_AT): New macros.
2303 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2304 Dominik Vogt <vogt@linux.vnet.ibm.com>
2307 * emit-rtl.c (init_emit): Remove ??? comment.
2308 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2309 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2310 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2311 alignment of 3 virtual registers to BITS_PER_WORD.
2313 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2315 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2317 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2318 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2320 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2323 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2325 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2327 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2328 with the exception of debug begin stmt markers appear before
2332 * final.c (rest_of_handle_final): Call variable_tracking_main only
2333 if !flag_var_tracking.
2335 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2336 Jakub Jelinek <jakub@redhat.com>
2340 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2341 labels debug stmts that can only appear after labels.
2343 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2345 PR rtl-optimization/82398
2346 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2347 EXPR_USEFULNESS in priority comparison.
2349 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2351 PR rtl-optimization/83393
2352 * combine.c (move_deaths): If reg_stat points to a too new insn in
2353 last_death, do not use it: find the proper insn instead.
2355 2017-12-12 Jeff Law <law@redhat.com>
2357 PR tree-optimization/83298
2358 PR tree-optimization/83362
2359 PR tree-optimization/83383
2360 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2361 push_value_range a public interface. Add new argument to
2362 record_ranges_from_stmt.
2363 * gimple-ssa-evrp-analyze.c
2364 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2365 Update comments. Handle recording temporary equivalences.
2366 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2367 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2368 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2369 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2370 gimple-ssa-evrp-analyze.h.
2371 (record_temporary_equivalences_from_phis): Add new argument. When
2372 the PHI arg is an SSA_NAME, set the result's range to the range
2374 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2375 from statements too.
2376 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2377 Pass it down to children as needed.
2378 (thread_outgoing_edges): Likewise.
2379 (thread_across_edge): Likewise. Push/pop range state as needed.
2380 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2382 2017-12-12 Julia Koval <julia.koval@intel.com>
2384 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2385 (PTA_CANNONLAKE): Remove PTA_CLWB.
2387 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2390 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2391 (vcondv2div2df): Likewise.
2392 (vconduv2dfv2di): Likewise.
2394 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2396 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2397 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2398 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2399 ATTR_MATHFN_FPROUNDING_ERRNO.
2401 2017-12-12 Richard Biener <rguenther@suse.de>
2403 PR tree-optimization/83385
2404 * graphite-scop-detection.c (get_order, order): Remove.
2405 (bb_to_rpo): New global.
2407 (build_scops): Sort pbbs in RPO order.
2409 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2411 * combine.c (simplify_set): Do not transform subregs to zero_extends
2412 if the destination mode is a vector mode.
2414 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2416 PR tree-optimization/83359
2417 * tree-cfg.h (fold_loop_internal_call): Declare.
2418 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2419 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2420 (find_loop_dist_alias): New function.
2421 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2422 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2425 PR tree-optimization/80631
2426 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2427 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2428 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2429 hardcoding zero as the value if COND_EXPR is never true. For
2430 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2431 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2432 hardcoding MAX_EXPR as the reduction operation.
2433 (is_nonwrapping_integer_induction): Allow negative step.
2434 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2435 vect_create_epilog_for_reduction, if no value is suitable, don't
2436 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2438 2017-12-12 Richard Biener <rguenther@suse.de>
2440 PR tree-optimization/81889
2441 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2442 range info from the non-wrapping IV instead of just the range
2445 2017-12-12 Julia Koval <julia.koval@intel.com>
2447 * config.gcc: Add vaesintrin.h.
2448 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2450 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2451 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2453 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2454 * config/i386/immintrin.h: Include vaesintrin.h.
2455 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2456 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2457 _mm_aesdec_epi128): New intrinsics.
2459 2017-12-12 Julia Koval <julia.koval@intel.com>
2461 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2462 OPTION_MASK_ISA_VAES_UNSET): New.
2463 (ix86_handle_option): Handle -mvaes.
2464 * config/i386/cpuid.h: Define bit_VAES.
2465 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2466 * config/i386/i386-c.c (__VAES__): New.
2467 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2468 (ix86_valid_target_attribute_inner_p): Ditto.
2469 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2470 * config/i386/i386.opt: Add -mvaes.
2471 * doc/invoke.texi: Ditto.
2473 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2475 * debug.h (gcc_debug_hooks): Add inline_entry.
2476 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2477 * debug.c (do_nothing_debug_hooks): Likewise.
2478 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2479 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2480 (dwarf2_lineno_debug_hooks): Likewise.
2482 * common.opt (gstatement-frontiers): New, setting
2483 debug_nonbind_markers_p.
2484 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2485 * toplev.c (process_options): Autodetect value for debug statement
2487 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2488 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2490 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2491 markers. Integrate source bind into debug stmt expand loop.
2492 (pass_expand::execute): Check debug marker limit. Avoid deep
2493 TER and expand debug locations for debug bind insns only.
2494 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2496 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2497 * final.c (reemit_insn_block_notes): Take current block from
2498 nonbind markers. Declare note where it's first set.
2499 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2500 begin stmt markers if enabled.
2501 (notice_source_line): Handle nonbind markers. Fail if their
2502 location is unknown or that of builtins.
2503 (rest_of_handle_final): Convert begin stmt markers to notes if
2504 var-tracking didn't run.
2505 (rest_of_clean_state): Skip begin stmt markers.
2506 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2508 * function.c (allocate_struct_function): Set begin_stmt_markers.
2509 * function.h (struct function): Add debug_marker_count counter
2510 and debug_nonbind_markers flag.
2511 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2512 * gimple-low.c (lower_function_body): Adjust
2513 debug_nonbind_markers.
2514 (lower_stmt): Drop or skip gimple debug stmts.
2515 (lower_try_catch): Skip debug stmts.
2516 * gimple.c (gimple_build_debug_begin_stmt): New.
2517 (gimple_copy): Increment debug_marker_count if copying one.
2518 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2519 * gimplify.c (rexpr_location): New.
2520 (rexpr_has_location): New.
2521 (warn_switch_unreachable_r): Handle gimple debug stmts.
2522 (shortcut_cond_r): Call expr_location.
2524 (find_goto_label): New.
2525 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2527 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2529 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2530 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2531 (LANG_HOOKS_INITIALIZER): ... this.
2532 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2533 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2535 (update_ebb_live_info): Skip debug insn markers.
2536 * lra.c (debug_insn_static_data): Rename to...
2537 (debug_bind_static_data): ... this.
2538 (debug_marker_static_data): New.
2539 (lra_set_insn_recog_data): Select one of the above depending
2541 (lra_update_isn_regno_info): Don't assume debug insns have
2543 (push_insns): Skip debug insns.
2544 * lto-streamer-in.c (input_function): Drop debug stmts
2545 depending on active options. Adjust debug_nonbind_markers.
2546 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2547 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2548 begin stmt marker notes.
2549 (print_insn): Likewise.
2550 * recog.c (extract_insn): Recognize rtl for debug markers.
2551 * rtl.def (DEBUG_MARKER): New.
2552 * tree-inline.c: Include params.h.
2553 (remap_gimple_stmt): Handle nonbind markers.
2554 (maybe_move_debug_stmts_to_successors): Likewise.
2555 (copy_debug_stmt): Likewise.
2556 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2557 markers regardless of no side effects.
2558 (tsi_link_before): Don't update container's side effects when adding
2559 a begin stmt marker.
2560 (tsi_link_after): Likewise.
2561 (expr_first): Skip begin stmt markers.
2562 (expr_last): Likewise.
2563 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2564 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2565 Disregard nonbind markers.
2566 * tree.c (make_node_stat): Don't set side effects for begin stmt
2568 (build1_stat): Likewise.
2569 * tree.def (DEBUG_BEGIN_STMT): New.
2570 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2571 * var-tracking.c (delete_debug_insns): Renamed to...
2572 (delete_vta_debug_insns): ... this.
2573 (reemit_marker_as_note): New.
2574 (vt_initialize): Reemit markers.
2575 (delete_vta_debug_insns): Likewise.
2576 (vt_debug_insns_local): Reemit or delete markers.
2577 (variable_tracking_main_1): Likewise.
2578 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2579 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2580 (gimple_debug_nonbind_marker_p): New.
2581 (gimple_build_debug_bind): Adjust.
2582 (gimple_build_debug_begin_stmt): New.
2583 * doc/invoke.texi (max-debug-marker-count): New param.
2584 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2585 (debug_parameter_ref, debug_marker): New.
2586 (NOTE_INSN_BEGIN_STMT): New.
2587 (DEBUG_INSN): Describe begin stmt markers.
2589 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2590 without debug insns, we wouldn't, but clean up debug insns
2591 after a control flow insn nevertheless.
2593 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2594 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2595 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2596 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2597 (rtl_tidy_fallthru_edge): Likewise.
2598 (rtl_verify_fallthru): Likewise.
2599 (rtl_verify_bb_layout): Likewise.
2600 (skip_insns_after_block): Likewise.
2601 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2602 * dwarf2out.c: Include print-rtl.h.
2603 (dwarf2out_next_real_insn): New.
2604 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2605 Dump debug binds in asm comments.
2606 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2607 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2608 callers to use gsi_start_nondebug_bb instead.
2609 (gsi_after_labels): Skip gimple debug stmts.
2610 (gsi_start_nondebug): New.
2611 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2612 (proper_loop_form_for_interchange): Adjust.
2613 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2614 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2615 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2616 (collect_fallthrough_labels): Likewise.
2617 (should_warn_for_implicit_fallthrough): Likewise.
2618 (warn_implicit_fallthrough_r): Likewise.
2619 (expand_FALLTHROUGH_r): Likewise.
2620 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2621 (graphite_copy_stmts_from_block): Skip nonbind markers.
2622 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2623 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2624 * jump.c (clean_barriers): Skip debug insns.
2625 * omp-expand.c (expand_parallel_call): Skip debug insns.
2626 (expand_task_call): Likewise.
2627 (remove_exit_barrier): Likewise.
2628 (expand_omp_taskreg): Likewise.
2629 (expand_omp_for_init_counts): Likewise.
2630 (expand_omp_for_generic): Likewise.
2631 (expand_omp_for_static_nochunk): Likewise.
2632 (expand_omp_for_static_chunk): Likewise.
2633 (expand_omp_simd): Likewise.
2634 (expand_omp_taskloop_for_outer): Likewise.
2635 (expand_omp_taskloop_for_inner): Likewise.
2636 (expand_oacc_for): Likewise.
2637 (expand_omp_sections): Likewise.
2638 (expand_omp_single): Likewise.
2639 (expand_omp_synch): Likewise.
2640 (expand_omp_atomic_load): Likewise.
2641 (expand_omp_atomic_store): Likewise.
2642 (expand_omp_atomic_fetch_op): Likewise.
2643 (expand_omp_atomic_pipeline): Likewise.
2644 (expand_omp_atomic_mutex): Likewise.
2645 (expand_omp_target): Likewise.
2646 (grid_expand_omp_for_loop): Likewise.
2647 (grid_expand_target_grid_body): Likewise.
2648 (build_omp_regions_1): Likewise.
2649 * omp-low.c (check_combined_parallel): Skip debug stmts.
2650 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2651 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2652 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2654 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2655 (make_edges): Likewise.
2656 (cleanup_dead_labels): Likewise.
2657 (gimple_can_merge_blocks_p): Likewise.
2658 (stmt_starts_bb_p): Likewise.
2659 (gimple_block_label): Likewise.
2660 (gimple_redirect_edge_and_branch): Likewise.
2661 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2663 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2665 * tree-pretty-print (print_declaration): Omit initializer in slim
2667 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2669 (eliminate_unnecessary_stmts): Stabilize block removal order.
2670 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2671 * var-tracking.c (get_first_insn): New.
2672 (vt_emit_notes): Call it.
2673 (vt_initialize): Walk any insns before the first BB.
2674 (delete_debug_insns): Likewise.
2676 * gimple.h (enum gimple_debug_subcode): Add
2677 GIMPLE_DEBUG_BEGIN_STMT.
2678 (gimple_debug_begin_stmt_p): New.
2679 (gimple_debug_nonbind_marker_p): New.
2680 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2681 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2682 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2683 * insn-notes.def (BEGIN_STMT): New.
2684 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2685 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2686 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2687 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2688 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2689 (INSN_DEBUG_MARKER_KIND): New.
2690 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2691 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2692 (INSN_VAR_LOCATION_PTR): New.
2693 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2695 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2696 deps only if debug bind insns are enabled.
2697 (pass_expand::execute): Avoid deep TER and expand
2698 debug locations for debug bind insns only.
2699 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2700 debug stmts special handling down to debug bind stmts.
2701 * combine.c (try_combine): Narrow debug insns special handling
2702 down to debug bind insns.
2703 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2704 Narrow debug insns preexisting special handling down to debug
2706 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2707 handling down to debug bind insns.
2708 * function.c (instantiate_virtual_regs): Skip debug markers,
2709 adjust handling of debug binds.
2710 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2711 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2712 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2713 insns to debug bind insns.
2714 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2715 special handling of debug stmts to debug bind stmts.
2716 * ipa-split.c (split_function): Likewise.
2717 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2718 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2720 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2721 * regrename.c (build_def_use): Likewise.
2722 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2723 (pass_cprop_hardreg): Narrow special casing of debug insns to
2725 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2726 * reload1.c (reload): Likewise.
2727 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2728 casing of debug stmts to debug bind stmts.
2729 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2730 * ssa-iterators.h (num_imm_uses): Likewise.
2731 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2732 debug stmts to debug bind stmts.
2733 * tree-inline.c (tree_function_versioning): Narrow special casing
2734 of debug stmts to debug bind stmts.
2735 * tree-loop-distribution.c (generate_loops_for_partition):
2736 Narrow special casing of debug stmts to debug bind stmts.
2737 * tree-sra.c (analyze_access_subtree): Narrow special casing
2738 of debug stmts to debug bind stmts.
2739 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2740 stmts to debug bind stmts.
2741 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2742 casing of debug stmts to debug bind stmts.
2743 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2744 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2745 casing of debug stmts to debug bind stmts.
2746 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2748 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2749 debug stmts to debug bind stmts.
2750 (gimple_replace_ssa_lhs): Likewise.
2751 (insert_debug_temp_for_var_def): Likewise.
2752 (insert_debug_temps_for_defs): Likewise.
2753 (reset_debug_uses): Likewise.
2754 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2755 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2756 (adjust_debug_stmts): Likewise.
2757 (adjust_phi_and_debug_stmts): Likewise.
2758 (vect_do_peeling): Likewise.
2759 * tree-vect-loop.c (vect_transform_loop): Likewise.
2760 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2761 * var-tracking.c (adjust_mems): Narrow special casing of debug
2762 insns to debug bind insns.
2763 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2764 (compute_bb_dataflow, vt_find_locations): Likewise.
2765 (vt_expand_loc, emit_notes_for_changes): Likewise.
2766 (vt_init_cfa_base): Likewise.
2767 (vt_emit_notes): Likewise.
2768 (vt_initialize): Likewise.
2769 (vt_finalize): Likewise.
2771 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2772 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2773 (next_nonnote_nondebug_insn_bb): New.
2774 (prev_nonnote_nondebug_insn_bb): New.
2775 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2776 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2777 (prev_nonnote_nondebug_insn_bb): Declare.
2778 (next_nonnote_nondebug_insn_bb): Declare.
2779 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2780 * cfgrtl.c (get_last_bb_insn): Likewise.
2781 * lra.c (push_insns): Likewise.
2783 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2786 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2788 * selftest.c (selftest::test_runner::test_runner): New ctor.
2789 (selftest::test_runner::~test_runner): New dtor.
2790 * selftest.h (class selftest::test_runner): New class.
2792 2017-12-11 Carl Love <cel@us.ibm.com>
2794 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
2795 vec_extract_fp32_from_shortl]): Add #defines.
2796 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
2797 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
2798 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
2799 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
2800 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
2801 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
2802 * doc/extend.texi: Add documentation for the added builtins.
2804 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
2806 PR rtl-optimization/80693
2807 PR rtl-optimization/81019
2808 PR rtl-optimization/81020
2809 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
2810 are not mentioned in i3. Place the REG_UNUSED note on i2,
2811 possibly modified to REG_DEAD, if it did not originate in i3.
2813 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2815 * recog.c (store_data_bypass_p_1): New function.
2816 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
2817 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
2819 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
2821 PR rtl-optimization/83361
2822 * ifcvt.c (if_convert): Call fixup_partitions.
2824 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
2826 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2827 early folding of splat_u{8,16,32}.
2829 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2831 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
2832 output_operand_lossage first argument with capital letter.
2833 (aarch64_override_options): Don't start error and sorry first argument
2834 with capital letter.
2836 2017-12-11 Andi Kleen <ak@linux.intel.com>
2838 PR gcov-profile/83355
2839 * auto-profile.c (string_table::get_index_by_decl): Don't
2840 recurse when abstract origin points to itself.
2842 2017-12-11 Bin Cheng <bin.cheng@arm.com>
2844 PR tree-optimization/83320
2845 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
2846 (prune_datarefs_not_in_loop): Ditto.
2848 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
2850 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
2852 2017-12-10 Jakub Jelinek <jakub@redhat.com>
2854 PR tree-optimization/83337
2855 * gimple-loop-interchange.cc (compute_access_stride): Handle
2856 bitfield DRs properly.
2858 2017-12-09 Jakub Jelinek <jakub@redhat.com>
2860 PR tree-optimization/83338
2861 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
2862 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
2863 vector of unsigned integers to vector of signed integers.
2865 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
2867 PR rtl-optimization/83317
2868 * lra-constraints.c (process_address_1): Add insn code check.
2870 2017-12-08 Michael Matz <matz@suse.de>
2872 Fix PR tree-optimization/83323
2873 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
2874 head-controlled loops and loop BBs.
2875 * common.opt (funroll-and-jam): Remove, instead ...
2876 (floop-unroll-and-jam): ... reuse this option.
2877 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
2878 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
2879 (-floop-unroll-and-jam): ... this option.
2881 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2883 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
2884 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
2885 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
2887 2017-12-08 Richard Biener <rguenther@suse.de>
2890 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
2891 handle accesses outside of zero-sized vars.
2893 2017-12-08 Martin Jambor <mjambor@suse.cz>
2895 PR tree-optimization/83141
2896 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
2897 test for MEM_REFs implicitely changing types with padding. Remove
2899 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
2901 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2903 * config/arc/arc.c (arc_attribute_table): Add exclusions to
2905 * config/avr/avr.c (avr_attribute_table): Likewise.
2906 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2907 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2908 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2909 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
2910 of struct attribute_spec.
2911 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2913 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
2916 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
2918 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
2921 * profile-count.c (profile_count::from_gcov_type): Move from
2922 profile-count.h; handle overflow.
2923 * profile-count.h (profile_count::from_gcov_type): Move offline.
2925 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
2927 PR rtl-optimization/83304
2928 * combine.c (move_deaths): If we do not know where a register died,
2931 2017-12-08 Richard Biener <rguenther@suse.de>
2933 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2934 Provide -fopt-info-loop feedback when we interchange in a nest.
2936 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2938 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
2939 for armv6 ARM CPU IDs.
2941 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2943 * common/config/arm/arm-common.c: Include <algorithm>.
2944 (INCLUDE_VECTOR): Define.
2945 (compare_opt_names): New function.
2946 (arm_rewrite_selected_arch): Only strip out extensions that can be
2947 expressed through -mfpu. Sort the remaining extensions
2950 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2952 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
2953 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
2954 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
2955 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
2957 2017-06-08 Tristan Gingold <gindold@adacore.com>
2960 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
2961 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
2963 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2965 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
2966 of struct attribute_spec.
2968 2017-12-08 Julia Koval <julia.koval@intel.com>
2970 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
2971 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
2972 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
2973 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
2974 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
2975 _mm_maskz_dpwssds_epi32): Ditto.
2977 2017-12-08 Richard Biener <rguenther@suse.de>
2979 PR tree-optimization/81303
2980 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
2981 conditions try to create a comparison vector type matching
2982 the data vector type.
2983 (vectorizable_condition): Adjust.
2984 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
2985 Leave invariant conditions alone in case we can vectorize those.
2987 2017-12-08 Julia Koval <julia.koval@intel.com>
2989 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
2990 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
2991 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
2992 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
2993 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
2995 2017-12-08 Julia Koval <julia.koval@intel.com>
2997 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
2998 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
2999 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3000 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3001 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3002 _mm_maskz_dpbusds_epi32): New intrinsics.
3004 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3006 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3008 (ld<bh>io_signed): Likewise.
3009 (st<bhw>io): Likewise.
3010 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3011 SMALL_INT12 constant integer operand.
3013 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3015 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3017 (nios2_large_constant_allowed): ...this. Adjust uses.
3018 (nios2_plus_symbolic_constant_p): Rename to...
3019 (nios2_plus_large_constant_p): ...this. Adjust uses.
3020 (nios2_legitimate_address_p): Correct CONST_INT handling.
3021 (nios2_symbolic_memory_operand_p): Rename to...
3022 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3023 (nios2_large_constant_p): Check for large constant integers too.
3024 (nios2_split_large_constant): Handle constant integers.
3025 (nios2_split_symbolic_memory_operand): Rename to...
3026 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3027 (nios2_legitimize_constant_address): Handle constant integers.
3028 (r0rel_constant_p): Handle small constant integers.
3029 (nios2_print_operand_address): Handle r0-relative integer addresses.
3030 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3031 * config/nios2/nios2.md: Adjust for renamed functions.
3033 2017-12-07 Andrew Waterman <andrew@sifive.com>
3035 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3036 (TARGET_HAVE_SRODATA_SECTION): New define.
3037 (riscv_select_section): New function.
3039 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3040 Alexander Monakov <amonakov@ispras.ru>
3041 Jakub Jelinek <jakub@redhat.com>
3044 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3046 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3048 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3049 a bool success value. Don't call output_operand_lossage here.
3050 (aarch64_print_ldpstp_address): Return a bool success value.
3051 (aarch64_print_operand_address): Call output_addr_const if
3052 aarch64_print_address_internal fails.
3053 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3054 'y'; call output_operand_lossage instead. Call output_operand_lossage
3055 if aarch64_print_ldpstp_address fails.
3057 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3059 * tree-vector-builder.h
3060 (tree_vector_builder::binary_encoded_nelts): Declare.
3061 * tree-vector-builder.c
3062 (tree_vector_builder::binary_encoded_nelts): New function.
3063 * fold-const.c (negate_expr_p): Likewise.
3064 (operand_equal_p, fold_checksum_tree): Likewise.
3065 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3066 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3067 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3068 (uniform_vector_p): Likewise.
3069 * varasm.c (const_hash_1, compare_constant): Likewise.
3070 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3071 (valid_lattice_transition): Operate directly on the VECTOR_CST
3073 * ipa-icf.c: Include tree-vector-builder.h.
3074 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3075 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3077 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3079 * tree.c (build_vector): Delete.
3080 * tree.h (build_vector): Make static and move into the self-testing
3083 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3085 * vector-builder.h (vector_builder::derived): New const overload.
3086 (vector_builder::elt): New function.
3087 * tree-vector-builder.h (tree_vector_builder::type): New function.
3088 (tree_vector_builder::apply_step): Declare.
3089 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3091 * gimple-fold.h (tree_vector_builder): Declare.
3092 (gimple_build_vector): Take a tree_vector_builder instead of a
3093 type and vector of elements.
3094 * gimple-fold.c (gimple_build_vector): Likewise.
3095 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3097 (get_initial_defs_for_reduction): Likewise.
3098 (vectorizable_induction): Likewise.
3100 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3102 * tree-vector-builder.h
3103 (tree_vector_builder::new_binary_operation): Declare.
3104 * tree-vector-builder.c
3105 (tree_vector_builder::new_binary_operation): New function.
3106 * fold-const.c (fold_relational_const): Use it.
3107 (const_binop): Likewise. Check that both input vectors have
3108 the same number of elements, thus excluding things like WIDEN_SUM.
3109 Check whether it is possible to operate directly on the encodings
3112 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3114 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3115 new_unary_operation, operating only on the encoded elements.
3116 (const_unop): Likewise.
3117 (exact_inverse): Likewise.
3118 (distributes_over_addition_p): New function.
3119 (const_binop): Use tree_vector_builder and new_unary_operation
3120 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3121 on the encoded elements unless the encoding is strided and the
3122 operation does not distribute over addition.
3123 (fold_convert_const): Use tree_vector_builder and
3124 new_unary_operation. Operate only on the encoded elements
3125 for truncating integer conversions, or for non-stepped encodings.
3127 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3129 * config/sparc/sparc.c: Include tree-vector-builder.h.
3130 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3131 * expmed.c: Include tree-vector-builder.h.
3132 (make_tree): Use tree_vector_builder instead of build_vector.
3133 * fold-const.c: Include tree-vector-builder.h.
3134 (const_binop): Use tree_vector_builder instead of build_vector.
3135 (const_unop): Likewise.
3136 (native_interpret_vector): Likewise.
3137 (fold_vec_perm): Likewise.
3138 (fold_ternary_loc): Likewise.
3139 * gimple-fold.c: Include tree-vector-builder.h.
3140 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3142 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3143 (simplify_vector_constructor): Use tree_vector_builder instead
3145 * tree-vect-generic.c: Include tree-vector-builder.h.
3146 (add_rshift): Use tree_vector_builder instead of build_vector.
3147 (expand_vector_divmod): Likewise.
3148 (optimize_vector_constructor): Likewise.
3149 * tree-vect-loop.c: Include tree-vector-builder.h.
3150 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3151 of build_vector. Explicitly use a stepped encoding for
3153 * tree-vect-slp.c: Include tree-vector-builder.h.
3154 (vect_get_constant_vectors): Use tree_vector_builder instead
3156 (vect_transform_slp_perm_load): Likewise.
3157 (vect_schedule_slp_instance): Likewise.
3158 * tree-vect-stmts.c: Include tree-vector-builder.h.
3159 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3160 (vect_gen_perm_mask_any): Likewise.
3161 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3162 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3164 (build_vector_from_val): Likewise. Explicitly use a duplicate
3167 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3169 * doc/generic.texi (VECTOR_CST): Describe new representation of
3171 * vector-builder.h: New file.
3172 * tree-vector-builder.h: Likewise.
3173 * tree-vector-builder.c: Likewise.
3174 * Makefile.in (OBJS): Add tree-vector-builder.o.
3175 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3176 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3177 (tree_vector): Change the number of elements to
3178 vector_cst_encoded_nelts.
3179 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3180 (VECTOR_CST_ELTS): Delete.
3181 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3182 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3183 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3184 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3185 (VECTOR_CST_ENCODED_ELT): Likewise.
3186 (vector_cst_encoded_nelts): New function.
3187 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3188 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3189 (vector_cst_int_elt, vector_cst_elt): Declare.
3190 * tree.c: Include tree-vector-builder.h.
3191 (tree_code_size): Abort if passed VECTOR_CST.
3192 (tree_size): Update for new VECTOR_CST layout.
3193 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3194 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3195 (build_vector): Use tree_vector_builder.
3196 (vector_cst_int_elt, vector_cst_elt): New functions.
3197 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3198 encoded elements and then create the vector in the canonical form.
3199 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3200 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3201 (tree_c_tests): Call test_vector_cst_patterns.
3202 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3204 (hash_tree): Likewise.
3205 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3206 (streamer_write_tree_header): Likewise.
3207 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3208 (streamer_alloc_tree): Likewise. Update call to make_vector.
3209 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3211 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3213 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3214 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3215 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3217 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3218 Richard Biener <rguenther@suse.de>
3220 PR tree-optimization/81303
3221 * Makefile.in (gimple-loop-interchange.o): New object file.
3222 * common.opt (floop-interchange): Reuse the option from graphite.
3223 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3224 -floop-interchange and mention it for -O3.
3225 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3226 * gimple-loop-interchange.cc: New file.
3227 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3228 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3229 * passes.def (pass_linterchange): New pass.
3230 * timevar.def (TV_LINTERCHANGE): New time var.
3231 * tree-pass.h (make_pass_linterchange): New declaration.
3232 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3233 interchange. Record IV before/after increment in new parameters.
3234 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3235 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3237 (check_reduction_path): ...New function here.
3238 * tree-vectorizer.h (check_reduction_path): New declaration.
3240 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3243 PR rtl-optimization/80818
3244 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3245 always early clobbered.
3246 * lra-lives.c (process_bb_lives): Check input hard regs for early
3247 clobbered non-operand hard reg.
3249 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3252 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3253 types_compatible_p, just that TYPE_MODE is the same.
3255 2017-12-07 Martin Sebor <msebor@redhat.com>
3258 * attribs.c (empty_attribute_table): Initialize new member of
3259 struct attribute_spec.
3260 (decl_attributes): Add argument. Handle mutually exclusive
3261 combinations of attributes.
3262 (selftests::test_attribute_exclusions): New function.
3263 (selftests::attribute_c_tests): Ditto.
3264 * attribs.h (decl_attributes): Add default argument.
3265 * selftest.h (attribute_c_tests): Declare.
3266 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3267 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3269 * doc/extend.texi (Common Function Attributes): Update const and pure.
3270 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3271 of struct attribute_spec.
3272 * config/arc/arc.c (arc_attribute_table): Same.
3273 * config/arm/arm.c (arm_attribute_table): Same.
3274 * config/avr/avr.c ( avr_attribute_table): Same.
3275 * config/bfin/bfin.c (bfin_attribute_table): Same.
3276 * config/cr16/cr16.c (cr16_attribute_table): Same.
3277 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3278 * config/h8300/h8300.c (h8300_attribute_table): Same.
3279 * config/i386/i386.c (ix86_attribute_table): Same.
3280 * config/ia64/ia64.c (ia64_attribute_table): Same.
3281 * config/m32c/m32c.c (m32c_attribute_table): Same.
3282 * config/m32r/m32r.c (m32r_attribute_table): Same.
3283 * config/m68k/m68k.c (m68k_attribute_table): Same.
3284 * config/mcore/mcore.c (mcore_attribute_table): Same.
3285 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3286 * config/mips/mips.c (mips_attribute_table): Same.
3287 * config/msp430/msp430.c (msp430_attribute_table): Same.
3288 * config/nds32/nds32.c (nds32_attribute_table): Same.
3289 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3290 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3291 * config/rl78/rl78.c (rl78__attribute_table): Same.
3292 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3293 * onfig/rx/rx.c (rx_attribute_table): Same.
3294 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3295 * config/sh/sh.c (sh_attribute_table): Same.
3296 * config/sparc/sparc.c (sparc_attribute_table): Same.
3297 * config/spu/spu.c (spu_attribute_table): Same.
3298 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3299 * config/v850/v850.c (v850_attribute_table): Same.
3300 * config/visium/visium.c (visium_attribute_table): Same.
3302 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3305 * config/arm/arm.c (INCLUDE_STRING): Define.
3306 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3307 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3310 2017-12-07 Michael Matz <matz@suse.de>
3312 Add unroll and jam pass
3314 * gimple-loop-jam.c: New file.
3315 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3316 * common.opt (funroll-and-jam): New option.
3317 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3318 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3319 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3320 * passes.def: Add pass_loop_jam.
3321 * timevar.def (TV_LOOP_JAM): Add.
3322 * tree-pass.h (make_pass_loop_jam): Declare.
3323 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3324 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3325 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3326 to flow_loop_tree_node_add.
3327 (duplicate_subloops, copy_loops_to): Append to sibling list.
3328 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3329 * doc/invoke.texi (-funroll-and-jam): Document new option.
3330 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3332 2017-12-07 Richard Biener <rguenther@suse.de>
3334 PR tree-optimization/83296
3335 PR tree-optimization/67769
3336 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3337 flow sensitive info in an unrelated BB.
3338 (value_replacement): Use reset_flow_sensitive_info.
3339 (minmax_replacement): Reset flow sensitive info on the def
3340 we move. Do not reset flow sensitive info in the whole BB
3341 we move the stmt to.
3342 (abs_replacement): Likewise.
3344 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3347 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3348 rs6000_cpu to the given -mcpu=, or to the default processor.
3350 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3352 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3353 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3354 instead of rs6000_cpu_attr.
3355 (rs6000_adjust_cost): Ditto.
3356 (is_microcoded_insn): Ditto.
3357 (rs6000_adjust_priority): Ditto.
3358 (rs6000_issue_rate): Ditto.
3359 (rs6000_use_sched_lookahead): Ditto.
3360 (rs6000_use_sched_lookahead_guard): Ditto.
3361 (rs6000_sched_reorder): Ditto.
3362 (force_new_group): Ditto.
3363 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3364 (group_ending_nop): Ditto.
3366 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3368 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3369 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3370 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3371 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3373 (rs6000_reassociation_width): Ditto.
3374 (rs6000_emit_epilogue): Ditto.
3375 (rs6000_adjust_cost): Ditto.
3376 (is_microcoded_insn): Ditto.
3377 (is_cracked_insn): Ditto.
3378 (rs6000_adjust_priority): Ditto.
3379 (rs6000_sched_reorder): Ditto.
3380 (rs6000_sched_reorder2): Ditto.
3381 (insn_must_be_first_in_group): Ditto.
3382 (insn_must_be_last_in_group): Ditto.
3383 (rs6000_register_move_cost): Ditto.
3384 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3387 2017-12-07 Julia Koval <julia.koval@intel.com>
3389 * config.gcc: Add vaesintrin.h.
3390 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3392 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3393 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3395 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3396 * config/i386/immintrin.h: Include vaesintrin.h.
3397 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3398 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3399 _mm_aesdec_epi128): New intrinsics.
3401 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3403 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3404 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3405 c-family/c-spellcheck.cc.
3406 (best_macro_match::best_macro_match): Likewise.
3408 (struct edit_distance_traits<cpp_hashnode *>): Move to
3409 c-family/c-spellcheck.h.
3410 (class best_macro_match): Likewise.
3412 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3414 PR tree-optimization/83293
3415 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3416 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3417 might insert into empty bb.
3420 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3421 simplify for plus with :c added, and pointer_plus without that.
3422 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3423 with undefined overflow and the conversion is not widening,
3424 perform negation in utype and only convert to type afterwards.
3425 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3426 simplify for plus with :c added, and pointer_plus without that.
3427 If type is integral with undefined overflow and the conversion is
3428 not widening, perform minus in utype and only convert to type
3429 afterwards. Move the last pointer_diff_expr simplify into the
3432 2017-12-06 Martin Sebor <msebor@redhat.com>
3434 PR tree-optimization/82646
3435 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3436 strncpy, not maxlen.
3438 2017-12-06 Martin Sebor <msebor@redhat.com>
3440 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3443 PR tree-optimization/83075
3444 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3445 strncat/strncpy don't change length of source string.
3447 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3450 2017-11-29 Martin Aberg <maberg@gaisler.com>
3452 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3453 to prevent b2bst errata sequence.
3454 (sqrtdf2_fix): Likewise.
3456 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3458 PR tree-optimization/81945
3459 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3460 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3461 to dest_cfun has orig_loop_num set, either remap it to the new
3462 loop number if the loop got moved too, or clear it.
3464 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3466 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3468 (thunderx2t99_nothing): New insn reservation.
3469 (thunderx2t99_mrs): New insn reservation.
3470 (thunderx2t99_multiple): New insn reservation.
3471 (thunderx2t99_alu_basi): Add bfx to reservation.
3472 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3474 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3477 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3479 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3481 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3482 tree-ssa-pre.c::remove_dead_inserted_code.
3483 * tree-ssa-dce.h: New file.
3484 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3485 (remove_dead_inserted_code): Move and rename to function
3486 tree-ssa-dce.c::simple_dce_from_worklist.
3487 (pass_pre::execute): Update use.
3489 2017-12-05 Richard Biener <rguenther@suse.de>
3491 PR tree-optimization/83277
3492 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3493 to code-gen liveout vars.
3495 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3497 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3498 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3499 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3502 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3504 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3506 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3507 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3510 2017-12-05 Richard Biener <rguenther@suse.de>
3512 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3513 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3514 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3515 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3517 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3519 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3521 2017-12-05 Martin Liska <mliska@suse.cz>
3522 Jakub Jelinek <jakub@redhat.com>
3524 * doc/invoke.texi: Document the options.
3525 * flag-types.h (enum sanitize_code): Add
3526 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3527 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3528 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3529 * opts.c: Define new sanitizer options.
3530 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3531 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3533 2017-12-05 Julia Koval <julia.koval@intel.com>
3535 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3536 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3537 (ix86_handle_option): Handle -mavx512vnni.
3538 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3539 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3540 * config/i386/i386-c (__AVX512VNNI__): New.
3541 * config/i386/i386.c (ix86_target_string): Handle new option.
3542 (ix86_valid_target_attribute_inner_p): Handle new option.
3543 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3544 * config/i386/i386.opt (mavx512vnni): New option.
3546 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3549 * athlon.md: Disable for generic.
3550 * haswell.md: Enable for generic.
3551 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3552 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3554 (ix86_adjust_cost): Move generic to haswell path.
3556 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3558 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3559 instead of MEM_P in a couple more places. Fix formatting issues.
3561 2017-12-04 Jim Wilson <jimw@sifive.com>
3563 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3564 instead of GP_REG_LAST-1.
3565 (riscv_adjust_libcall_cfi_prologue): Likewise.
3566 (riscv_adjust_libcall_cri_epilogue): Likewise.
3567 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3570 2017-12-04 Luis Machado <luis.machado@linaro.org>
3572 * ipa-pure-const.c (check_decl): Add missing newline.
3573 (state_from_flags): Likewise.
3575 2017-12-04 Jeff Law <law@redhat.com>
3577 PR tree-optimizatin/78496
3578 * gimple-ssa-evrp-analyze.h
3579 (evrp_range_analyzer::get_vr_values): Simplify.
3580 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3581 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3582 and gimple-ssa-evrp-analyze.h.
3583 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3584 (simplify_stmt_for_jump_threading): Copy a blob of code from
3585 tree-vrp.c to use ranges to simplify statements.
3586 (dom_opt_dom_walker::before_dom_children): Call
3587 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3588 (dom_opt_dom_walker::after_dom_children): Similarly for
3589 evrp_range_analyzer::leave.
3590 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3593 * gimple-ssa-evrp-analyze.c
3594 (evrp_range_analyzer::extract_range_from_stmt): Always use
3595 vr_values::update_value_range so preexisting range info is
3596 medged with new range info, even if the new range is VR_VARYING.
3598 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3600 * combine.c: Adjust comment.
3601 (use_crosses_set_p): Delete.
3602 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3603 (try_combine): Ditto.
3605 2017-12-04 Richard Biener <rguenther@suse.de>
3607 PR tree-optimization/83255
3608 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3609 Re-add zero-iteration check.
3611 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3613 PR rtl-optimization/83245
3614 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3615 hard registers as earlyclobber, also if not in an asm.
3617 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3621 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3624 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3625 rs6000_cpu based on cpu_index, not tune_index.
3627 2017-12-04 Richard Biener <rguenther@suse.de>
3629 PR tree-optimization/83238
3630 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3631 code match comment, rejecting invalid SESE regions.
3633 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3635 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3636 require base operand is a REG_POINTER prior to reload on targets
3637 with non-equivalent space registers.
3639 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3641 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3642 (update_specialized_profile): Fix updating of counts.
3643 (perhaps_add_new_callers): Likewise.
3645 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3648 * x86-tune.def: Remove obsolette FIXMEs.
3649 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3650 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3651 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3653 (X86_TUNE_PAD_RETURNS): Disable for generic.
3655 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3657 PR tree-optimization/83170
3658 PR tree-optimization/83241
3659 * gimple-ssa-store-merging.c
3660 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3661 gimple_vuse (ins_stmt) in case it has changed.
3662 (imm_store_chain_info::output_merged_store): Likewise.
3664 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3668 * tree-cfg.c (pass_warn_function_return::execute): Handle
3669 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3670 with BUILTINS_LOCATION.
3674 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3675 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3676 instead of DECL_MODE.
3678 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3679 last argument to rtx pointer.
3680 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3681 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3682 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3683 depending on the chosen ISAs.
3684 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3685 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3686 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3688 * config/i386/sse.md (mov<mode>_internal): Likewise.
3689 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3691 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3693 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3695 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3697 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3698 parameters from prototype.
3699 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3700 parameters. Don't print a comment.
3701 (emit_fusion_gpr_load): Adjust.
3702 (emit_fusion_load_store): Adjust.
3703 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3704 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3705 comment on the second line.
3707 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3710 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3711 rs6000_cpu based on cpu_index, not tune_index.
3713 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3715 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3716 which_alternative instead of which_alternative + 1.
3717 (output_asm_insn): Print an extra tab if the template is short.
3719 2017-12-01 Jim Wilson <jimw@sifive.com>
3721 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3723 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3724 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3726 * doc/tm.texi: Regenerate.
3728 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3731 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3732 whether we can allocate pseudos before trying to fix an address.
3733 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3734 memory address is indexed or indirect.
3735 (floatuns_<mode>si2_hw2): Likewise.
3737 2017-12-01 Jason Merrill <jason@redhat.com>
3739 * Makefile.in (TAGS): Add c-family/*.cc.
3741 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3743 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3744 (call_value_insn): Likewise.
3745 (sibcall_insn): Likewise.
3746 (sibcall_value_insn): Likewise.
3747 (movsi_aarch64): Likewise.
3748 (movdi_aarch64): Likewise.
3749 (add_losym_): Likewise.
3750 (ldr_got_small_): Likewise.
3751 (ldr_got_small_sidi): Likewise.
3752 (ldr_got_small_28k_): Likewise.
3753 (ldr_got_small_28k_sidi): Likewise.
3754 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3755 Move output_addr_const to symbolic case. Add error check.
3757 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3759 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3760 * config/s390/s390.md: Disable MVC merging peephole if it would
3761 disable operand forwarding.
3762 (new peephole2): Split MVCs if it would turn them into up to 2
3765 2017-12-01 Richard Biener <rguenther@suse.de>
3767 PR tree-optimization/83232
3768 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3769 detection of same access. Instead of breaking the group here
3770 do not consider the duplicate. Add comment explaining real fix.
3772 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3774 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3776 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3778 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3780 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3782 * function.h (struct function): Remove cilk_frame_decl,
3783 is_cilk_function and calls_cilk_spawn fields.
3784 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3786 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3787 cilk_elemental field.
3788 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3789 * target.def: Adjust comment.
3790 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3791 Don't test cilk_elemental.
3793 PR tree-optimization/83233
3794 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
3795 bswap_stat name for the struct.
3798 * tree.h (SWITCH_BREAK_LABEL_P): Define.
3799 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
3800 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
3801 SWITCH_BREAK_LABEL_P set on the label.
3802 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
3803 added for default case if it was missing and not all cases covered.
3804 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
3805 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
3807 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
3808 to avoid -Wimplicit-fallthrough warning.
3809 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
3810 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
3812 PR tree-optimization/83221
3813 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
3815 (init_reassoc): Formatting fix.
3818 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
3819 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
3823 * tree-cfg.c: Include asan.h.
3824 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
3826 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3828 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
3830 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3832 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
3833 vpcmpeqd instruction.
3835 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3837 * config/i386/i386.c (standard_sse_constant_opcode): Fix
3838 registers type for 128bit mode.
3840 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3842 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
3844 * gengtype-state.c (read_state_pair): Likewise.
3845 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
3846 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
3847 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
3848 * ifcvt.c (noce_try_store_flag_constants): Likewise.
3849 * tree-ssa-ccp.c (ccp_finalize): Likewise.
3850 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
3851 * builtins.c (fold_builtin_3): Likewise.
3852 * graphite-scop-detection.c
3853 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
3854 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
3857 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
3860 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
3861 parameter. Return true if ignore_decl_rtl_set_p is true and other
3862 conditions are satisfied.
3863 * asan.h (asan_protect_global): Add new parameter.
3864 * varasm.c (categorize_decl_for_section): Pass true as second parameter
3865 to asan_protect_global calls.
3867 2017-11-30 Jim Wilson <jimw@sifive.com>
3869 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
3870 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
3871 -mstrict-align, add info on default value. Delete redundant lines for
3872 -mabi. Add missing -mexplicit-relocs docs.
3874 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3876 * config/arc/arc.md (trap): New pattern.
3878 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3880 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
3881 instruction to end into a delay slot.
3882 * config/arc/arc.md (cond_delay_insn): Check if the instruction
3883 can be placed into a delay slot against reg_note.
3885 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3887 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
3888 labels number of usages.
3890 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3892 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
3894 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
3896 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3899 * internal-fn.c (expand_mul_overflow): Optimize unsigned
3900 multiplication by power of 2 constant into two shifts + comparison.
3902 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
3905 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
3907 2017-11-30 Richard Biener <rguenther@suse.de>
3909 PR tree-optimization/83202
3910 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3911 allow_peel argument and guard peeling.
3912 (canonicalize_loop_induction_variables): Likewise.
3913 (canonicalize_induction_variables): Pass false.
3914 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
3915 peeling from cunrolli.
3917 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3919 * combine.c (try_combine): Print a message to dump file whenever
3920 I0, I1, or I2 cannot be combined into I3.
3922 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3924 PR rtl-optimization/83156
3925 PR rtl-optimization/82621
3926 * combine.c (try_combine): Don't split an I2 if one of the dests is
3927 set again before I3. Allow unused dests.
3929 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3931 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
3933 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
3935 PR rtl-optimization/80818
3936 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
3937 recursively. Use insn code for clobber.
3938 (lra_set_insn_recog_data): Pass the new arg to
3939 collect_non_operand_hard_regs.
3940 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
3942 (lra_update_insn_regno_info): Pass insn to
3943 add_regs_to_insn_regno_info.
3945 2017-11-29 Jim Wilson <jimw@sifive.com>
3946 Andrew Waterman <andrew@sifive.com>
3948 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
3949 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
3950 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
3951 use SINGLE_SHIFT_COST.
3952 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
3953 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
3955 2017-11-29 Julia Koval <julia.koval@intel.com>
3957 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
3958 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
3959 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
3960 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
3961 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
3962 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
3963 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
3964 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
3965 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
3966 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
3967 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
3968 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
3969 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
3970 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
3971 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
3972 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
3973 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
3974 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
3975 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
3976 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
3977 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
3978 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
3979 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
3980 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
3981 __builtin_ia32_vpshldv_v2di_mask,
3982 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
3983 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
3984 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
3986 2017-11-29 Julia Koval <julia.koval@intel.com>
3988 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
3989 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
3990 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
3991 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
3992 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
3993 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
3994 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
3995 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
3996 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
3997 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
3998 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
3999 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4000 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4001 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4002 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4003 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4004 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4005 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4006 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4007 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4008 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4010 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4012 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4013 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4016 2017-11-29 Martin Aberg <maberg@gaisler.com>
4018 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4019 to prevent b2bst errata sequence.
4020 (sqrtdf2_fix): Likewise.
4022 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4024 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4025 (div_sqrt_insn_p): New function.
4026 (sparc_do_work_around_errata): Insert NOP instructions to
4027 prevent sequences that could trigger the TN-0013 errata for
4028 certain LEON3 processors.
4029 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4030 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4031 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4032 (in_branch_delay): Prevent div and sqrt in delay slot if
4034 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4036 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4038 * config/sparc/sparc.c (atomic_insn_p): New function.
4039 (sparc_do_work_around_errata): Insert NOP instructions to
4040 prevent sequences that could trigger the TN-0010 errata for
4042 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4043 instruction referable in atomic_insns_p.
4045 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4047 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4048 (atomic_compare_and_swap_leon3_1): Likewise.
4051 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4053 * config/sparc/sparc.c (fpop_insn_p): New function.
4054 (sparc_do_work_around_errata): Insert NOP instructions to
4055 prevent sequences that could trigger the TN-0012 errata for
4057 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4058 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4059 (in_branch_annul_delay): Prevent floating-point instructions
4060 in delay slot of annulled integer branch.
4062 2017-11-29 Richard Biener <rguenther@suse.de>
4064 PR tree-optimization/83202
4065 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4067 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4068 nodes vectorized to the same stmts multiple times.
4069 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4070 (vect_analyze_slp_instance): Adjust.
4071 (scalar_stmts_to_slp_tree_map_t): New typedef.
4072 (vect_schedule_slp_instance): Add a map recording the SLP node
4073 representing the vectorized stmts for a set of scalar stmts.
4074 Avoid code-generating redundancies.
4075 (vect_schedule_slp): Allocate map and pass it down.
4077 2017-11-29 Nathan Sidwell <nathan@acm.org>
4080 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4083 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4085 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4086 cases for printing LDP/STP memory addresses.
4087 (aarch64_print_address_internal): Renamed from
4088 aarch64_print_operand_address, added parameter, add Pmode check.
4089 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4090 (aarch64_print_operand_address): Indirect to
4091 aarch64_print_address_internal.
4092 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4093 'y' operand output specifier.
4095 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4098 * tree.c (build_simple_mem_ref_loc): Handle
4099 get_addr_base_and_unit_offset returning a MEM_REF.
4102 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4105 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4108 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4110 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4112 (layout::move_to_column): Likewise.
4114 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4117 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4118 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4120 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4121 Martin Jambor <mjambor@suse.cz>
4124 * tree.h (expr_type_first_operand_type_p): Declare
4125 * tree.c (expr_type_first_operand_type_p): New function.
4126 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4127 (ipa_value_from_jfunc): Adjust declaration.
4128 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4129 Use it as result type for arithmetics, unless it is NULL in which case
4130 be more conservative.
4131 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4132 ipa_get_jf_pass_through_result.
4133 (propagate_vals_across_pass_through): Likewise.
4134 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4135 is to propagate_vals_across_pass_through.
4136 (propagate_constants_across_call): Pass PARM_TYPE to
4137 propagate_scalar_across_jump_function.
4138 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4139 ipa_value_from_jfunc.
4140 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4141 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4142 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4143 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4144 target_type, pass it to ipa_value_from_jfunc.
4145 (update_indirect_edges_after_inlining): Pass parameter type to
4146 try_make_edge_direct_simple_call.
4148 2017-11-28 Jeff Law <law@redhat.com>
4150 * gimple-ssa-evrp-analyze.c
4151 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4152 refine ranges if scev_initialized_p returns true.
4153 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4155 2017-11-28 Julia Koval <julia.koval@intel.com>
4157 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4158 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4159 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4160 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4161 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4162 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4163 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4164 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4165 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4166 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4167 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4168 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4169 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4170 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4171 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4172 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4173 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4174 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4175 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4176 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4177 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4178 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4180 2017-11-28 Julia Koval <julia.koval@intel.com>
4182 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4183 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4184 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4185 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4186 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4187 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4188 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4189 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4190 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4191 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4192 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4193 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4194 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4195 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4196 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4197 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4198 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4199 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4200 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4201 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4202 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4203 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4204 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4205 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4206 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4207 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4208 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4209 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4210 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4211 __builtin_ia32_vpshld_v2di_mask): New builtins.
4212 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4213 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4215 2017-11-28 Richard Biener <rguenther@suse.de>
4217 PR tree-optimization/80776
4218 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4220 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4222 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4223 If the incoming edge is an effective fallthru because the other
4224 edge only reaches a __builtin_unreachable () then record ranges
4225 derived from the controlling condition in SSA info.
4226 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4227 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4229 2017-11-28 Olivier Hainque <hainque@adacore.com>
4231 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4232 on mingw build hosts.
4234 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4237 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4239 2017-11-28 Richard Biener <rguenther@suse.de>
4242 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4243 copies generated from memcpy use a character array as reference
4246 2017-11-28 Julia Koval <julia.koval@intel.com>
4247 Sebastian Peryt <sebastian.peryt@intel.com>
4249 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4250 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4251 cilk-common.o, cilk.h, cilk-common.c): Remove.
4253 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4254 * builtins.c (is_builtin_name): Remove cilkplus condition.
4255 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4256 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4257 cilk-builtins.def, cilkplus.def): Remove.
4258 * cif-code.def (CILK_SPAWN): Remove.
4259 * cilk-builtins.def: Delete.
4260 * cilk-common.c: Ditto.
4262 * cilkplus.def: Ditto.
4263 * config/darwin.h (fcilkplus): Delete.
4264 * cppbuiltin.c: Ditto.
4265 * doc/extend.texi: Remove cilkplus doc.
4266 * doc/generic.texi: Ditto.
4267 * doc/invoke.texi: Ditto.
4268 * doc/passes.texi: Ditto.
4269 * gcc.c (fcilkplus): Remove.
4270 * gengtype.c (cilk.h): Remove.
4271 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4273 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4275 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4277 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4278 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4279 cilkplus conditions.
4280 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4281 inline_read_section): Ditto.
4282 * ipa-inline-analysis.c (cilk.h): Remove.
4283 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4284 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4285 append_linker_options): Remove condition for fcilkplus.
4286 * lto/lto-lang.c (cilk.h): Remove.
4287 (lto_init): Remove condition for fcilkplus.
4288 * omp-expand.c (expand_cilk_for_call): Delete.
4289 (expand_omp_taskreg, expand_omp_for_static_chunk,
4290 expand_omp_for): Remove cilkplus
4292 (expand_cilk_for): Delete.
4293 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4294 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4295 execute_lower_omp, diagnose_sb_0): Ditto.
4296 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4297 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4298 * tree-nested.c: Ditto.
4299 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4300 (dump_generic_node): Ditto.
4301 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4302 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4304 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4306 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4308 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4309 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4310 for vector mode and !TARGET_SIMD.
4312 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4314 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4316 * tree.h (SWITCH_LABELS): Remove.
4317 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4318 assert SWITCH_BODY is non-NULL.
4319 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4321 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4323 PR tree-optimization/80788
4324 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4325 has TREE_OVERFLOW set, call drop_tree_overflow.
4327 2017-11-28 Richard Biener <rguenther@suse.de>
4329 PR tree-optimization/83158
4330 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4332 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4335 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4336 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4338 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4340 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4341 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4343 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4346 * varasm.c (compare_constant): Take the mode of the constants into
4347 account when comparing floating point constants.
4349 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4351 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4352 from explicit instantiation of debug_helper.
4353 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4355 2017-11-27 Richard Biener <rguenther@suse.de>
4357 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4360 2017-11-27 Richard Biener <rguenther@suse.de>
4362 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4363 (type_hash_canon): Also clear int_cst_hash_table entry for
4365 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4367 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4369 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4370 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4371 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4373 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4375 * hash-map.h (gt_cleare_cache): Avoid UB.
4377 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4379 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4380 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4381 (decide_unroll_runtime_iterations): Likewise.
4382 (decide_unroll_stupid): Likewise.
4384 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4387 * config/i386/i386.md: Add a loop with incssp.
4389 2017-11-27 Martin Jambor <mjambor@suse.cz>
4391 PR tree-optimization/81248
4392 * tree-sra.c (splice_param_accesses): Remove size check.
4393 (decide_one_param_reduction): Fix size check.
4394 * gimple-pretty-print.c (dump_profile): Silence warning.
4395 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4397 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4400 * dbxout.c (lastlineno): New variable.
4401 (dbx_debug_hooks): Use dbxout_switch_text_section as
4402 switch_text_section debug hook.
4403 (dbxout_function_end): Switch to current_function_section
4404 rather than function_section. If crtl->has_bb_partition,
4405 output just one N_FUN, depending on in_cold_section_p.
4406 (dbxout_source_line): Remember last lineno in lastlineno.
4407 (dbxout_switch_text_section): New function.
4408 (dbxout_function_decl): Adjust dbxout_block caller.
4409 (dbx_block_with_cold_children): New function.
4410 (dbxout_block): Return true if any LBRAC/RBRAC have been
4411 emitted. Use dbx_block_with_cold_children at depth == 0
4412 in second partition. Add PARENT_BLOCKNUM argument, pass
4413 it optionally adjusted to children. Output LBRAC/RBRAC
4414 around recursive call only if the block is in the current
4415 partition, if not and anything was output, emit empty
4417 * final.c (final_scan_insn): Compute cold_function_name
4418 before calling switch_text_section debug hook. Call
4419 that hook even if dwarf2out_do_frame if not emitting
4423 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4424 TREE_READONLY decls.
4426 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4428 PR rtl-optimization/82488
4429 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4431 2017-11-26 Julia Koval <julia.koval@intel.com>
4433 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4435 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4438 2017-11-26 Julia Koval <julia.koval@intel.com>
4440 * config/i386/driver-i386.c (host_detect_local_cpu):
4441 Detect skylake-avx512.
4443 2017-11-26 Julia Koval <julia.koval@intel.com>
4445 * config.gcc: Add -march=cannonlake.
4446 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4447 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4448 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4449 (PTA_CANNONLAKE): New.
4450 (processor_target_table): Add cannonlake.
4451 (ix86_option_override_internal): Ditto.
4452 (fold_builtin_cpu): Ditto.
4453 (get_builtin_code_for_version): Handle cannonlake.
4454 (M_INTEL_COREI7_CANNONLAKE): New.
4455 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4456 * doc/invoke.texi: Add -march=cannonlake.
4458 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4460 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4461 (try_init_one_plugin): Alternative implementation for MinGW.
4462 * Makefile.in (plugin_implib): New.
4463 (gengtype-lex.c): Fix broken AIX workaround.
4464 * configure: Regenerate.
4465 * doc/plugins.texi: Document support for MinGW.
4467 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4469 PR rtl-optimization/81553
4470 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4471 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4472 is a shift where C1 has different mode than the whole shift, use C1's
4473 mode for MULT rather than the shift's mode.
4476 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4477 builtins not enabled in the currently selected ISA.
4479 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4481 PR tree-optimization/71026
4482 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4483 (insert_reciprocals): Change to insert reciprocals before a division
4484 by a square and to insert the square of a reciprocal.
4485 (execute_cse_reciprocals_1): Change to consider division by a square.
4486 (register_division_in): Add importance parameter.
4488 2017-11-24 Richard Biener <rguenther@suse.de>
4490 PR tree-optimization/82402
4491 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4492 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4494 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4496 * match.pd (0-ptr): New transformation.
4498 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4501 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4502 when skipping checking.
4504 2017-11-24 Richard Biener <rguenther@suse.de>
4506 PR tree-optimization/83128
4507 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4508 (vn_reference_lookup_3): Likewise.
4510 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4513 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4514 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4517 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4518 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4519 check for the latter with gimple_call_builtin_p. Do not handle
4520 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4522 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4524 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4527 2017-11-23 Julia Koval <julia.koval@intel.com>
4529 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4530 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4531 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4532 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4533 _mm512_maskz_expandloadu_epi16): New intrinsics.
4534 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4535 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4536 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4537 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4538 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4539 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4540 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4541 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4542 _mm256_maskz_expandloadu_epi8): New intrinsics.
4543 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4544 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4545 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4546 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4547 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4548 * config/i386/sse.md (VI248_VLBW): New iterator.
4549 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4551 2017-11-23 Julia Koval <julia.koval@intel.com>
4553 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4554 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4555 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4556 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4557 _mm512_mask_compressstoreu_epi16): New.
4558 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4559 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4560 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4561 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4562 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4563 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4564 _mm256_mask_compressstoreu_epi8): New.
4565 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4566 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4567 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4568 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4569 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4570 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4571 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4572 __builtin_ia32_compresshi128_mask,
4573 __builtin_ia32_compressstoreuqi512_mask,
4574 __builtin_ia32_compressstoreuhi512_mask,
4575 __builtin_ia32_compressstoreuqi256_mask,
4576 __builtin_ia32_compressstoreuqi128_mask,
4577 __builtin_ia32_compressstoreuhi256_mask,
4578 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4579 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4581 (ix86_expand_special_args_builtin): Handle new types.
4582 (s4fma_expand): Handle new builtin array.
4583 * config/i386/immintrin.h: Include new headers.
4584 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4585 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4587 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4589 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4590 fixes. Declare temp and g variables at the top in order to avoid
4591 {} in most of the cases.
4593 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4595 * match.pd (ptr-0): New transformation.
4597 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4599 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4600 (struct addr_mode_cost_table): New.
4601 (struct tune_params): Add field addr_mode_costs.
4602 * config/arm/arm.c (generic_addr_mode_costs): New.
4603 (arm_slowmul_tune): Initialise addr_mode_costs field.
4604 (arm_fastmul_tune): Likewise.
4605 (arm_strongarm_tune): Likewise.
4606 (arm_xscale_tune): Likewise.
4607 (arm_9e_tune): Likewise.
4608 (arm_marvell_pj4_tune): Likewise.
4609 (arm_v6t2_tune): Likewise.
4610 (arm_cortex_tune): Likewise.
4611 (arm_cortex_a8_tune): Likewise.
4612 (arm_cortex_a7_tune): Likewise.
4613 (arm_cortex_a15_tune): Likewise.
4614 (arm_cortex_a35_tune): Likewise.
4615 (arm_cortex_a53_tune): Likewise.
4616 (arm_cortex_a57_tune): Likewise.
4617 (arm_exynosm1_tune): Likewise.
4618 (arm_xgene1_tune): Likewise.
4619 (arm_cortex_a5_tune): Likewise.
4620 (arm_cortex_a9_tune): Likewise.
4621 (arm_cortex_a12_tune): Likewise.
4622 (arm_cortex_a73_tune): Likewise.
4623 (arm_v7m_tune): Likewise.
4624 (arm_cortex_m7_tune): Likewise.
4625 (arm_v6m_tune): Likewise.
4626 (arm_fa726te_tune): Likewise.
4627 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4629 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4631 * config/arm/arm.c (arm_mem_costs): New function.
4632 (arm_rtx_costs_internal): Use arm_mem_costs.
4634 2017-11-23 Mark Wielaard <mark@klomp.org>
4636 * dwarf2out.c (init_sections_and_labels): Use generation to create
4637 unique ranges_section_label and ranges_base_label. Return generation.
4638 (output_rnglists): Add generation argument. Use generation to create
4639 unique ranges labels.
4640 (dwarf2out_finish): Get generation from init_sections_and_labels
4641 and pass generation to output_rnglists.
4643 2017-11-23 Mike Stump <mikestump@comcast.net>
4644 Eric Botcazou <ebotcazou@adacore.com>
4646 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4647 * cfgloop.h (struct loop): Add unroll field.
4648 * function.h (struct function): Add has_unroll bitfield.
4649 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4650 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4651 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4652 (pass_rtl_unroll_loops::gate): Likewise.
4653 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4654 for which loop->unroll==1.
4655 (decide_unroll_constant_iterations): Use note for consistency and
4656 take loop->unroll into account. Return early if loop->unroll is set.
4657 Fix thinko in existing test.
4658 (decide_unroll_runtime_iterations): Use note for consistency and
4659 take loop->unroll into account.
4660 (decide_unroll_stupid): Likewise.
4661 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4662 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4663 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4665 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4666 (print_loop): Print loop->unroll if set.
4667 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4668 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4669 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4671 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4672 loop->unroll is set and smaller than the trip count. Otherwise bypass
4673 entirely the heuristics if loop->unroll is set. Remove dead note.
4674 Fix off-by-one bug in other note.
4675 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4676 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4678 (tree_unroll_loops_completely): Make static.
4679 (pass_complete_unroll::execute): Use correct type for variable.
4680 (pass_complete_unrolli::execute): Fix formatting.
4681 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4683 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4685 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4686 enable when TARGET_PREFER_AVX128 is set.
4688 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4690 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4692 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4693 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4695 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4697 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4698 frequencies. Fix estimation of aggregate parameters.
4700 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4702 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4705 2017-11-23 Tom de Vries <tom@codesourcery.com>
4707 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4709 2017-11-23 Tom de Vries <tom@codesourcery.com>
4711 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4714 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4717 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4718 sibcall_value_pcrel_fdpic): Use local variable instead of
4720 (calli_tbr_rel): Add missing operand 2.
4721 (call_valuei_tbr_rel): Add missing operand 3.
4723 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4726 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4727 bitpos/bitsize covering the whole destination, use store_expr only if
4728 the complex mode is the same. Otherwise, use expand_normal and if
4729 it returns CONCAT, subreg each part separately instead of trying to
4730 subreg the whole result.
4732 2017-11-23 Richard Biener <rguenther@suse.de>
4734 PR tree-optimization/23094
4735 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4736 come here from walking over backedges in the first iteration.
4737 (vn_reference_lookup_3): Skip clobbers that store the same value.
4739 2017-11-23 Richard Biener <rguenther@suse.de>
4741 PR tree-optimization/81403
4742 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4743 a block we need a leader relative to.
4744 (phi_translate_1): For nary processing require a leader from
4745 get_representative_for given we run expression simplification
4746 using match-and-simplify. Remove previous fix.
4748 2017-11-22 Jeff Law <law@redhat.com>
4750 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4751 Use new method allocate_value_range rather than accessing the
4752 vrp_value_range_pool data member directly.
4753 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4754 to use extract_range_from_stmt method to avoid need for
4755 extract_range_from_assignment method.
4756 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4757 method rather than setting values_propgated data member directly.
4758 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4759 and values propagated data members and extract_range_from_assignment
4760 method. Reorder private data members to conform to standards.
4761 Add new methods set_lattice_propagation_complete and
4762 allocate_value_range.
4764 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4766 PR rtl-optimization/83030
4767 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4768 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4769 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4770 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4771 CROSSING_JUMP_P flag is already set before setting it.
4772 * cfgrtl.c (fixup_partition_crossing): Likewise.
4773 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4776 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4778 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4779 Handle the case where both arguments are using gen_const_vec_series.
4781 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4784 * pretty-print.c (pp_format): Move quoting implementation to
4785 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4786 to pass address of "quote" local.
4787 (pp_begin_quote): New function.
4788 (pp_end_quote): New function.
4789 * pretty-print.h (printer_fn): Convert penultimate param from bool
4791 (pp_begin_quote): New decl.
4792 (pp_end_quote): New decl.
4793 * tree-diagnostic.c (default_tree_printer): Convert penultimate
4794 param from bool to bool *.
4795 * tree-diagnostic.h (default_tree_printer): Likewise.
4797 2017-11-22 Jeff Law <law@redhat.com>
4799 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
4800 Initialize vr_values.
4801 (evrp_range_analyzer::try_find_new_range): Call methods attached to
4802 vr_values via vr_values class instance rather than delegators.
4803 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
4804 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4805 (evrp_range_analyzer::push_value_range): Likewise.
4806 (evrp_range_analyzer::pop_value_range): Likewise.
4807 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
4808 most delegators. Those remaining are exposed as public interfaces.
4809 Make vr_values a pointer and private.
4810 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
4812 (evrp_range_analyzer::get_vr_value): New method.
4813 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
4814 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
4815 (class evrp_dom_walker): Attach evrp_folder class, initialize
4816 it in the ctor. Remove temporary delegators.
4817 (evrp_dom_walker::before_dom_children): Call methods in attached
4818 evrp_range_analyzer class via class instance pointer. Use
4819 free value_range_constant_singleton to remove need for
4820 op_with_constant_singleton_value delegator method. Do not
4821 create a vrp_prop class instance for every call! Narrow
4822 scope of a couple variables.
4823 (evrp_dom_walker::cleanup): Call methods in attached
4824 evrp_range_analyzer class via class instance pointer.
4825 * vr-values.h (class vr_values): Privatize many methods and
4828 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4830 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
4831 * cfgexpand.c (expand_debug_expr): Remove handling for them.
4832 * expr.c (expand_expr_real_2): Likewise.
4833 * fold-const.c (const_unop): Likewise.
4834 * optabs-tree.c (optab_for_tree_code): Likewise.
4835 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
4836 * tree-inline.c (estimate_operator_cost): Likewise.
4837 * tree-pretty-print.c (dump_generic_node): Likewise.
4838 (op_code_prio): Likewise.
4839 (op_symbol_code): Likewise.
4840 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
4841 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
4842 * internal-fn.c (direct_internal_fn_optab): New function.
4843 (direct_internal_fn_array, direct_internal_fn_supported_p
4844 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
4845 * fold-const-call.c (fold_const_reduction): New function.
4846 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
4848 * tree-vect-loop.c: Include internal-fn.h.
4849 (reduction_code_for_scalar_code): Rename to...
4850 (reduction_fn_for_scalar_code): ...this and return an internal
4852 (vect_model_reduction_cost): Take an internal_fn rather than
4854 (vect_create_epilog_for_reduction): Likewise. Build calls rather
4856 (vectorizable_reduction): Use internal functions rather than tree
4857 codes for the reduction operation. Update calls to the functions
4859 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
4860 Use calls to internal functions rather than REDUC tree codes.
4861 * config/aarch64/aarch64-simd.md: Update comment accordingly.
4863 2017-11-22 Olivier Hainque <hainque@adacore.com>
4865 * config/vxworks.c (vxworks_override_options): Pick default
4866 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
4867 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
4868 DWARF_GNAT_ENCODINGS_DEFAULT.
4869 * config/vxworksae.h: Likewise.
4871 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
4873 PR tree-optimization/83104
4874 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
4877 2017-11-22 Marek Polacek <polacek@redhat.com>
4878 H.J. Lu <hongjiu.lu@intel.com>
4879 Jason Merrill <jason@redhat.com>
4884 * calls.c (initialize_argument_information): Call
4885 warn_parameter_passing_abi target hook.
4886 (store_one_arg): Use 0 for empty record size. Don't push 0 size
4887 argument onto stack.
4888 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
4889 * common.opt: Update -fabi-version description.
4890 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
4891 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
4893 (ix86_is_empty_record): New function.
4894 (ix86_warn_parameter_passing_abi): New function.
4895 (TARGET_EMPTY_RECORD_P): Redefine.
4896 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
4897 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
4898 * doc/tm.texi: Regenerated.
4899 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
4900 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
4901 * dwarf2out.c (get_ultimate_context): Move to tree.c.
4902 * explow.c (hard_function_value): Call arg_int_size_in_bytes
4903 instead of int_size_in_bytes.
4904 * expr.c (copy_blkmode_to_reg): Likewise.
4905 * function.c (aggregate_value_p): Return 0 for empty types.
4906 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
4908 (locate_and_pad_parm): Call arg size_in_bytes instead
4910 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
4911 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
4912 * target.def (empty_record_p, warn_parameter_passing_abi): New target
4914 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
4915 (std_gimplify_va_arg_expr): Skip empty records. Call
4916 arg_size_in_bytes instead size_in_bytes.
4917 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
4918 * tree-core.h (tree_type_common): Add empty_flag.
4919 (tree_decl_common): Update comments.
4920 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
4922 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4923 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
4925 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4926 * tree.c (default_is_empty_type): New function.
4927 (default_is_empty_record): New function.
4928 (arg_int_size_in_bytes): New function.
4929 (arg_size_in_bytes): New function.
4930 (get_ultimate_context): New function.
4931 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
4932 TRANSLATION_UNIT_WARN_EMPTY_P.
4933 (default_is_empty_record, arg_int_size_in_bytes,
4934 arg_size_in_bytes, get_ultimate_context): Declare.
4936 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4938 * config/arm/arm.c (cmse_clear_registers): New function.
4939 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
4940 code by call to cmse_clear_registers.
4941 (cmse_nonsecure_entry_clear_before_return): Likewise.
4943 2017-11-22 Tamar Christina <tamar.christina@arm.com>
4945 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
4946 (vdot_s32, vdotq_s32): New.
4947 (vdot_lane_u32, vdotq_lane_u32): New.
4948 (vdot_lane_s32, vdotq_lane_s32): New.
4951 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4954 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
4955 for unsigned values with fewer HWIs than the precision.
4956 (test_overflow): New function.
4957 (wide_int_cc_tests): Call it.
4959 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4961 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
4962 has the same mode class as Pmode.
4964 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4966 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
4967 padding_bits_to_clear_ptr.
4968 (cmse_nonsecure_entry_clear_before_return): Likewise.
4970 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4972 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
4973 auto_sbitap instead of integer bitfield to control register needing
4976 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4978 PR tree-optimization/83044
4979 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
4980 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
4981 that need the upper bound. Subtract offset from
4982 get_addr_base_and_unit_offset only if positive and subtract it
4983 before division by eltsize rather than after it.
4986 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
4987 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
4988 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
4992 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
4993 expand_binop with *mul_widen_optab, make sure at least one of the
4994 operands doesn't have VOIDmode.
4997 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
4999 PR rtl-optimization/82044
5000 PR tree-optimization/82042
5001 * dse.c (record_store): Check for overflow.
5002 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5003 add_wild_read instead of clear_rhs_from_active_local_stores on
5004 overflow and log it into dump_file.
5006 2017-11-22 Richard Biener <rguenther@suse.de>
5008 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5009 fake edges to exit when looking for a place to insert.
5010 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5012 (insert_into_preds_of_block): Commit edge insertion immediately,
5013 assert that doesn't require new BBs.
5014 (fini_pre): Release expressions.
5015 (pass_pre::execute): Shuffle things around a bit, if the fn
5016 is too large do not compute AVAIL either as this is really the
5019 2017-11-22 Richard Biener <rguenther@suse.de>
5021 PR tree-optimization/83089
5022 * tree-if-conv.c (pass_if_conversion::execute): If anything
5023 changed reset SCEV and free the number of iteration estimates.
5025 2017-11-21 Martin Sebor <msebor@redhat.com>
5027 PR tree-optimization/82945
5028 * calls.h (warn_nonstring_bound): Remove unused function.
5030 2017-11-21 Martin Sebor <msebor@redhat.com>
5032 PR tree-optimization/82945
5033 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5034 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5035 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5037 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5038 * calls.h (get_attr_nonstring_decl): Declare new function.
5039 * doc/extend.texi (attribute nonstring): Update.
5040 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5041 get_attr_nonstring_decl and handle it.
5042 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5043 detection of nul-termination.
5044 (strlen_to_stridx): Change to a pointer.
5045 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5046 (pass_strlen::execute): Same.
5048 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5050 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5051 for the new option -mprefer-vector-width=[none|128|256|512].
5052 * config/i386/i386.c (ix86_target_string): remove old style options
5053 -mprefer-avx256 and make -mprefer-avx128 as alias.
5054 (ix86_option_override_internal): Apply defaults for the
5055 -mprefer-vector-width=[128|256] option.
5056 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5057 Implement macros to work with -mprefer-vector-width=.
5058 * config/i386/i386.opt: Implemented option
5059 -mprefer-vector-width=[none|128|256|512].
5060 * doc/invoke.texi: Documentation for
5061 -mprefer-vector-width=[none|128|256|512].
5063 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5065 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5067 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5069 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5071 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5073 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5075 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5076 * expr.c (expand_expr_real_2): Likewise.
5077 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5078 fold_binary_loc): Likewise.
5079 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5080 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5081 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5082 MINUS_EXPR transformations.
5083 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5084 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5085 * tree-inline.c (estimate_operator_cost): Likewise.
5086 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5087 op_symbol_code): Likewise.
5088 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5089 * vr-values.c (extract_range_from_binary_expr): Likewise.
5090 * varasm.c (initializer_constant_valid_p_1): Likewise.
5091 * tree.def: New tree code POINTER_DIFF_EXPR.
5093 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5095 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5096 integer suffix to movbe mnemonic.
5097 (*bswaphi2_movbe): Ditto.
5098 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5100 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5103 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5104 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5105 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5107 2017-11-21 Martin Liska <mliska@suse.cz>
5109 * tree-inline.c (expand_call_inline): Remove not needed
5112 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5113 Jakub Jelinek <jakub@redhat.com>
5116 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5117 Remove static keyword from f variable.
5119 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5121 PR tree-optimization/83086
5122 * gimple-ssa-store-merging.c
5123 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5124 rather than n.base_addr.
5126 2017-11-21 Martin Liska <mliska@suse.cz>
5128 PR rtl-optimization/82044
5129 PR tree-optimization/82042
5130 * dse.c (check_mem_read_rtx): Check for overflow.
5132 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5134 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5137 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5139 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5140 and its corresponding call arg location note.
5142 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5143 Andrew Burgess <andrew.burgess@embecosm.com>
5145 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5147 (arc_return_slot_offset): Likewise.
5148 (arc_eh_return_address_location): New declaration.
5149 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5150 (MUST_SAVE_REGISTER): Add exception handler case.
5151 (MUST_SAVE_RETURN_ADDR): Likewise.
5152 (arc_frame_pointer_required): Likewise.
5153 (arc_frame_pointer_needed): New function.
5154 (arc_compute_frame_size): Changed.
5155 (arc_expand_prologue): Likewise.
5156 (arc_expand_epilogue): Likewise.
5157 (arc_initial_elimination_offset): Likewise.
5158 (arc_return_slot_offset): Delete.
5159 (arc_eh_return_address_location): New function.
5160 (arc_builtin_setjmp_frame_value): Likewise.
5161 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5162 (EH_RETURN_STACKADJ_RTX): Define.
5163 (EH_RETURN_HANDLER_RTX): Likewise.
5164 * config/arc/arc.md (eh_return): Delete.
5166 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5168 * print-tree.h (debug_vec_tree): Remove prototype.
5169 * gdbinit.in (pvt): Remove macro.
5171 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5173 PR tree-optimization/83047
5174 * gimple-ssa-store-merging.c
5175 (imm_store_chain_info::output_merged_store): If the loads with the
5176 same vuse are in different basic blocks, for load_gsi pick a load
5177 location that is dominated by the other loads.
5180 * config/i386/i386.c (ix86_memmodel_check): Start
5181 -Winvalid-memory-model diagnostics with lowercase letter.
5184 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5185 set in_cold_section_p to the partition containing loc_list->first.
5186 When seeing loc_list->last_before_switch node, update secname and
5187 perform range_across_switch second partition handling only after that.
5190 * run-rtl-passes.c: Include debug.h.
5191 (run_rtl_passes): Call debug_hooks->assembly_start.
5192 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5196 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5197 OPTAB_DIRECT in calls to expand_simple_binop.
5199 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5202 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5204 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5207 * spellcheck.h (best_match::blithely_get_best_candidate): New
5210 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5212 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5213 with lower case letter.
5215 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5217 * config/i386/i386.md (bswaphi2): New expander.
5218 (*bswaphi2_movbe): New insn pattern.
5219 (bswaphi -> rorhi pepehole2): New peephole pattern.
5221 2017-11-20 Jeff Law <law@redhat.com>
5223 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5224 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5225 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5226 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5227 gimple-ssa-evrp-analyze.h.
5229 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5230 set BB_VISITED here.
5231 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5233 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5235 * tree-predcom.c: Add general comment on Store-Store chains.
5236 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5238 (get_chain_last_ref_at): Rename into...
5239 (get_chain_last_write_at): ...this.
5240 (get_chain_last_write_before_load): New function.
5241 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5242 CT_STORE_STORE when write reference is added.
5243 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5244 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5245 (initialize_root_vars_store_elim_1): Ditto.
5246 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5247 definition is created.
5248 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5249 chain by replacing it with dominant stored value.
5251 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5253 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5255 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5257 * vr-values.c (extract_range_from_binary_expr): Use a full range
5260 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5262 * config/arm/arm.md (R4_REGNUM): Define constant.
5263 (nonsecure_call_internal): Remove r4 clobber.
5264 (nonsecure_call_value_internal): Likewise.
5265 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5266 clobber and resequence match_operands.
5267 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5268 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5269 (nonsecure_call_value_reg_thumb2): Likewise.
5271 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5273 PR tree-optimization/78821
5274 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5275 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5276 to the address of the base rather than the base itself.
5277 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5278 (find_bswap_or_nop_finalize): New function.
5279 (find_bswap_or_nop): Use it.
5280 (bswap_replace): Return a tree rather than bool, change first
5281 argument from gimple * to gimple_stmt_iterator, allow inserting
5282 into an empty sequence, allow ins_stmt to be NULL - then emit
5283 all stmts into gsi. Fix up MEM_REF address gimplification.
5284 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5285 (struct store_immediate_info): Add N and INS_STMT non-static
5287 (store_immediate_info::store_immediate_info): Initialize them
5288 from newly added ctor args.
5289 (merged_store_group::apply_stores): Formatting fixes. Sort by
5291 (stmts_may_clobber_ref_p): For stores call also
5292 refs_anti_dependent_p.
5293 (gather_bswap_load_refs): New function.
5294 (imm_store_chain_info::try_coalesce_bswap): New method.
5295 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5296 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5297 (imm_store_chain_info::output_merged_store): Fail if number of
5298 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5299 and NOP_EXPR rhs_code.
5300 (pass_store_merging::process_store): Compute n and ins_stmt, if
5301 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5302 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5304 (pass_store_merging::execute): Calculate dominators.
5306 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5307 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5308 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5309 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5310 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5311 class pass_optimize_bswap, bswap_replace,
5312 pass_optimize_bswap::execute): Moved to ...
5313 * gimple-ssa-store-merging.c: ... this file.
5314 Include optabs-tree.h.
5315 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5316 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5317 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5318 anonymous namespace, remove static keywords.
5319 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5320 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5322 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5325 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5327 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5329 * vec.h (debug_helper): New function.
5330 (DEFINE_DEBUG_VEC): New macro.
5331 * hash-set.h (debug_helper): New function.
5332 (DEFINE_DEBUG_HASH_SET): New macro.
5333 * cfg.c (debug_slim (edge)): New function.
5334 Call DEFINE_DEBUG_VEC for edges.
5335 Call DEFINE_DEBUG_HASH_SET for edges.
5336 * cfghooks.c (debug_slim (basic_block)): New function.
5337 Call DEFINE_DEBUG_VEC for basic blocks.
5338 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5339 * print-tree.c (debug_slim): New function to handle trees.
5340 Call DEFINE_DEBUG_VEC for trees.
5341 Call DEFINE_DEBUG_HASH_SET for trees.
5342 (debug (vec<tree, va_gc>) &): Remove.
5343 (debug (<vec<tree, va_gc>) *): Remove.
5344 * print-rtl.c (debug_slim): New function to handle const_rtx.
5345 Call DEFINE_DEBUG_VEC for rtx_def.
5346 Call DEFINE_DEBUG_VEC for rtx_insn.
5347 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5348 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5349 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5350 (debug (vec<rtx_insn *> *ptr): Remove.
5351 (debug_insn_vector): Remove.
5352 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5354 2017-11-20 Tom de Vries <tom@codesourcery.com>
5356 PR rtl-optimization/82020
5357 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5358 IF_THEN_ELSE condition.
5360 2017-11-19 Jeff Law <law@redhat.com>
5362 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5363 of degenerates resulting from ignoring an edge.
5365 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5368 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5370 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5373 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5374 for uninitialied counts.
5376 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5379 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5381 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5384 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5386 2017-11-19 Tom de Vries <tom@codesourcery.com>
5388 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5391 2017-11-19 Tom de Vries <tom@codesourcery.com>
5393 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5396 2017-11-19 Tom de Vries <tom@codesourcery.com>
5398 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5401 2017-11-19 Tom de Vries <tom@codesourcery.com>
5403 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5405 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5407 2017-11-19 Tom de Vries <tom@codesourcery.com>
5409 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5411 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5412 semicolon after MCORE_EXPORT_NAME call.
5414 2017-11-19 Tom de Vries <tom@codesourcery.com>
5417 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5418 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5420 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5422 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5423 (cgraph_edge::clone): Cleanup updating of profile.
5424 * ipa-cp.c (update_profiling_info): Likewise.
5425 * ipa-inline-transform.c (inline_transform): Likewise.
5426 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5427 * ipa-split.c (execute_split_functions): Do not split when function
5429 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5430 * profile-count.c (profile_count::dump): Add global0.
5431 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5433 (profile_count::to_sreal_scale): Likewise.
5434 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5435 (profile_count::combine_with_ipa_count): New function.
5436 * profile-count.h (profile_guessed_global0adjusted): New.
5437 (profile_count::adjusted_zero): New.
5438 (profile_count::global0adjusted): New.
5439 (profile_count::combine_with_ipa_count): New.
5440 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5441 correct profile of return block of split functions.
5442 (copy_cfg_body): Remove unused profile_count.
5443 (copy_body): Likewise.
5444 (expand_call_inline): Update.
5445 (tree_function_versioning): Update.
5447 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5449 * hash-set.h (hash_set::empty): New.
5450 * tree-ssa-threadbackward.h: Delete.
5451 * tree-ssa-threadbackward.c (class thread_jumps): New.
5452 Move max_threaded_paths into class.
5453 (fsm_find_thread_path): Remove arguments that are now in class.
5454 (profitable_jump_thread_path): Rename to...
5455 (thread_jumps::profitable_jump_thread_path): ...this.
5456 (convert_and_register_jump_thread_path): Rename to...
5457 (thread_jumps::convert_and_register_current_path): ...this.
5458 (check_subpath_and_update_thread_path): Rename to...
5459 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5460 (register_jump_thread_path_if_profitable): Rename to...
5461 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5462 (handle_phi): Rename to...
5463 (thread_jumps::handle_phi): ...this.
5464 (handle_assignment): Rename to...
5465 (thread_jumps::handle_assignment): ...this.
5466 (fsm_find_control_statement_thread_paths): Rename to...
5467 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5468 (find_jump_threads_backwards): Rename to...
5469 (thread_jumps::find_jump_threads_backwards): ...this.
5470 Initialize path local data.
5471 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5472 from within thread_jumps class.
5473 (pass_early_thread_jumps::execute): Same.
5475 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5477 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5479 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5481 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5482 to canonical location.
5484 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5487 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5489 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5491 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5493 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5494 rcrt1.o%s/grcrt1.o%s for -static-pie.
5496 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5498 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5499 ix86_shift_rotate_cost): Break out from ...
5500 (ix86_rtx_costs): ... here.
5501 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5504 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5506 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5507 when entry block was promoted unlikely.
5508 (estimate_bb_frequencies): Increase frequency scale.
5509 * profile-count.h (profile_count): Export precision info.
5511 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5513 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5514 disturbin profile of entry block.
5516 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5518 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5521 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5523 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5524 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5526 2017-11-17 Jeff Law <law@redhat.com>
5528 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5529 from evrp_dom_walker class. Various methods moved into new class.
5530 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5531 (evrp_range_analyzer::enter): New method.
5532 (evrp_range_analyzer::leave): New method.
5533 (evrp_dom_walker): Remove delegators no longer needed by this class.
5534 Replace vr_values data member with evrp_range_analyzer
5536 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5537 method extracted from evrp_dom_walker::before_dom_children.
5538 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5539 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5541 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5542 Add private copy constructor and move assignment operators.
5543 Privatize methods and class data where trivially possible.
5544 (evrp_dom_walker::cleanup): New function, extracted from
5545 execute_early_vrp. Simplify access to class data.
5547 * vr-values.h (get_output_for_vrp): Prototype.
5548 * vr-values.c (get_output_for_vrp): New function extracted from
5549 vrp_visit_assignment_or_call and extract_range_from_stmt.
5550 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5552 2017-11-17 Luis Machado <luis.machado@linaro.org>
5554 * config/aarch64/aarch64.c
5555 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5556 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5557 tune_params::AUTOPREFETCHER_WEAK.
5559 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5562 * config/arm/arm.c (arm_valid_target_attribute_rec):
5563 Parse "arch=" and "+<ext>".
5564 (arm_valid_target_attribute_tree): Re-init global options.
5565 (arm_option_override): Make non-static.
5566 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5567 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5568 (__ARM_FEATURE_CRC32): Support undef.
5569 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5570 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5572 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5574 * gdbinit.in (break-on-diagnostic): New command.
5576 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5578 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5579 return to indirect jump for EH return if control-flow
5580 protection is enabled. Change explicit 'false' argument in
5581 pro_epilogue_adjust_stack with a value of
5583 * config/i386/i386.md (simple_return_indirect_internal):
5584 Remove SImode restriction to support 64-bit.
5586 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5588 * combine.c (added_notes_insn): New.
5589 (try_combine): Handle added_notes_insn like added_links_insn.
5590 Rewrite return value code.
5591 (distribute_notes): Set added_notes_insn to the earliest insn we added
5594 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5596 PR rtl-optimization/82621
5597 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5598 dest of one of those SETs is unused.
5600 2017-11-17 Richard Biener <rguenther@suse.de>
5603 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5604 * tree-pretty-print.c (dump_generic_node): Handle
5605 annot_expr_parallel_kind.
5606 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5607 * gimplify.c (gimple_boolify): Likewise.
5609 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5611 * config.gcc (extra_headers): Add cet.h for x86 targets.
5612 * config/i386/cet.h: New file.
5613 * doc/install.texi: Add --enable-cet/--disable-cet.
5615 2017-11-17 Richard Biener <rguenther@suse.de>
5617 PR tree-optimization/83017
5618 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5619 (gen_parallel_loop): Properly count iterations.
5620 (parallelize_loops): Handle loop->can_be_parallel independent
5621 of flag_loop_parallelize_all. Make static profitability test match
5623 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5624 * invoke.texi (parloops-min-per-thread): Document.
5626 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5628 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5629 upstreaming review comments.
5631 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5633 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5634 with fast unaligned access.
5635 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5637 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5639 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5640 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5642 2017-11-17 Richard Biener <rguenther@suse.de>
5644 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5645 folding of references.
5647 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5650 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5651 of replacing call to strncmp with corresponding call to strcmp when
5654 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5656 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5657 option prefer-avx256 for skylake-avx512 configuration.
5658 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5659 (get_builtin_code_for_version): Ditto.
5661 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5662 Monk Chiang <sh.chiang04@gmail.com>
5664 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5665 (FIXED_REGISTERS): Reserve more register numbers.
5666 (CALL_USED_REGISTERS): Likewise.
5667 (REG_ALLOC_ORDER): Likewise.
5668 (REG_CLASS_CONTENTS): Likewise.
5669 (REGISTER_NAMES): Likewise.
5671 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5672 Kito Cheng <kito.cheng@gmail.com>
5674 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5676 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5678 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5680 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5681 parameter list for vec_splats.
5683 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5685 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5687 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5689 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5692 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5693 with the required version of automake 1.11.6.
5695 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5697 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5699 (DU_C2_3_power9): Correct reservation combinations.
5700 (FP_DIV_power9, VEC_DIV_power9): New.
5701 (power9-alu): Split out rotate/shift...
5702 (power9-rot): ...to here, correct dispatch resource.
5703 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5705 (power9-fp): Correct latency.
5706 (power9-sdiv): Add div/sqrt resource.
5707 (power9-ddiv): Correct latency, add div/sqrt resource.
5708 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5709 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5711 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5713 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5715 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5716 switch statement mapping KF built-ins to TF built-ins if we don't
5717 have the proper ISA 3.0 assembler support.
5719 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5721 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5722 (lower_emutls_function_body): Do not compute it.
5724 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5726 * ipa-split.c (split_bb_info): Turn time to sreal.
5727 (split_point): Likewise.
5728 (dump_split_point): Likewise.
5729 (fine_split_points): Likewise.
5730 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5732 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5734 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5737 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5739 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5741 (expensive_function_p): Remove useless assert.
5742 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5744 2017-11-16 Martin Sebor <msebor@redhat.com>
5746 PR tree-optimization/82588
5747 PR tree-optimization/82583
5748 * tree-vrp.c (check_array_ref): Handle flexible array members,
5749 string literals, and inner indices.
5750 (search_for_addr_array): Add detail to diagnostics.
5752 2017-11-16 Nathan Sidwell <nathan@acm.org>
5756 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5757 (SET_DECL_ASSEMBLER_NAME): Forward to
5758 overwrite_decl_assembler_name.
5759 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5760 (overwrite_decl_assembler_name): Declare.
5761 * tree.c (overwrite_decl_assembler_name): New.
5762 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5763 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5764 (LANG_HOOKS_INITIALIZER): Add it.
5765 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5766 * langhooks.c (lhd_set_decl_assembler_name): Use
5767 SET_DECL_ASSEMBLER_NAME.
5768 (lhd_overwrite_decl_assembler_name): Default implementation.
5770 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5771 Jackson Woodruff <jackson.woodruff@arm.com>
5773 PR tree-optimization/71026
5774 * match.pd: Canonicalize constant multiplies in division.
5776 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5778 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5779 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5780 Remove OPT_fomit_frame_pointer entry.
5781 * common/config/arc/arc-common.c: Likewise.
5782 * common/config/arm/arm-common.c: Likewise.
5783 * common/config/avr/avr-common.c: Likewise.
5784 * common/config/c6x/c6x-common.c: Likewise.
5785 * common/config/cr16/cr16-common.c: Likewise.
5786 * common/config/cris/cris-common.c: Likewise.
5787 * common/config/epiphany/epiphany-common.c: Likewise.
5788 * common/config/fr30/fr30-common.c: Likewise.
5789 * common/config/frv/frv-common.c: Likewise.
5790 * common/config/ia64/ia64-common.c: Likewise.
5791 * common/config/iq2000/iq2000-common.c: Likewise.
5792 * common/config/lm32/lm32-common.c: Likewise.
5793 * common/config/m32r/m32r-common.c: Likewise.
5794 * common/config/mcore/mcore-common.c: Likewise.
5795 * common/config/microblaze/microblaze-common.c: Likewise.
5796 * common/config/mips/mips-common.c: Likewise.
5797 * common/config/mmix/mmix-common.c: Likewise.
5798 * common/config/mn10300/mn10300-common.c: Likewise.
5799 * common/config/nios2/nios2-common.c: Likewise.
5800 * common/config/pa/pa-common.c: Likewise.
5801 * common/config/pdp11/pdp11-common.c: Likewise.
5802 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
5803 * common/config/riscv/riscv-common.c: Likewise.
5804 * common/config/rs6000/rs6000-common.c: Likewise.
5805 * common/config/rx/rx-common.c: Likewise.
5806 * common/config/s390/s390-common.c: Likewise.
5807 * common/config/sh/sh-common.c: Likewise.
5808 * common/config/sparc/sparc-common.c: Likewise.
5809 * common/config/tilegx/tilegx-common.c: Likewise.
5810 * common/config/tilepro/tilepro-common.c: Likewise.
5811 * common/config/v850/v850-common.c: Likewise.
5812 * common/config/visium/visium-common.c: Likewise.
5813 * common/config/xstormy16/xstormy16-common.c: Likewise.
5814 * common/config/xtensa/xtensa-common.c: Likewise.
5815 * invoke.texi (-fomit-frame-pointer): Update documentation.
5817 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5819 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
5821 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5823 * cfg.c (scale_bbs_frequencies_int,
5824 cale_bbs_frequencies_gcov_type): Remove.
5825 * cfg.h (scale_bbs_frequencies_int,
5826 cale_bbs_frequencies_gcov_type): Remove.
5828 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5830 * tree-ssa-loop-manip.c
5831 (scale_dominated_blocks_in_loop): Update to profile counts.
5832 (tree_transform_and_unroll_loop): Likewise.
5834 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5836 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
5837 scale_bbs_frequencies_int.
5839 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5841 * final.c (compute_alignments): Use counts rather than frequencies.
5843 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5845 * cfgloopanal.c: Include sreal.h
5846 (average_num_loop_insns): Use counts and sreal for accounting.
5848 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5850 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
5853 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5854 Kito Cheng <kito.cheng@gmail.com>
5856 * config/nds32/constraints.md: Provide more constraints.
5857 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
5858 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
5859 support constraints usage.
5861 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5863 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
5865 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5866 Kito Cheng <kito.cheng@gmail.com>
5868 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
5869 * config/nds32/nds32.opt: Refine the layout.
5870 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
5871 TARGET_EXT_STRING): Support new options.
5872 * config/nds32/nds32.h: Likewise.
5873 * config/nds32/nds32.md: Likewise.
5874 * config/nds32/nds32-predicates.c: Likewise.
5875 * config/nds32/constraints.md: Likewise.
5876 * common/config/nds32/nds32-common.c: Likewise.
5878 2017-11-16 Julia Koval <julia.koval@intel.com>
5881 * config/i386/gfniintrin.h: Add sse check.
5882 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
5884 2017-11-16 Julia Koval <julia.koval@intel.com>
5886 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
5887 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
5888 (ix86_handle_option): Handle -mavx512vbmi2.
5889 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
5890 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
5891 * config/i386/i386-c.c (__AVX512VBMI2__): New.
5892 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
5893 (ix86_valid_target_attribute_inner_p): Ditto.
5894 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
5895 * config/i386/i386.opt (mavx512vbmi2): New option.
5896 * doc/invoke.texi: Add new option.
5898 2017-11-16 Julia Koval <julia.koval@intel.com>
5900 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
5901 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
5902 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
5903 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
5904 _mm512_gf2p8mul_epi8): New intrinsics.
5905 * config/i386/i386-builtin-types.def
5906 (V64QI_FTYPE_V64QI_V64QI): New type.
5907 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
5908 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
5909 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
5910 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
5911 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
5912 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
5914 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
5916 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
5917 explicitly as a stream of bytes.
5919 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5921 * config/rs6000/altivec.h (vec_xst_be): New #define.
5922 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
5923 and externalize from *altivec_vperm_<mode>_internal.
5924 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
5926 (XL_BE_V8HI): Likewise.
5927 (XL_BE_V4SI): Likewise.
5928 (XL_BE_V4SI): Likewise.
5929 (XL_BE_V2DI): Likewise.
5930 (XL_BE_V4SF): Likewise.
5931 (XL_BE_V2DF): Likewise.
5932 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
5933 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
5934 all array entries with these keys: VSX_BUILTIN_VEC_XL,
5935 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
5936 VSX_BUILTIN_VEC_XST_BE.
5937 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
5938 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
5940 (altivec_init_builtins): Replace conditional calls to def_builtin
5941 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
5942 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
5943 with unconditional calls. Remove calls to def_builtin for
5944 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
5945 __builtin_vec_xst_be.
5946 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
5947 to define_expand, and add alternate RTL generation for P8.
5948 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
5949 vsx_ld_elemrev_v8hi.
5950 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
5951 add alternate RTL generation for P8.
5952 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
5953 vsx_ld_elemrev_v16qi.
5954 (vsx_st_elemrev_v8hi): Convert define_insn
5955 to define_expand, and add alternate RTL generation for P8.
5956 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
5957 vsx_st_elemrev_v8hi.
5958 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
5959 add alternate RTL generation for P8.
5960 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
5961 vsx_st_elemrev_v16qi.
5963 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
5966 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
5967 TARGET_AVX512ER check.
5968 (ix86_option_override_internal): Set MASK_VZEROUPPER if
5969 neither -mzeroupper nor -mno-zeroupper is used and
5970 TARGET_EMIT_VZEROUPPER is set.
5971 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
5972 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
5974 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
5976 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5977 folding of vector compares.
5978 (fold_build_vec_cmp): New helper function.
5979 (fold_compare_helper): New helper function.
5980 (builtin_function_type): Add compare builtins to the list of functions
5981 having unsigned arguments. Cosmetic updates to comment indentation.
5982 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
5983 the not+eq combination.
5985 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5987 PR tree-optimization/82726
5988 PR tree-optimization/70754
5989 * tree-predcom.c (order_drefs_by_pos): New function.
5990 (combine_chains): Move code setting has_max_use_after to...
5991 (try_combine_chains): ...here. New parameter. Sort combined chains
5992 according to position information.
5993 (tree_predictive_commoning_loop): Update call to above function.
5994 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
5996 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5998 PR tree-optimization/82726
6000 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6002 PR tree-optimization/70754
6003 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6004 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6005 combined stmt before it if not NULL.
6006 (combine_chains): Process refs reversely and compute dominance point
6010 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6012 PR tree-optimization/79663
6013 * tree-predcom.c (combine_chains): Process refs in reverse order
6014 only for ZERO length chains, and add explaining comment.
6016 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6018 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6019 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6020 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6022 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6024 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6026 2017-11-15 Martin Liska <mliska@suse.cz>
6028 * tree-cfg.c (pass_warn_function_return::execute):
6029 Compare warn_return_type for greater than zero.
6031 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6035 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6036 to return true on Xeon and not on Xeon Phi.
6037 (ix86_check_avx256_register): Changed to ...
6038 (ix86_check_avx_upper_register): ... this. Add extra check for
6039 VALID_AVX512F_REG_OR_XI_MODE.
6040 (ix86_avx_u128_mode_needed): Changed
6041 ix86_check_avx256_register to ix86_check_avx_upper_register.
6042 (ix86_check_avx256_stores): Changed to ...
6043 (ix86_check_avx_upper_stores): ... this. Changed
6044 ix86_check_avx256_register to ix86_check_avx_upper_register.
6045 (ix86_avx_u128_mode_after): Changed
6046 avx_reg256_found to avx_upper_reg_found. Changed
6047 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6048 (ix86_avx_u128_mode_entry): Changed
6049 ix86_check_avx256_register to ix86_check_avx_upper_register.
6050 (ix86_avx_u128_mode_exit): Ditto.
6051 * config/i386/i386.h: (host_detect_local_cpu): New define.
6053 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6055 * config/arm/xgene1.md (xgene1): Split into automatons
6056 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6057 (xgene1_f_load): Adjust reservations and/or types.
6058 (xgene1_f_store): Likewise.
6059 (xgene1_load_pair): Likewise.
6060 (xgene1_store_pair): Likewise.
6061 (xgene1_fp_load1): Likewise.
6062 (xgene1_load1): Likewise.
6063 (xgene1_store1): Likewise.
6064 (xgene1_move): Likewise.
6065 (xgene1_alu): Likewise.
6066 (xgene1_simd): Likewise.
6067 (xgene1_bfm): Likewise.
6068 (xgene1_neon_load1): Likewise.
6069 (xgene1_neon_store1): Likewise.
6070 (xgene1_neon_logic): Likewise.
6071 (xgene1_neon_st1): Likewise.
6072 (xgene1_neon_ld1r): Likewise.
6073 (xgene1_alu_cond): Added.
6074 (xgene1_shift_reg): Likwise.
6075 (xgene1_bfx): Likewise.
6076 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6078 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6081 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6083 (can_widen_mult_without_libcall): New function.
6084 (expand_mul_overflow): If only checking unsigned mul overflow,
6085 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6086 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6087 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6088 (expand_DIVMOD): Formatting fix.
6089 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6090 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6091 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6093 PR tree-optimization/82977
6094 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6095 constructed temporary to strlen_to_stridx.put.
6097 2017-11-15 Martin Liska <mliska@suse.cz>
6099 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6100 * configure: Regenerate.
6102 2017-11-15 Martin Liska <mliska@suse.cz>
6105 * config/sh/sh-mem.cc: Use proper probability for
6108 2017-11-14 Jeff Law <law@redhat.com>
6110 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6111 the red zone for stack_clash_protection_final_dynamic_probe targets
6112 when the total dynamic stack size is zero bytes.
6114 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6115 blocks is post order.
6117 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6119 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6120 * final.c (rest_of_clean_state): Set it for the
6121 -fcompare-debug dump.
6122 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6123 class when TDF_COMPARE_DEBUG is set.
6125 * dwarf2out.c (gen_producer_string): Discard
6128 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6131 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6132 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6133 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6134 (__TGMATH_REAL_2_3): Remove macros.
6135 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6136 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6138 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6140 (remquo): Define using __TGMATH_REAL_3.
6142 2017-11-14 Jeff Law <law@redhat.com>
6144 * vr-values.c: New file with contents extracted from tree-vrp.c.
6145 * Makefile.in (OBJS): Add vr-values.o
6146 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6147 (set_value_range, set_and_canonicalize_value_range): Likewise.
6148 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6149 (value_range_constant_singleton, symbolic_range_p): Likewise.
6150 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6151 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6152 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6153 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6154 (find_case_label_range, find_case_label_index): Likewise.
6155 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6156 (range_int_cst_singleton_p, value_inside_range): Likewise.
6157 (get_single_symbol): Likewise.
6158 (switch_update): Move structure definition here.
6159 (to_remove_edges, to_update_switch_stmts): Provide externs.
6160 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6161 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6162 (vrp_val_is_min, set_value_range): Likewise.
6163 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6164 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6165 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6166 (range_is_nonnull, range_int_cst_p): Likewwise.
6167 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6168 (get_single_symbol, operand_less_p): Likewise
6169 (compare_values_warnv, compare_values): Likewise.
6170 (value_inside_range, value_range_constant_singleton): Likewise.
6171 (zero_nonzero_bitgs_from_vr): Likewise.
6172 (extract_range_from_binary_expr_1): Likewise.
6173 (overflow_comparison_p): Likewise.
6174 (to_remove_edges, to_update_switch_stmts): Likewise.
6175 (find_case_label-index, find_case_label_range): Likewise.
6176 (switch_update, set_value_range_to_nonnegative): Remove.
6177 (set_value_range_to_truthvalue): Likewise.
6178 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6179 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6180 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6181 (find_case_label_ranges, test_for_singularity): Likewise.
6182 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6183 (x_vr_values): Move to its remaining use site.
6185 2017-11-10 Jeff Law <law@redhat.com>
6187 * vr-values.h (VR_INITIALIZER): Move #define here.
6188 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6189 * Makefile.in (OBJS): Add tree-evrp.o
6190 * tree-vrp.h (assert_info): Move structure definition here.
6191 (set_value_range_to_varying): Prototype.
6192 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6193 (infer_value_range, register_edge_assert_for): Likewise.
6194 (stmt_interesting_for_vrp): Likewise.
6195 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6196 (set_value_range_to_varying): No longer static.
6197 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6198 (infer_value_range, register_edge_assert_for): Likewise.
6200 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6202 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6203 generate the XXBRD instruction.
6205 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6206 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6207 as being compatible if -mabi=ieeelongdouble.
6208 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6209 to setup float128 built-ins with hardware support.
6210 (BU_FLOAT128_HW_2): Likewise.
6211 (BU_FLOAT128_HW_3): Likewise.
6212 (BU_FLOAT128_HW_VSX_1): Likewise.
6213 (BU_FLOAT128_HW_VSX_2): Likewise.
6214 (scalar_extract_expq): Change float128 built-in functions to
6215 accommodate having both KFmode and TFmode functions. Use the
6216 KFmode variant as the default.
6217 (scalar_extract_sigq): Likewise.
6218 (scalar_test_neg_qp): Likewise.
6219 (scalar_insert_exp_q): Likewise.
6220 (scalar_insert_exp_qp): Likewise.
6221 (scalar_test_data_class_qp): Likewise.
6222 (sqrtf128_round_to_odd): Delete processing the round to odd
6223 built-in functions as special built-in functions, and define them
6224 as float128 built-ins. Use the KFmode variant as the default.
6225 (truncf128_round_to_odd): Likewise.
6226 (addf128_round_to_odd): Likewise.
6227 (subf128_round_to_odd): Likewise.
6228 (mulf128_round_to_odd): Likewise.
6229 (divf128_round_to_odd): Likewise.
6230 (fmaf128_round_to_odd): Likewise.
6231 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6232 support for KFmode and TFmode xststdcqp calls.
6233 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6234 point, switch the built-in handlers for the get/set float128
6235 exponent, get float128 mantissa, float128 test built-ins, and the
6236 float128 round to odd built-in functions. Eliminate creating the
6237 float128 round to odd built-in functions as special built-ins.
6238 (rs6000_init_builtins): Eliminate special creation of the float128
6239 round to odd built-in functions.
6240 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6241 function insns to support both TFmode and KFmode variants.
6242 (xsxsigqp_<mode>): Likewise.
6243 (xsiexpqpf_<mode>): Likewise.
6244 (xsiexpqp_<mode>): Likewise.
6245 (xststdcqp_<mode>): Likewise.
6246 (xststdcnegqp_<mode>): Likewise.
6247 (xststdcqp_<mode>): Likewise.
6249 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6251 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6252 unused path_in_freq_ptr parameter.
6253 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6255 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6257 * ipa-inline.c (edge_badness): Dump sreal frequency.
6258 (compute_inlined_call_time): Match natural implementaiton ...
6259 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6260 forgotten division by CGRAPH_FREQ_BASE.
6262 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6264 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6265 Solaris 11. Update comment.
6266 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6268 * config/sol2.h (STARTFILE_SPEC): Likewise.
6269 * configure: Regenerate.
6271 2017-11-14 Carl Love <cel@us.ibm.com>
6273 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6274 le_ and be_ prefixes to swap* variables. Remove
6275 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6278 2017-11-14 Jason Merrill <jason@redhat.com>
6280 Support GTY((cache)) on hash_map.
6281 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6282 (ggc_cache_remove): Override it instead of ggc_mx.
6283 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6284 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6285 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6286 (simple_cache_map_traits): Override maybe_mx.
6287 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6288 (hash_map): Friend gt_cleare_cache.
6289 (gt_cleare_cache): New.
6290 * tree.h (tree_cache_traits): New hash_map traits class.
6291 (tree_cache_map): New typedef.
6293 2017-11-14 Richard Biener <rguenther@suse.de>
6295 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6296 paramter and handling.
6297 (cleanup_control_flow_bb): Likewise.
6298 (cleanup_control_flow_pre): New helper performing a DFS walk
6299 to call cleanup_control_flow_bb in PRE order.
6300 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6301 via cleanup_control_flow_pre.
6303 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6305 * config/aarch64/aarch64-simd.md
6306 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6307 (*aarch64_simd_bsl<mode>_alt): Likewise.
6308 (aarch64_simd_bsldi_internal): New.
6309 (aarch64_simd_bsldi_alt): Likewise.
6311 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6313 * tracer.c (better_p): Do not compare frequencies.
6314 * reg-stack.c (better_edge): Likewise.
6315 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6318 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6320 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6321 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6322 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6323 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6324 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6325 * predict.c (maybe_hot_frequency_p): Inline to ...
6326 (maybe_hot_count_p): ... here; rewrite to counts.
6327 (counts_to_freqs): Rename to ...
6328 (update_max_bb_count): ... this one.
6329 (expensive_function_p): Use counts.
6330 (estimate_bb_frequencies): Update.
6331 (rebuild_frequencies): Update.
6332 * predict.h (counts_to_freqs): Rename to ...
6333 (update_max_bb_count): ... this one.
6334 * profile.c (compute_branch_probabilities): Add debug info
6335 * tree-inline.c (expand_call_inline): Update debug info.
6336 (optimize_inline_calls): Use update_max_bb_count..
6337 (tree_function_versioning): Use update_max_bb_count..
6338 * value-prof.c (gimple_value_profile_transformations):
6339 Do not use update_max_bb_count.
6341 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6343 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6344 always use frequencies.
6346 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6348 * bb-reorder.c: Remove frequencies from comments.
6349 (better_edge_p): Use profile counts.
6350 (find_traces): Dump profile counts.
6351 (rotate_loop): Use profile counts.
6352 (find_traces_1_round): Likewise.
6353 (connect_better_edge_p): Use counts instead of probabilities for
6355 (copy_bb_p): Drop early check for non-0 frequency.
6356 (sanitize_hot_paths): Update comments.
6358 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6360 * ipa-split.c (struct split_point): Add count.
6361 (consider_split): Do not compute incoming frequency; compute incoming
6362 count and store it to split_point.
6363 (split_function): Set count of the call to split part correctly.
6365 2017-11-13 Carl Love <cel@us.ibm.com>
6367 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6369 2017-11-13 Tom Tromey <tom@tromey.com>
6371 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6373 2017-11-13 Carl Love <cel@us.ibm.com>
6375 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6376 Add support for builtins:
6377 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6378 vector {un,}signed {char,int,short},
6379 vector {un,}signed {char,int,short}) arguments.
6380 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6381 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6382 Add BU_P9V_AV_2 expansions for the builtins.
6383 * config/rs6000/altivec.h (vec_first_match_index,
6384 vec_first_mismatch_index, vec_first_match_or_eos_index,
6385 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6386 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6387 new extern declaration.
6388 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6389 * config/rs6000/vsx.md (first_match_index_<mode>,
6390 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6391 first_mismatch_or_eos_index_<mode>): Add define expand.
6392 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6393 * doc/extend.texi: Update the built-in documenation file for the new
6396 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6398 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6399 into the min/max operations for _Float<N> and _Float<N>X types.
6401 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6404 * dwarf2out.c (do_eh_frame): New static variable.
6405 (dwarf2out_begin_prologue): Set it.
6406 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6408 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6410 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6412 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6414 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6417 * debug.h (dwarf2out_do_eh_frame): Declare.
6418 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6419 (dwarf2out_do_frame): Use it.
6420 (dwarf2out_do_cfi_asm): Likewise.
6421 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6422 (dwarf2out_assembly_start): Likewise.
6423 (dwarf2out_begin_prologue): Fix comment.
6424 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6425 if the target needs either debug or unwind DWARF2 info.
6426 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6427 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6428 -ftrapping-math, -ftrapv and -fwrapv.
6430 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6432 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6433 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6434 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6435 (estimate_edge_size_and_time): Likewise.
6436 (ipa_merge_fn_summary_after_inlining): Likewise.
6437 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6438 (compute_uninlined_call_time): Use sreal_frequency.
6439 (compute_inlined_call_time): Likewise.
6440 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6441 * profile-count.c: Include sreal.h.
6442 (profile_count::to_sreal_scale): New.
6443 * profile-count.h: Forward declare sreal.
6444 (profile_count::to_sreal_scale): Declare.
6446 2017-11-13 Nathan Sidwell <nathan@acm.org>
6448 * diagnostic.c (maybe_line_and_column): New.
6449 (diagnostic_get_location_text): Use it.
6450 (diagnostic_report_current_module): Likewise.
6451 (test_diagnostic_get_location_text): Add tests.
6453 2017-11-13 Luis Machado <luis.machado@linaro.org>
6455 * doc/md.texi (Specifying processor pipeline description): Fix
6456 incorrect latency for the div instruction example.
6458 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6460 PR tree-optimization/78821
6461 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6462 that bit_not_p is the same.
6463 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6464 (split_group): Count precisely bit_not_p bits in each statement.
6465 (invert_op): New function.
6466 (imm_store_chain_info::output_merged_store): Use invert_op to
6467 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6468 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6470 2017-11-13 Martin Liska <mliska@suse.cz>
6472 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6473 (struct source_info): Likewise.
6474 (add_branch_counts): Likewise.
6475 (add_line_counts): Likewise.
6476 (function_summary): Likewise.
6477 (output_intermediate_line): Likewise.
6478 (generate_results): Likewise.
6480 2017-11-13 Martin Liska <mliska@suse.cz>
6482 * gcov.c (struct block_info): Remove typedef for block_t.
6483 (struct line_info): Likewise.
6484 (line_info::has_block): Likewise.
6485 (EXIT_BLOCK): Likewise.
6486 (unblock): Likewise.
6487 (circuit): Likewise.
6488 (get_cycles_count): Likewise.
6489 (process_file): Likewise.
6490 (read_graph_file): Likewise.
6491 (solve_flow_graph): Likewise.
6492 (find_exception_blocks): Likewise.
6493 (add_line_counts): Likewise.
6494 (accumulate_line_info): Likewise.
6495 (output_line_details): Likewise.
6497 2017-11-13 Martin Liska <mliska@suse.cz>
6499 * gcov.c (struct arc_info): Remove typedef for arc_t.
6500 (struct line_info): Likewise.
6501 (add_branch_counts): Likewise.
6502 (output_branch_count): Likewise.
6503 (function_info::~function_info): Likewise.
6504 (circuit): Likewise.
6505 (output_intermediate_line): Likewise.
6506 (read_graph_file): Likewise.
6507 (solve_flow_graph): Likewise.
6508 (find_exception_blocks): Likewise.
6509 (add_line_counts): Likewise.
6510 (accumulate_line_info): Likewise.
6511 (output_line_details): Likewise.
6512 (output_function_details): Likewise.
6514 2017-11-13 Martin Liska <mliska@suse.cz>
6516 * gcov.c (struct function_info): Remove typedef for function_t.
6517 (struct source_info): Likewise.
6518 (source_info::get_functions_at_location): Likewise.
6519 (solve_flow_graph): Likewise.
6520 (find_exception_blocks): Likewise.
6521 (add_line_counts): Likewise.
6522 (output_intermediate_file): Likewise.
6523 (process_file): Likewise.
6524 (generate_results): Likewise.
6525 (release_structures): Likewise.
6526 (read_graph_file): Likewise.
6527 (read_count_file): Likewise.
6528 (accumulate_line_counts): Likewise.
6529 (output_lines): Likewise.
6531 2017-11-13 Martin Liska <mliska@suse.cz>
6533 * gcov.c (function_info::function_info): Remove num_counts
6534 and add vector<gcov_type>.
6535 (function_info::~function_info): Use the vector.
6536 (process_file): Likewise.
6537 (read_graph_file): Likewise.
6538 (read_count_file): Likewise.
6539 (solve_flow_graph): Likewise.
6541 2017-11-13 Martin Liska <mliska@suse.cz>
6543 * gcov.c (function_info::is_artificial): New function.
6544 (process_file): Erase all artificial early.
6545 (generate_results): Skip as all artificial are already
6548 2017-11-13 Martin Liska <mliska@suse.cz>
6550 * gcov.c (read_graph_file): Store to global vector of functions.
6551 (read_count_file): Iterate the vector.
6552 (process_file): Likewise.
6553 (generate_results): Likewise.
6554 (release_structures): Likewise.
6556 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6558 PR tree-optimization/82954
6559 * gimple-ssa-store-merging.c
6560 (imm_store_chain_info::coalesce_immediate_stores): If
6561 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6563 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6565 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6566 Upddate call to ENDIAN_LANE_N.
6567 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6568 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6569 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6570 and use aarch64_endian_lane_rtx.
6571 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6573 2017-11-12 Tom de Vries <tom@codesourcery.com>
6575 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6577 2017-11-12 Tom de Vries <tom@codesourcery.com>
6579 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6581 2017-11-12 Tom de Vries <tom@codesourcery.com>
6583 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6585 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6586 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6587 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6589 2017-11-11 Martin Sebor <msebor@redhat.com>
6592 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6595 * prefic.c (translate_name): Replace strncpy with memcpy to
6596 avoid -Wstringop-truncation.
6598 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6600 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6602 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6604 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6606 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6608 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6611 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6613 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6614 enable generating XXBRH if the value is in a vector register.
6615 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6616 value is in a vector register.
6617 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6618 register to register bswap64's instead of doing the GPR sequence
6619 used on previous machines.
6620 (bswapdi2_xxbrd): New insn.
6621 (bswapdi2_reg): Disallow on ISA 3.0.
6622 (register to register bswap64 splitter): Do not split the insn on
6623 ISA 3.0 systems that use XXBRD.
6625 2017-11-10 Martin Sebor <msebor@redhat.com>
6628 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6630 (find_subframework_file): Same.
6632 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6634 * auto-profile.c (afdo_indirect_call): Drop frequency.
6635 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6636 (cgraph_node::create_edge): Drop frequency argument.
6637 (cgraph_node::create_indirect_edge): Drop frequency argument.
6638 (cgraph_edge::make_speculative): Drop frequency arguments.
6639 (cgraph_edge::resolve_speculation): Do not update frequencies
6640 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6641 (cgraph_node::dump): Check consistency in IPA mode.
6642 (cgraph_edge::maybe_hot_p): Use IPA counter.
6643 (cgraph_edge::verify_count_and_frequency): Rename to ...
6644 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6645 (cgraph_node::verify_node): Update.
6646 * cgraph.h (struct cgraph_edge): Drop frequency.
6647 (cgraph_edge::frequency): New function.
6648 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6650 (cgraph_edge::rebuild_edges): Likewise.
6651 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6652 (duplicate_thunk_for_node): Do not pass frequency.
6653 (cgraph_node::create_clone): Scale only counts.
6654 (cgraph_node::create_virtual_clone): Do not pass frequency.
6655 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6656 (cgraph_node::create_version_clone): Do not pass frequency.
6657 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6658 (cgraph_node::expand_thunk): Do not pass frequency.
6659 (cgraph_node::create_wrapper): Do not pass frequency.
6660 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6662 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6663 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6664 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6665 (gather_caller_stats): Use frequency function.
6666 (ipcp_cloning_candidate_p): Use frequency function.
6667 (ipcp_propagate_stage): Use frequency function.
6668 (get_info_about_necessary_edges): Use frequency function.
6669 (update_profiling_info): Update only IPA profile.
6670 (update_specialized_profile): Use frequency functoin.
6671 (perhaps_add_new_callers): Update only IPA profile.
6672 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6673 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6674 (dump_ipa_call_summary): Use frequency function.
6675 (estimate_edge_size_and_time): Use frequency function.
6676 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6677 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6678 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6679 (update_noncloned_counts): ... ths one; scale counts only.
6680 (clone_inlined_nodes): Do not scale frequency.
6681 (inline_call): Do not pass frequency.
6682 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6683 (compute_inlined_call_time): Use IPA profile.
6684 (want_inline_small_function_p): Use IPA profile.
6685 (want_inline_self_recursive_call_p): Use IPA profile.
6686 (edge_badness): Use IPA profile.
6687 (lookup_recursive_calls): Use IPA profile.
6688 (recursive_inlining): Do not pass frequency.
6689 (resolve_noninline_speculation): Do not update frequency.
6690 (inline_small_functions): Collect max of IPA profile.
6691 (dump_overall_stats): Dump IPA porfile.
6692 (dump_inline_stats): Dump IPA porfile.
6693 (ipa_inline): Collect IPA stats.
6694 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6695 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6696 (ipa_propagate_frequency): Use frequency function.
6697 (ipa_profile): Cleanup.
6698 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6699 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6700 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6701 (input_node): Do not stream frequency.
6702 (input_edge): Do not stream frequency.
6703 (merge_profile_summaries): Scale only IPA profiles.
6704 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6705 * predict.c (drop_profile): Do not recompute frequency.
6706 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6707 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6708 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6709 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6710 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6711 * tree-inline.c (copy_bb): Do not scale frequency.
6712 (expand_call_inline): Do not scale frequency.
6713 (tree_function_versioning): Do not scale frequency.
6714 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6716 2017-11-10 Julia Koval <julia.koval@intel.com>
6718 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6719 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6720 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6721 (_mm256_maskz_gf2p8affine_epi64_epi8)
6722 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6723 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6724 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6725 (__builtin_ia32_vgf2p8affineqb_v32qi)
6726 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6727 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6729 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6733 (arm_option_override): Refactor.
6734 (arm_option_reconfigure_globals): New.
6735 (arm_options_perform_arch_sanity_checks): New.
6736 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6738 (arm_options_perform_arch_sanity_checks): Likewise
6740 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6742 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6743 (power9-qpmul): New.
6744 * rs6000/rs6000.md ("type" attr): Add qmul.
6745 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6746 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6747 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6749 2017-11-10 Martin Sebor <msebor@redhat.com>
6752 * builtins.c (compute_objsize): Handle arrays that
6753 compute_builtin_object_size likes to fail for. Make extern.
6754 * builtins.h (compute_objsize): Declare.
6755 (check_strncpy_sizes): New function.
6756 (expand_builtin_strncpy): Call check_strncpy_sizes.
6757 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6758 -Wstringop-truncation.
6759 (gimple_fold_builtin_strncat): Same.
6760 * gimple.c (gimple_build_call_from_tree): Set call location.
6761 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6762 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6763 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6764 (handle_builtin_strlen): Use strlen_to_stridx.
6765 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6767 Use strlen_to_stridx.
6768 (pass_strlen::execute): Release strlen_to_stridx.
6769 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6770 (-Wstringop-truncation): Document new option.
6772 2017-11-10 Martin Liska <mliska@suse.cz>
6774 PR gcov-profile/82702
6775 * gcov.c (main): Handle intermediate files in a different
6777 (get_gcov_intermediate_filename): New function.
6778 (output_gcov_file): Remove support of intermediate files.
6779 (generate_results): Allocate intermediate file.
6780 (release_structures): Clean-up properly fn_end.
6781 (output_intermediate_file): Start iterating with line 1.
6783 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6785 PR tree-optimization/82929
6786 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6787 ops_swapped_p non-static data member.
6788 (store_immediate_info::store_immediate_info): Clear it.
6789 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6790 ops set ops_swapped_p.
6791 (count_multiple_uses): Handle ops_swapped_p.
6793 2017-11-10 Martin Liska <mliska@suse.cz>
6795 * coverage.c (coverage_init): Stream information about
6796 support of has_unexecuted_blocks.
6797 * doc/gcov.texi: Document that.
6798 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
6799 * gcov.c (read_graph_file): Likewise.
6800 (output_line_beginning): Fix a small issue with
6803 2017-11-10 Bin Cheng <bin.cheng@arm.com>
6805 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
6806 reference of trivial component.
6808 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6811 * gimple-ssa-store-merging.c
6812 (pass_store_merging::terminate_all_aliasing_chains): For
6813 gimple_store_p stmts also call refs_output_dependent_p.
6815 PR rtl-optimization/82913
6816 * compare-elim.c (try_merge_compare): Punt if def_insn is not
6819 2017-11-09 Jeff Law <law@redhat.com>
6821 * vr-values.h: New file with vr_values class.
6822 * tree-vrp.c: Include vr-values.h
6823 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
6824 data objects into the vr_values class.
6825 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
6826 (get_value_range): Make it a member function within vr_values class.
6827 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
6828 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
6829 (op_with_constant_singleton_value_range): Likewise.
6830 (extract_range_for_var_from_comparison_expr): Likewise.
6831 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
6832 (extract_range_from_binary_expr): Likewise.
6833 (extract_range_from_unary_expr): Likewise.
6834 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
6835 (check_for_binary_op_overflow, extract_range_basic): Likewise.
6836 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
6837 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
6838 (compare_name_with_value, compare_names): Likewise.
6839 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
6840 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
6841 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
6842 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
6843 (extract_range_from_phi_node): Likewise.
6844 (simplify_truth_ops_using_ranges): Likewise.
6845 (simplify_div_or_mod_using_ranges): Likewise.
6846 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
6847 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
6848 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
6849 (simplify_float_conversion_using_ranges): Likewise.
6850 (simplify_internal_call_using_ranges): Likewise.
6851 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
6852 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
6853 poitner via x_vr_values for calls into gimple folder.
6854 (vrp_initialize_lattice): Make this the vr_values ctor.
6855 (vrp_free_lattice): Make this the vr_values dtor.
6856 (set_vr_value): New function.
6857 (class vrp_prop): Add vr_values data member. Add various member
6858 functions as well as member functions that delegate to vr_values.
6859 (check_array_ref): Make a member function within vrp_prop class.
6860 (search_for_addr_array, vrp_initialize): Likewise.
6861 (vrp_finalize): Likewise. Revamp to avoid direct access to
6862 vr_value, values_propagated, etc.
6863 (check_array_bounds): Extract vrp_prop class instance pointer from
6864 walk info structure. Use it to call member functions.
6865 (check_all_array_refs): Make a member function within vrp_prop class.
6866 Smuggle class instance pointer via walk info structure.
6867 (x_vr_values): New local static.
6868 (vrp_valueize): Use x_vr_values to get class instance.
6869 (vr_valueize_1): Likewise.
6870 (class vrp_folder): Add vr_values data member. Add various member
6871 functions as well as member functions that delegate to vr_values.
6872 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
6873 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
6874 class instance from vr_values. Use it to call member functions.
6875 (vrp_dom_walker): Add vr_values data member.
6876 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
6877 instance via x_vr_values.
6878 (identify_jump_threads): Accept vr_values as argument. Store
6879 it into the walker structure.
6880 (evrp_dom_walker): Add vr_values class data member. Add various
6882 (evrp_dom_walker::try_find_new_range): Use vr_values data
6883 member to access the memory allocator.
6884 (evrp_dom_walker::before_dom_children): Store vr_values class
6885 instance into the vrp_folder class.
6886 (evrp_dom_walker::push_value_range): Rework to avoid direct
6887 access to num_vr_values and vr_value.
6888 (evrp_dom_walker::pop_value_range): Likewise.
6889 (execute_early_vrp): Remove call to vrp_initialize_lattice.
6890 Use vr_values to get to dump_all_value_ranges member function.
6891 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
6892 and simplify_cond_using_ranges_2 via vrp_prop class instance.
6893 Pass vr_values class instance down to identify_jump_threads.
6894 Remove call to vrp_free_lattice.
6895 (debug_all_value_ranges): Remove.
6897 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
6898 (vrp_folder): Likewise.
6900 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
6901 Get it from the existing bitmap instead.
6902 (vrp_intersect_ranges_1): Likewise.
6904 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6906 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6908 (store_immediate_info::store_immediate_info): Add bitnotp argument,
6909 set bit_not_p to it.
6910 (imm_store_chain_info::coalesce_immediate_stores): Break group
6911 if bit_not_p is different.
6912 (count_multiple_uses, split_group,
6913 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
6914 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
6915 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
6916 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
6917 to store_immediate_info ctor.
6919 2017-11-09 Jim Wilson <jimw@sifive.com>
6921 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
6922 (scan_prog_file): Likewise.
6924 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6926 * bb-reorder.c (max_entry_frequency): Remove.
6927 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
6928 connect_traces, push_to_next_round_p): Remove prototypes.
6929 (find_traces_1_round): Use counts only.
6930 (push_to_next_round_p): Likewise.
6931 (find_traces): Likewise.
6932 (rotate_loop): Likewise.
6933 (find_traces_1_round): Likewise.
6934 (connect_traces): Likewise.
6935 (edge_order): Likewise.
6937 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
6939 * config/arm/arm.c (output_return_instruction): Add comments to
6940 indicate requirement for cmse_nonsecure_entry return to account
6941 for the size of clearing instruction output here.
6942 (thumb_exit): Likewise.
6943 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
6944 return in hardfloat mode.
6946 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
6948 * config/rs6000/rs6000.c (machine_function): Add a bool,
6949 "toc_is_wrapped_separately".
6950 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
6951 if it wasn't explicitly set or unset, we are optimizing for speed, and
6952 doing separate shrink-wrapping.
6953 (rs6000_get_separate_components): Enable the TOC component if
6954 saving the TOC register in the prologue.
6955 (rs6000_components_for_bb): Handle the TOC component.
6956 (rs6000_emit_prologue_components): Store the TOC register where needed.
6957 (rs6000_set_handled_components): Mark TOC as handled, if handled.
6958 (rs6000_emit_prologue): Don't save the TOC if that is already done.
6960 2017-11-09 Martin Jambor <mjambor@suse.cz>
6962 * ipa-param-manipulation.c: New file.
6963 * ipa-param-manipulation.h: Likewise.
6964 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
6965 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
6966 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
6967 (ipa_parm_adjustment): Likewise.
6968 (ipa_parm_adjustment_vec): Likewise.
6969 (ipa_get_vector_of_formal_parms): Moved declaration to
6970 ipa-param-manipulation.h.
6971 (ipa_get_vector_of_formal_parm_types): Likewise.
6972 (ipa_modify_formal_parameters): Likewise.
6973 (ipa_modify_call_arguments): Likewise.
6974 (ipa_combine_adjustments): Likewise.
6975 (ipa_dump_param_adjustments): Likewise.
6976 (ipa_modify_expr): Likewise.
6977 (ipa_get_adjustment_candidate): Likewise.
6978 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
6979 ipa-param-manipulation.c.
6980 (ipa_get_vector_of_formal_parm_types): Likewise.
6981 (ipa_modify_formal_parameters): Likewise.
6982 (ipa_modify_call_arguments): Likewise.
6983 (ipa_modify_expr): Likewise.
6984 (get_ssa_base_param): Likewise.
6985 (ipa_get_adjustment_candidate): Likewise.
6986 (index_in_adjustments_multiple_times_p): Likewise.
6987 (ipa_combine_adjustments): Likewise.
6988 (ipa_dump_param_adjustments): Likewise.
6989 * tree-sra.c: Also include ipa-param-manipulation.h
6990 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
6993 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6994 Alan Hayward <alan.hayward@arm.com>
6995 David Sherwood <david.sherwood@arm.com>
6997 * doc/sourcebuild.texi (vect_masked_store): Document.
6999 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7000 Alan Hayward <alan.hayward@arm.com>
7001 David Sherwood <david.sherwood@arm.com>
7003 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7005 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7006 Alan Hayward <alan.hayward@arm.com>
7007 David Sherwood <david.sherwood@arm.com>
7009 * doc/sourcebuild.texi (vect_variable_length): Document.
7011 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7012 Alan Hayward <alan.hayward@arm.com>
7013 David Sherwood <david.sherwood@arm.com>
7015 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7017 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7018 Alan Hayward <alan.hayward@arm.com>
7019 David Sherwood <david.sherwood@arm.com>
7021 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7023 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7024 Alan Hayward <alan.hayward@arm.com>
7025 David Sherwood <david.sherwood@arm.com>
7027 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7028 previously undocumented selectors.
7029 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7031 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7033 * doc/rtl.texi (const_vector): Say that elements can be
7034 const_wide_ints too.
7035 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7036 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7037 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7038 * optabs.c (expand_vector_broadcast): Likewise.
7040 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7041 Alan Hayward <alan.hayward@arm.com>
7042 David Sherwood <david.sherwood@arm.com>
7044 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7045 scaled index even if the unscaled address was invalid.
7046 Don't increase the complexity of using a scale in that case.
7048 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7049 Alan Hayward <alan.hayward@arm.com>
7050 David Sherwood <david.sherwood@arm.com>
7052 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7053 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7054 UNITS_PER_WORD bytes.
7055 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7056 based on REGMODE_NATURAL_SIZE of the inner mode.
7057 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7058 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7059 * expmed.c (lowpart_bit_field_p): Divide the value up into
7060 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7061 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7062 whether something is likely to occupy more than one register.
7064 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7067 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7068 profile_count::adjust_for_ipa_scaling.
7069 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7070 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7072 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7074 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7076 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7077 (split_group): Add total_orig and total_new arguments, estimate the
7078 number of statements related to the store group without store merging
7079 and with store merging.
7080 (imm_store_chain_info::output_merged_store): Adjust split_group
7081 callers, punt if estimated number of statements with store merging
7082 is not smaller than estimated number of statements without it.
7084 (handled_load): Remove has_single_use checks.
7085 (pass_store_merging::process_store): Likewise.
7087 2017-11-09 Richard Biener <rguenther@suse.de>
7089 PR tree-optimization/82902
7090 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7092 2017-11-09 Martin Liska <mliska@suse.cz>
7095 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7098 2017-11-09 Martin Liska <mliska@suse.cz>
7100 PR tree-optimization/82669
7101 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7103 2017-11-09 Martin Liska <mliska@suse.cz>
7105 PR gcov-profile/48463
7106 * coverage.c (coverage_begin_function): Output also end locus
7107 of a function and information whether the function is
7109 * gcov-dump.c (tag_function): Parse and print the information.
7110 * gcov.c (INCLUDE_MAP): Add include.
7111 (INCLUDE_SET): Likewise.
7112 (struct line_info): Move earlier in the source file because
7113 of vector<line_info> in function_info structure.
7114 (line_info::line_info): Likewise.
7115 (line_info::has_block): Likewise.
7116 (struct source_info): Add new member index.
7117 (source_info::get_functions_at_location): New function.
7118 (function_info::group_line_p): New function.
7119 (output_intermediate_line): New function.
7120 (output_intermediate_file): Use the mentioned function.
7121 (struct function_start): New.
7122 (struct function_start_pair_hash): Likewise.
7123 (process_file): Add code that identifies group functions.
7124 Assign lines either to global or function scope.
7125 (generate_results): Skip artificial functions.
7126 (find_source): Assign index for each source file.
7127 (read_graph_file): Read new flag artificial and end_line.
7128 (add_line_counts): Assign it either to global of function scope.
7129 (accumulate_line_counts): Isolate core of the function to
7130 accumulate_line_info and call it for both function and global
7132 (accumulate_line_info): New function.
7133 (output_line_beginning): Fix GNU coding style.
7134 (print_source_line): New function.
7135 (output_line_details): Likewise.
7136 (output_function_details): Likewise.
7137 (output_lines): Iterate both source (global) scope and function
7139 (struct function_line_start_cmp): New class.
7140 * doc/gcov.texi: Reflect changes in documentation.
7142 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7145 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7146 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7147 and similarly for not instead of neg.
7149 2017-11-08 Andi Kleen <ak@linux.intel.com>
7151 * config/i386/i386.opt: Add -mforce-indirect-call.
7152 * config/i386/predicates.md: Check for flag_force_indirect_call.
7153 * doc/invoke.texi: Document -mforce-indirect-call
7155 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7157 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7159 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7160 config/riscv/riscv.c (predict.h): New include.
7161 (riscv_slow_unaligned_access_p): No longer static.
7162 (riscv_block_move_straight): Add require.
7163 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7165 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7168 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7169 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7170 nonimmediate_operand predicate for operand 1 instead of
7173 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7175 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7177 * config/aarch64/constraints.md (Uml): New constraint.
7178 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7181 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7183 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7184 of two vec_duplicates into a vec_concat.
7186 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7188 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7189 Simplify vec_merge of vec_duplicate and vec_concat.
7190 * config/aarch64/constraints.md (Utq): New constraint.
7191 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7194 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7196 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7197 Simplify vec_merge of vec_duplicate and const_vector.
7198 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7200 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7201 mode iterator. Update predicate on operand 1 to
7202 handle non-const_vec constants. Delete constraints.
7203 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7205 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7207 PR tree-optimization/78821
7208 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7210 (store_operand_info::store_operand_info): Initialize it to false.
7211 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7212 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7213 store in the group, and if chain_info is non-NULL, to ignore altogether
7215 (compatible_load_p): Fail if bit_not_p does not match.
7216 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7217 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7218 (pass_store_merging::process_store): Adjust
7219 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7220 call terminate_all_aliasing_chains newly when adding a store into
7221 a chain with non-NULL chain_info.
7223 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7225 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7227 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7229 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7231 (aarch64_layout_frame): Initialise emit_frame_chain.
7232 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7233 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7235 2017-11-08 Martin Liska <mliska@suse.cz>
7237 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7238 of gimple_call_internal_p.
7240 2017-11-07 Tom de Vries <tom@codesourcery.com>
7242 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7244 2017-11-07 Tom de Vries <tom@codesourcery.com>
7246 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7249 2017-11-08 Martin Liska <mliska@suse.cz>
7252 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7254 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7256 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7258 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7261 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7262 (*cmp<mode>_ccz_1): New insn with $k alternative.
7265 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7266 RTX_COMM_COMPARE as commutative as well.
7267 (ix86_binary_operator_ok): Formatting fix.
7268 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7269 *<code><mode>3<mask_name><round_saeonly_name>,
7270 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7271 *<s>mul<mode>3_highpart<mask_name>,
7272 *vec_widen_umult_even_v16si<mask_name>,
7273 *vec_widen_umult_even_v8si<mask_name>,
7274 *vec_widen_umult_even_v4si<mask_name>,
7275 *vec_widen_smult_even_v16si<mask_name>,
7276 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7277 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7278 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7279 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7280 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7281 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7282 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7283 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7284 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7285 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7286 ix86_binary_operator_ok. Formatting fixes.
7287 (*<plusminus_insn><mode>3<mask_name><round_name>,
7288 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7291 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7293 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7294 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7295 eq and ne if TARGET_ISEL.
7296 (cmp): New code_iterator.
7297 (UNS, UNSU_, UNSIK): New code_attrs.
7298 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7299 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7301 ("eq<mode>3"): ... this.
7302 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7304 ("ne<mode>3"): ... this.
7306 2017-11-07 Julia Koval <julia.koval@intel.com>
7309 * common/config/i386/i386-common.c
7310 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7311 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7312 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7313 * config/i386/i386.opt: Ditto.
7314 * config/i386/i386.c (ix86_target_string): Ditto.
7315 (ix86_option_override_internal): Ditto.
7316 (ix86_init_mpx_builtins): Move MPX to args2.
7317 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7318 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7319 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7320 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7321 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7322 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7323 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7325 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7328 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7329 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7330 (zero-extendsidi peephole2): Remove peephole.
7332 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7335 * stor-layout.c: Include attribs.h.
7336 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7337 explicit lookup of "aligned" attribute.
7339 2017-11-07 Andrew Waterman <andrew@sifive.com>
7341 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7342 (riscv_expand_block_move): Likewise.
7343 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7345 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7346 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7347 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7348 (riscv_adjust_block_mem): Likewise.
7349 (riscv_block_move_loop): Likewise.
7350 (riscv_expand_block_move): Likewise.
7351 * config/riscv/riscv.md (movmemsi): New pattern.
7353 2017-11-07 Michael Clark <michaeljclark@mac.com>
7355 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7356 (MUSL_DYNAMIC_LINKER): Likewise.
7358 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7360 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7361 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7364 2017-11-07 Richard Biener <rguenther@suse.de>
7366 * match.pd: Fix build.
7368 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7369 Jackson Woodruff <jackson.woodruff@arm.com>
7371 PR tree-optimization/71026
7372 * match.pd: Canonicalize negate in division.
7374 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7377 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7379 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7381 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7382 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7384 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7386 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7387 non-scalar integral types.
7388 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7389 (-(A-B), -(~A)): New transformations.
7391 2017-11-07 Tom de Vries <tom@codesourcery.com>
7393 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7394 semicolon after "do {} while (0)".
7395 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7396 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7397 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7398 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7399 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7401 2017-11-07 Tom de Vries <tom@codesourcery.com>
7403 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7404 after "do {} while (0)".
7405 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7406 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7407 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7408 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7409 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7411 2017-11-07 Tom de Vries <tom@codesourcery.com>
7414 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7416 (arm_rtx_costs_internal): Add missing semicolon after
7417 HANDLE_NARROW_SHIFT_ARITH call.
7419 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7421 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7422 disable isel if it was not set explicitly.
7424 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7426 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7427 (add_type_attribute) likewise.
7429 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7431 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7432 of struct ix86_frame.
7433 (ix86_initial_elimination_offset): Likewise.
7434 (ix86_expand_split_stack_prologue): Likewise.
7436 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7438 * tree-vrp.h (enum value_range_type): Update stale comment.
7440 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7441 Alan Hayward <alan.hayward@arm.com>
7442 David Sherwood <david.sherwood@arm.com>
7444 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7445 (aarch64_expand_vec_perm_const): Take the number of units too.
7446 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7447 (aarch64_expand_vec_perm_const): Likewise.
7448 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7449 (vec_perm<mode>): Update accordingly.
7451 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7452 Alan Hayward <alan.hayward@arm.com>
7453 David Sherwood <david.sherwood@arm.com>
7455 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7456 Take the number of units too.
7457 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7458 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7459 but check for a vector mode before rather than after the call.
7460 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7461 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7462 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7463 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7464 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7465 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7466 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7467 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7468 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7469 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7470 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7471 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7472 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7473 (aarch64_sqdmull2_n<mode>): Update accordingly.
7475 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7476 Alan Hayward <alan.hayward@arm.com>
7477 David Sherwood <david.sherwood@arm.com>
7479 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7480 the number of units too.
7481 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7482 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7483 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7484 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7485 (vec_store_lanesxi<mode>): Update accordingly.
7487 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7488 Alan Hayward <alan.hayward@arm.com>
7489 David Sherwood <david.sherwood@arm.com>
7491 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7492 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7493 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7494 of units rather than the mode.
7495 * config/aarch64/iterators.md (nunits): New mode attribute.
7496 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7497 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7498 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7499 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7500 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7501 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7502 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7503 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7504 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7505 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7506 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7507 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7508 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7509 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7510 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7511 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7512 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7513 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7514 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7515 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7516 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7517 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7518 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7519 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7520 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7521 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7522 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7523 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7524 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7525 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7526 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7527 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7528 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7529 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7530 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7531 (aarch64_simd_vec_setv2di): Likewise.
7533 2017-11-06 Carl Love <cel@us.ibm.com>
7535 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7537 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7538 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7540 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7542 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7543 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7544 Add power 8 macro expansions.
7545 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7546 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7547 power 8 instructions. (VSX_XXBR): Add iterator.
7549 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7551 * config/arm/arm.md (predicable_short_it): Change default to "no",
7552 improve documentation, remove uses that are identical to the default.
7553 (enabled_for_depr_it): Rename to enabled_for_short_it.
7554 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7555 Remove default uses.
7556 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7557 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7558 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7559 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7561 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7564 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7565 float128 helper macros, which are no longer used after deleting
7566 the old 'q' built-in functions, and moving the round to odd
7567 built-in functions to being special built-in functions.
7568 (BU_FLOAT128_2): Likewise.
7569 (BU_FLOAT128_1_HW): Likewise.
7570 (BU_FLOAT128_2_HW): Likewise.
7571 (BU_FLOAT128_3_HW): Likewise.
7572 (FABSQ): Delete old 'q' built-in functions.
7573 (COPYSIGNQ): Likewise.
7574 (SQRTF128_ODD): Move round to odd built-in functions to be
7575 special built-in functions, so that we can handle
7576 -mabi=ieeelongdouble.
7577 (TRUNCF128_ODD): Likewise.
7578 (ADDF128_ODD): Likewise.
7579 (SUBF128_ODD): Likewise.
7580 (MULF128_ODD): Likewise.
7581 (DIVF128_ODD): Likewise.
7582 (FMAF128_ODD): Likewise.
7583 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7584 built-in names to 'f128'.
7585 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7586 old 'q' built-in functions, as the machine independent code for
7587 'f128' built-in functions handles this.
7588 (rs6000_expand_builtin): Add expansion for float128 round to odd
7589 functions, keying off on -mabi=ieeelongdouble of whether to use
7590 the KFmode or TFmode variant.
7591 (rs6000_init_builtins): Initialize the _Float128 round to odd
7593 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7594 _Float128 'q' built-in functions are now mapped into the new
7595 'f128' built-in functions.
7597 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7599 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7600 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7602 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7605 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7606 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7607 * toplev.c: Include "ipa-fnsummary.h".
7608 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7610 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7612 PR tree-optimization/82838
7613 * gimple-ssa-store-merging.c
7614 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7615 on a separate gimple_seq which is then appended to seq.
7617 2017-11-06 Jeff Law <law@redhat.com>
7620 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7621 (get_probe_interval): New functions.
7622 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7623 (ix86_adjust_stack_and_probe): Likewise.
7624 (output_adjust_stack_and_probe): Likewise.
7625 (ix86_emit_probe_stack_range): Likewise.
7626 (ix86_expand_prologue): Likewise.
7628 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7630 PR tree-optimization/82816
7631 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7632 if the modes of the two types are the same.
7633 (convert_plusminus_to_widen): Likewise.
7635 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7637 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7639 (usadv16qi): New define_expand.
7640 (usadv8hi): New define_expand.
7642 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7645 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7647 (inline_transform): Likewise.
7648 * predict.c (counts_to_freqs): Remove useless conditional.
7649 * profile-count.h (profile_count::apply_scale): Move sanity check.
7650 * tree-inline.c (copy_bb): Always scale.
7651 (copy_cfg_body): Likewise.
7653 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7656 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7658 (*cmp_ite0): Add enabled_for_depr_it attribute.
7659 (*cmp_ite1): Likewise.
7661 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7663 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7666 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7668 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7669 return the value by pointer.
7670 (extract_range_from_multiplicative_op_1): Update accordingly.
7671 Return as soon as an operation fails.
7673 2017-11-05 Tom de Vries <tom@codesourcery.com>
7676 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7677 (DEF_SANITIZER_BUILTIN): ... here.
7678 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7679 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7681 2017-11-05 Tom de Vries <tom@codesourcery.com>
7684 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7686 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7687 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7688 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7690 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7691 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7692 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7694 2017-11-05 Tom de Vries <tom@codesourcery.com>
7697 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7700 2017-11-04 Michael Clark <michaeljclark@mac.com>
7702 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7703 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7705 (*addsi3_extended): Likewise.
7706 (*addsi3_extended2): Likewise.
7707 (<optab>si3): Likewise.
7708 (<optab>di3): Likewise.
7709 (<optab><mode>3): Likewise.
7710 (<*optabe>si3_internal): Likewise.
7711 (zero_extendqi<SUPERQI:mode>2): Likewise.
7712 (*add<mode>hi3): Likewise.
7713 (*xor<mode>hi3): Likewise.
7714 (<optab>di3): Likewise.
7715 (*<optab>si3_extend): Likewise.
7716 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7717 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7718 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7720 2017-11-04 Andrew Waterman <andrew@sifive.com>
7722 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7723 TARGET_STRICT_ALIGN based upon -mtune argument.
7725 2017-11-04 Andrew Waterman <andrew@sifive.com>
7727 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7729 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7731 * config/i386/i386.c (choose_basereg): Use optional scratch
7732 register and add assertion.
7733 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7734 needed, and don't allocate stack.
7735 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7736 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7737 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7739 2017-11-03 Jeff Law <law@redhat.com>
7741 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7742 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7743 to probe at the start of a noreturn function.
7745 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7747 PR tree-optimization/78821
7748 * gimple-ssa-store-merging.c: Update the file comment.
7749 (MAX_STORE_ALIAS_CHECKS): Define.
7750 (struct store_operand_info): New type.
7751 (store_operand_info::store_operand_info): New constructor.
7752 (struct store_immediate_info): Add rhs_code and ops data members.
7753 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7754 and op1r arguments to the ctor, initialize corresponding data members.
7755 (struct merged_store_group): Add load_align_base and load_align
7757 (merged_store_group::merged_store_group): Initialize them.
7758 (merged_store_group::do_merge): Update them.
7759 (merged_store_group::apply_stores): Pick the constant for
7760 encode_tree_to_bitpos from one of the two operands, or skip
7761 encode_tree_to_bitpos if neither operand is a constant.
7762 (class pass_store_merging): Add process_store method decl. Remove
7763 bool argument from terminate_all_aliasing_chains method decl.
7764 (pass_store_merging::terminate_all_aliasing_chains): Remove
7765 var_offset_p argument and corresponding handling.
7766 (stmts_may_clobber_ref_p): New function.
7767 (compatible_load_p): New function.
7768 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7769 if there is overlap and rhs_code is not INTEGER_CST. For
7770 non-overlapping stores terminate group if rhs is not mergeable.
7771 (get_alias_type_for_stmts): Change first argument from
7772 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7773 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7774 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7776 (get_location_for_stmts): Change first argument from
7777 auto_vec<gimple *> & to vec<gimple *> &.
7778 (struct split_store): Remove orig_stmts data member, add orig_stores.
7779 (split_store::split_store): Create orig_stores rather than orig_stmts.
7780 (find_constituent_stmts): Renamed to ...
7781 (find_constituent_stores): ... this. Change second argument from
7782 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7783 to info structures rather than the statements.
7784 (split_group): Rename ALLOW_UNALIGNED argument to
7785 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7786 it. Adjust find_constituent_stores caller.
7787 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7788 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7789 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7790 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7791 (mem_valid_for_store_merging): New function.
7792 (handled_load): New function.
7793 (pass_store_merging::process_store): New method.
7794 (pass_store_merging::execute): Use process_store method. Adjust
7795 terminate_all_aliasing_chains caller.
7797 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7799 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
7800 Return true for more constants, symbols and label references.
7801 (aarch64_valid_floating_const): Remove unused function.
7803 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7806 * config/aarch64/aarch64.c (aarch64_layout_frame):
7807 Undo forcing of LR at bottom of frame.
7809 2017-11-03 Jeff Law <law@redhat.com>
7812 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
7813 for int_registers_saved.
7815 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
7816 extracted from tree-ssa-dom.c.
7817 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
7818 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
7819 (record_equivalences_from_incoming_edge): Add additional argument
7820 to single_pred_edge_ignoring_loop_edges call.
7821 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
7822 (uncprop_dom_walker::before_dom_children): Add additional argument
7823 to single_pred_edge_ignoring_loop_edges call.
7824 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
7825 single_pred_edge_ignoring_loop_edges rather than open coding.
7826 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
7828 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
7830 * match.pd (-(-A)): Rewrite.
7832 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7834 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
7835 (rs6000_emit_int_cmove): New declaration.
7836 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
7837 (rs6000_emit_sISEL): Delete.
7838 (rs6000_emit_int_cmove): Make non-static.
7839 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
7840 instead of rs6000_emit_sISEL.
7842 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
7844 * asan.c (create_cond_insert_point): Maintain profile.
7845 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
7847 * basic-block.h (struct basic_block_def): Remove frequency.
7848 (EDGE_FREQUENCY): Use to_frequency
7849 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
7851 (find_traces): Update to use to_frequency.
7852 (find_traces_1_round): Likewise; use only IPA counts.
7853 (bb_to_key): Likewise.
7854 (connect_traces): Use IPA counts only.
7855 (copy_bb_p): Update to use to_frequency.
7856 (fix_up_crossing_landing_pad): Likewise.
7857 (sanitize_hot_paths): Likewise.
7858 * bt-load.c (basic_block_freq): Likewise.
7859 * cfg.c (init_flow): Set count_max to uninitialized.
7860 (check_bb_profile): Remove frequencies; check counts.
7861 (dump_bb_info): Do not dump frequencies.
7862 (update_bb_profile_for_threading): Update counts only.
7863 (scale_bbs_frequencies_int): Likewise.
7864 (MAX_SAFE_MULTIPLIER): Remove.
7865 (scale_bbs_frequencies_gcov_type): Update counts only.
7866 (scale_bbs_frequencies_profile_count): Update counts only.
7867 (scale_bbs_frequencies): Update counts only.
7868 * cfg.h (struct control_flow_graph): Add count-max.
7869 (update_bb_profile_for_threading): Update prototype.
7870 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
7871 (find_many_sub_basic_blocks): Likewise.
7872 * cfgcleanup.c (try_forward_edges): Likewise.
7873 (try_crossjump_to_edge): Likewise.
7874 * cfgexpand.c (expand_gimple_cond): Likewise.
7875 (expand_gimple_tailcall): Likewise.
7876 (construct_init_block): Likewise.
7877 (construct_exit_block): Likewise.
7878 * cfghooks.c (verify_flow_info): Check consistency of counts.
7879 (dump_bb_for_graph): Do not dump frequencies.
7880 (split_block_1): Do not update frequencies.
7881 (split_edge): Do not update frequencies.
7882 (make_forwarder_block): Do not update frequencies.
7883 (duplicate_block): Do not update frequencies.
7884 (account_profile_record): Do not update frequencies.
7885 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
7886 for global heuristics.
7887 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
7888 (expected_loop_iterations_unbounded): Use counts only.
7889 * cfgloopmanip.c (scale_loop_profile): Simplify.
7890 (create_empty_loop_on_edge): Simplify
7892 (duplicate_loop_to_header_edge): Simplify
7893 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
7894 (update_br_prob_note): Take care of removing note when profile
7896 (relink_block_chain): Do not dump frequency.
7897 (rtl_account_profile_record): Use to_frequency.
7898 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
7899 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
7900 (cgraph_update_edges_for_call_stmt_node): Likewise.
7901 (cgraph_edge::verify_count_and_frequency): Update.
7902 (cgraph_node::verify_node): Temporarily disable frequency verification.
7903 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
7904 to_cgraph_frequency.
7905 (cgraph_edge::rebuild_edges): Convert to ipa counts.
7906 * cgraphunit.c (init_lowered_empty_function): Do not initialize
7908 (cgraph_node::expand_thunk): Update profile.
7909 * except.c (dw2_build_landing_pads): Do not update frequency.
7910 * final.c (compute_alignments): Use to_frequency.
7911 (dump_basic_block_info): Do not dump frequency.
7912 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
7913 (dump_gimple_bb_header): Do not dump frequency.
7914 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
7916 * gimple-streamer-in.c (input_bb): Do not stream frequency.
7917 * gimple-streamer-out.c (output_bb): Do not stream frequency.
7918 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
7919 (init_before_recovery): Do not update frequency.
7920 (sched_create_recovery_edges): Do not update frequency.
7921 * hsa-gen.c (convert_switch_statements): Do not update frequency.
7922 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
7923 (ipa_cp_c_finalize): Set max_count to uninitialized.
7924 * ipa-fnsummary.c (get_minimal_bb): Use counts.
7925 (param_change_prob): Use counts.
7926 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
7928 * ipa-split.c (consider_split): Use to_frequency.
7929 (split_function): Use to_frequency.
7930 * ira-build.c (loop_compare_func): Likewise.
7931 (mark_loops_for_removal): Likewise.
7932 (mark_all_loops_for_removal): Likewise.
7933 * loop-doloop.c (doloop_modify): Do not update frequency.
7934 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
7936 * lto-streamer-in.c (input_function): Update count_max.
7937 * omp-expand.c (expand_omp_taskreg): Update count_max.
7938 * omp-simd-clone.c (simd_clone_adjust): Update profile.
7939 * predict.c (maybe_hot_frequency_p): Use to_frequency.
7940 (maybe_hot_count_p): Use ipa counts only.
7941 (maybe_hot_bb_p): Simplify.
7942 (maybe_hot_edge_p): Simplify.
7943 (probably_never_executed): Do not take frequency argument.
7944 (probably_never_executed_bb_p): Do not pass frequency.
7945 (probably_never_executed_edge_p): Likewise.
7946 (combine_predictions_for_bb): Check that profile is nonzero.
7947 (propagate_freq): Do not set frequency.
7948 (drop_profile): Simplify.
7949 (counts_to_freqs): Simplify.
7950 (expensive_function_p): Use to_frequency.
7951 (propagate_unlikely_bbs_forward): Simplify.
7952 (determine_unlikely_bbs): Simplify.
7953 (estimate_bb_frequencies): Add hack to silence graphite issues.
7954 (compute_function_frequency): Use ipa counts.
7955 (pass_profile::execute): Update.
7956 (rebuild_frequencies): Use counts only.
7957 (force_edge_cold): Use counts only.
7958 * profile-count.c (profile_count::dump): Dump new count types.
7959 (profile_count::differs_from_p): Check compatiblity.
7960 (profile_count::to_frequency): New function.
7961 (profile_count::to_cgraph_frequency): New function.
7962 * profile-count.h (struct function): Declare.
7963 (enum profile_quality): Add profile_guessed_local and
7964 profile_guessed_global0.
7965 (class profile_proability): Decrease number of bits to 29;
7966 update from_reg_br_prob_note and to_reg_br_prob_note.
7967 (class profile_count: Update comment; decrease number of bits
7968 to 61. Check compatibility.
7969 (profile_count::compatible_p): New private member function.
7970 (profile_count::ipa_p): New member function.
7971 (profile_count::operator<): Handle global zero correctly.
7972 (profile_count::operator>): Handle global zero correctly.
7973 (profile_count::operator<=): Handle global zero correctly.
7974 (profile_count::operator>=): Handle global zero correctly.
7975 (profile_count::nonzero_p): New member function.
7976 (profile_count::force_nonzero): New member function.
7977 (profile_count::max): New member function.
7978 (profile_count::apply_scale): Handle IPA scalling.
7979 (profile_count::guessed_local): New member function.
7980 (profile_count::global0): New member function.
7981 (profile_count::ipa): New member function.
7982 (profile_count::to_frequency): Declare.
7983 (profile_count::to_cgraph_frequency): Declare.
7984 * profile.c (OVERLAP_BASE): Delete.
7985 (compute_frequency_overlap): Delete.
7986 (compute_branch_probabilities): Do not use compute_frequency_overlap.
7987 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
7988 * sched-ebb.c (rank): Use counts only.
7989 * shrink-wrap.c (handle_simple_exit): Use counts only.
7990 (try_shrink_wrapping): Use counts only.
7991 (place_prologue_for_one_component): Use counts only.
7992 * tracer.c (find_best_predecessor): Use to_frequency.
7993 (find_trace): Use to_frequency.
7994 (tail_duplicate): Use to_frequency.
7995 * trans-mem.c (expand_transaction): Do not update frequency.
7996 * tree-call-cdce.c: Do not update frequency.
7997 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
7998 (gimple_merge_blocks): Likewise.
7999 (gimple_split_edge): Likewise.
8000 (gimple_duplicate_sese_region): Likewise.
8001 (gimple_duplicate_sese_tail): Likewise.
8002 (move_sese_region_to_fn): Likewise.
8003 (gimple_account_profile_record): Likewise.
8004 (insert_cond_bb): Likewise.
8005 * tree-complex.c (expand_complex_div_wide): Likewise.
8006 * tree-eh.c (lower_resx): Update profile.
8007 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8009 (initialize_cfun): Do not initialize frequencies
8010 (freqs_to_counts): Delete.
8011 (copy_cfg_body): Ignore count parameter.
8012 (copy_body): Update.
8013 (expand_call_inline): Update count_max.
8014 (optimize_inline_calls): Update count_max.
8015 (tree_function_versioning): Update count_max.
8016 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8017 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8019 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8020 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8021 (try_peel_loop): Likewise.
8022 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8024 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8025 (tree_transform_and_unroll_loop): Do not use frequencies
8026 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8027 Use reliable prediction only.
8028 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8029 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8030 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8031 probability scaling.
8032 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8033 not update frequency
8034 (any_remaining_duplicated_blocks): Likewise.
8035 (update_profile): Likewise.
8036 (estimated_freqs_path): Delete.
8037 (freqs_to_counts_path): Delete.
8038 (clear_counts_path): Delete.
8039 (ssa_fix_duplicate_block_edges): Likewise.
8040 (duplicate_thread_path): Likewise.
8041 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8042 * tree-tailcall.c (decrease_profile): Do not update frequency.
8043 (eliminate_tail_call): Likewise.
8044 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8045 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8046 (optimize_mask_stores): Likewise.
8047 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8048 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8049 (ubsan_expand_ptr_ifn): Update profile.
8050 * value-prof.c (gimple_ic): Simplify.
8051 * value-prof.h (gimple_ic): Update prototype.
8052 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8053 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8055 (want_inline_self_recursive_call_p): Likewise.
8056 (resolve_noninline_speculation): Only cummulate defined counts.
8057 (inline_small_functions): Use nonzero_p.
8058 (ipa_inline): Do not access freed node.
8060 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8062 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8063 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8065 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8067 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8068 non-legitimate address.
8070 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8072 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8073 (*lt0_<mode>di, *lt0_<mode>si): New.
8075 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8077 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8078 TARGET_PAIRED_FLOAT.
8080 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8081 Jim Wilson <jim.wilson@linaro.org>
8083 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8084 * config/aarch64/aarch64-tune.md: Regenerated.
8085 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8086 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8088 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8090 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8091 (arc_expand_prologue): Restore blink for millicode.
8092 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8094 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8097 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8098 gen_vec_duplicate after forcing the scalar into a register.
8100 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8102 * combine (try_combine): Print the insns input to try_combine to the
8105 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8108 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8109 Remove second argument from aarch64_process_target_attr call.
8110 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8112 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8114 (aarch64_handle_attr_arch): Remove second argument.
8115 (aarch64_handle_attr_cpu): Ditto.
8116 (aarch64_handle_attr_tune): Ditto.
8117 (aarch64_handle_attr_isa_flags): Ditto.
8118 (aarch64_process_one_target_attr): Ditto.
8119 (aarch64_process_target_attr): Ditto.
8120 (aarch64_option_valid_attribute_p): Remove second argument.
8121 on aarch64_process_target_attr call.
8123 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8125 * diagnostic.c: Include "selftest-diagnostic.h".
8126 (selftest::assert_location_text): New function.
8127 (selftest::test_diagnostic_get_location_text): New function.
8128 (selftest::diagnostic_c_tests): Call it.
8130 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8132 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8133 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8134 (class selftest::test_diagnostic_context): Move to...
8135 * selftest-diagnostic.c: New file.
8136 * selftest-diagnostic.h: New file.
8138 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8140 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8141 offset range for FT32B.
8142 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8143 * config/ft32/ft32.md: Add TARGET_NOPM.
8144 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8145 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8147 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8149 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8151 2017-11-02 Jeff Law <law@redhat.com>
8153 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8154 virtual keyword on FINAL OVERRIDE members.
8156 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8157 virtuals together. Add virtual destructor.
8158 (substitute_and_fold_engine): Similarly.
8160 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8162 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8164 2017-11-02 Richard Biener <rguenther@suse.de>
8166 PR tree-optimization/82795
8167 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8169 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8171 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8173 (gcc_GAS_CHECK_FEATURE): Remove.
8174 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8175 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8176 * configure: Regenerate.
8178 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8180 * config/arc/arc.c (hwloop_optimize): Account for empty
8183 2017-11-02 Richard Biener <rguenther@suse.de>
8186 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8187 Truncate ARRAY_REF index and element size.
8189 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8191 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8193 2017-11-01 Jeff Law <law@redhat.com>
8195 * tree-ssa-ccp.c (ccp_folder): New class derived from
8196 substitute_and_fold_engine.
8197 (ccp_folder::get_value): New member function.
8198 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8199 (ccp_fold_stmt): Remove prototype.
8200 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8201 * tree-ssa-copy.c (copy_folder): New class derived from
8202 substitute_and_fold_engine.
8203 (copy_folder::get_value): Renamed from get_value.
8204 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8205 * tree-vrp.c (vrp_folder): New class derived from
8206 substitute_and_fold_engine.
8207 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8208 (vrp_folder::get_value): New member function.
8209 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8210 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8211 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8212 provide a class interface to folder/substitute routines.
8213 (ssa_prop_fold_stmt_fn): Remove typedef.
8214 (ssa_prop_get_value_fn): Likewise.
8215 (subsitute_and_fold): Remove prototype.
8216 (replace_uses_in): Likewise.
8217 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8218 Renamed from replace_uses_in. Call the virtual member function
8219 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8220 (substitute_and_fold_dom_walker): Remove initialization of
8221 data member entries for calbacks. Add substitute_and_fold_engine
8222 member and initialize it.
8223 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8224 member functions for get_value, replace_phi_args_in c
8225 replace_uses_in, and fold_stmt calls.
8226 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8227 substitute_and_fold. Remove assert. Update ctor call.
8229 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8230 (ssa_prop_visit_phi_fn): Likewise.
8231 (class ssa_propagation_engine): New class to provide an interface
8233 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8235 (ssa_prop_visit_phi): Likewise.
8236 (ssa_propagation_engine::simulate_stmt): Moved into class.
8237 Call visit_phi/visit_stmt from the class rather than via
8238 file scoped static variables.
8239 (ssa_propagation_engine::simulate_block): Moved into class.
8240 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8241 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8242 set file scoped statics for the visit_stmt/visit_phi callbacks.
8243 * tree-complex.c (complex_propagate): New class derived from
8244 ssa_propagation_engine.
8245 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8246 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8247 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8249 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8250 ssa_propagation_engine.
8251 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8252 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8253 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8254 * tree-ssa-copy.c (copy_prop): New class derived from
8255 ssa_propagation_engine.
8256 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8257 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8258 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8259 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8260 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8261 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8262 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8264 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8266 PR rtl-optimization/82778
8267 PR rtl-optimization/82597
8268 * compare-elim.c (struct comparison): Add in_a_setter field.
8269 (find_comparison_dom_walker::before_dom_children): Remove killed
8270 bitmap and df_simulate_find_defs call, instead walk the defs.
8271 Compute last_setter and initialize in_a_setter. Merge definitions
8272 with first initialization for a few variables.
8273 (try_validate_parallel): Use insn_invalid_p instead of
8274 recog_memoized. Return insn rather than just the pattern.
8275 (try_merge_compare): Fix up comment. Don't uselessly test if
8276 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8278 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8281 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8282 Alan Hayward <alan.hayward@arm.com>
8283 David Sherwood <david.sherwood@arm.com>
8285 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8286 aarch64_hard_regno_nregs to get the number of registers
8289 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8290 Alan Hayward <alan.hayward@arm.com>
8291 David Sherwood <david.sherwood@arm.com>
8293 * config/aarch64/constraints.md (Upl): Rename to...
8295 * config/aarch64/aarch64.md
8296 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8299 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8300 Alan Hayward <alan.hayward@arm.com>
8301 David Sherwood <david.sherwood@arm.com>
8303 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8304 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8307 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8308 Alan Hayward <alan.hayward@arm.com>
8309 David Sherwood <david.sherwood@arm.com>
8311 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8312 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8313 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8315 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8316 of a single element.
8317 * config/aarch64/iterators.md: Add a comment above the permute
8318 unspecs to say that they are generated directly by
8319 aarch64_expand_vec_perm_const.
8320 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8322 2017-11-01 Nathan Sidwell <nathan@acm.org>
8324 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8326 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8328 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8329 and medany code models, and describe what they do.
8331 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8333 Revert accidental duplicate:
8335 * combine.c (can_change_dest_mode): Reject changes in
8336 REGMODE_NATURAL_SIZE.
8338 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8340 PR rtl-optimization/64682
8341 PR rtl-optimization/69567
8342 PR rtl-optimization/69737
8343 PR rtl-optimization/82683
8344 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8345 register mentioned in the note, drop the note, unless it came from I3,
8346 in which case it should go to I3 again.
8348 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8350 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8351 and return false if not.
8352 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8354 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8356 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8357 range is known to be empty.
8359 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8360 Alan Hayward <alan.hayward@arm.com>
8361 David Sherwood <david.sherwood@arm.com>
8363 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8364 and CONST_VECTOR_NUNITS instead of computing the number of units from
8365 the byte sizes of the vector and element.
8366 (simplify_binary_operation_1): Likewise.
8367 (simplify_const_binary_operation): Likewise.
8368 (simplify_ternary_operation): Likewise.
8370 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8371 Alan Hayward <alan.hayward@arm.com>
8372 David Sherwood <david.sherwood@arm.com>
8374 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8375 (int_mem_offset): ...this new function.
8376 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8377 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8378 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8381 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8382 Alan Hayward <alan.hayward@arm.com>
8383 David Sherwood <david.sherwood@arm.com>
8385 * lower-subreg.c (interesting_mode_p): New function.
8386 (compute_costs, find_decomposable_subregs, decompose_register)
8387 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8388 (resolve_clobber, dump_choices): Use it.
8390 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8391 Alan Hayward <alan.hayward@arm.com>
8392 David Sherwood <david.sherwood@arm.com>
8394 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8396 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8397 Alan Hayward <alan.hayward@arm.com>
8398 David Sherwood <david.sherwood@arm.com>
8400 * alias.c (find_base_value, find_base_term): Only process integer
8401 truncations. Check the precision rather than the size.
8403 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8404 Alan Hayward <alan.hayward@arm.com>
8405 David Sherwood <david.sherwood@arm.com>
8407 * machmode.h (is_narrower_int_mode): New function
8408 * optabs.c (expand_float, expand_fix): Use it.
8409 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8411 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8412 Alan Hayward <alan.hayward@arm.com>
8413 David Sherwood <david.sherwood@arm.com>
8415 * rtl.h (narrower_subreg_mode): New function.
8416 * ira-color.c (update_costs_from_allocno): Use it.
8418 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8419 Alan Hayward <alan.hayward@arm.com>
8420 David Sherwood <david.sherwood@arm.com>
8422 * optabs-query.h (convert_optab_p): New function, split out from...
8423 (convert_optab_handler): ...here.
8424 (widening_optab_handler): Delete.
8425 (find_widening_optab_handler): Remove permit_non_widening parameter.
8426 (find_widening_optab_handler_and_mode): Likewise. Provide an
8427 override that operates on mode class wrappers.
8428 * optabs-query.c (widening_optab_handler): Delete.
8429 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8430 parameter. Assert that the two modes are the same class and that
8431 the "from" mode is narrower than the "to" mode. Use
8432 convert_optab_handler instead of widening_optab_handler.
8433 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8434 instead of widening_optab_handler.
8435 * expr.c (expand_expr_real_2): Update calls to
8436 find_widening_optab_handler.
8437 * optabs.c (expand_widen_pattern_expr): Likewise.
8438 (expand_binop_directly): Take the insn_code as a parameter.
8439 (expand_binop): Only call find_widening_optab_handler for
8440 conversion optabs; use optab_handler otherwise. Update calls
8441 to find_widening_optab_handler and expand_binop_directly.
8442 Use convert_optab_handler instead of widening_optab_handler.
8443 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8444 find_widening_optab_handler and use scalar_mode rather than
8446 (convert_plusminus_to_widen): Likewise.
8448 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8449 Alan Hayward <alan.hayward@arm.com>
8450 David Sherwood <david.sherwood@arm.com>
8452 * machmode.h (fixed_size_mode): New class.
8453 * rtl.h (get_pool_mode): Return fixed_size_mode.
8454 * gengtype.c (main): Add fixed_size_mode.
8455 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8456 (get_raw_arg_mode): Likewise.
8457 * doc/tm.texi: Regenerate.
8458 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8459 * targhooks.c (default_get_reg_raw_mode): Likewise.
8460 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8461 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8462 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8463 (msp430_get_raw_result_mode): Likewise.
8464 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8465 * dbxout.c (dbxout_parms): Require fixed-size modes.
8466 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8467 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8468 * omp-low.c (lower_oacc_reductions): Likewise.
8469 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8470 (simplify_subreg): Update accordingly.
8471 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8472 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8473 that aren't fixed-size.
8474 (get_pool_mode): Return a fixed_size_mode.
8475 (output_constant_pool_2): Take a fixed_size_mode.
8477 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8478 Alan Hayward <alan.hayward@arm.com>
8479 David Sherwood <david.sherwood@arm.com>
8481 * doc/rtl.texi (vec_series): Document.
8482 (const): Say that the operand can be a vec_series.
8483 * rtl.def (VEC_SERIES): New rtx code.
8484 * rtl.h (const_vec_series_p_1): Declare.
8485 (const_vec_series_p): New function.
8486 * emit-rtl.h (gen_const_vec_series): Declare.
8487 (gen_vec_series): Likewise.
8488 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8489 (gen_vec_series): Likewise.
8490 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8491 * simplify-rtx.c (simplify_unary_operation): Handle negations
8493 (simplify_binary_operation_series): New function.
8494 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8495 (test_vector_ops_series): New function.
8496 (test_vector_ops): Call it.
8497 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8498 gen_const_vec_series.
8499 (altivec_lvsr): Likewise.
8500 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8502 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8503 Alan Hayward <alan.hayward@arm.com>
8504 David Sherwood <david.sherwood@arm.com>
8506 * doc/rtl.texi (const): Update description of address constants.
8507 Say that vector constants are allowed too.
8508 * common.md (E, F): Use CONSTANT_P instead of checking for
8510 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8511 checking for CONST_VECTOR.
8512 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8514 * expr.c (expand_expr_real_2): Check for vector modes instead
8515 of checking for CONST_VECTOR.
8516 * rtl.h (const_vec_p): New function.
8517 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8518 (unwrap_const_vec_duplicate): Handle them here too.
8520 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8521 David Malcolm <dmalcolm@redhat.com>
8522 Alan Hayward <alan.hayward@arm.com>
8523 David Sherwood <david.sherwood@arm.com>
8525 * rtl.h (vec_duplicate_p): New function.
8526 * selftest-rtl.c (assert_rtx_eq_at): New function.
8527 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8528 (assert_rtx_eq_at): Declare.
8529 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8530 * selftest-run-tests.c (selftest::run_tests): Call it.
8531 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8532 (simplify_unary_operation_1): Recursively handle vector duplicates.
8533 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8535 (simplify_subreg): Handle subregs of vector duplicates.
8536 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8537 (selftest::simplify_rtx_c_tests): New functions.
8539 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8540 Alan Hayward <alan.hayward@arm.com>
8541 David Sherwood <david.sherwood@arm.com>
8543 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8544 (gen_vec_duplicate): Likewise.
8545 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8547 (gen_const_vector): ...here.
8548 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8549 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8550 whose elements are all equal.
8551 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8552 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8553 (simplify_relational_operation): Likewise.
8554 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8556 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8557 (aarch64_expand_vector_init): Likewise.
8558 * config/arm/arm.c (neon_vdup_constant): Likewise.
8559 (neon_expand_vector_init): Likewise.
8560 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8561 (arm_block_set_unaligned_vect): Likewise.
8562 (arm_block_set_aligned_vect): Likewise.
8563 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8564 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8565 (expand_vec_perm_even_odd_pack): Likewise.
8566 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8567 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8568 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8569 gen_const_vec_duplicate.
8570 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8571 * config/mips/mips.c (mips_gen_const_int_vector): Use
8572 gen_const_vec_duplicate.
8573 (mips_expand_vector_init): Use CONST0_RTX.
8574 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8575 (define_split): Use gen_const_vec_duplicate.
8576 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8577 (define_split): Use gen_const_vec_duplicate.
8578 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8579 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8580 * config/spu/spu.c (spu_const): Likewise.
8582 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8583 Alan Hayward <alan.hayward@arm.com>
8584 David Sherwood <david.sherwood@arm.com>
8586 * combine.c (can_change_dest_mode): Reject changes in
8587 REGMODE_NATURAL_SIZE.
8589 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8591 * configure.ac (--enable-libssp): New.
8592 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8593 trying to determine target-specific default. Adjust indentation.
8594 * configure: Regenerated.
8595 * doc/install.texi (Configuration): Expand --disable-libssp
8598 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8600 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8603 2017-10-31 Martin Jambor <mjambor@suse.cz>
8606 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8608 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8610 * auto-profile.c (autofdo_source_profile::read): Use
8611 UNKNOWN_LOCATION rather than 0.
8612 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8613 (warning_at): ...this overload.
8614 (warning_at_rich_loc_n): Rename to...
8615 (warning_n): ...this overload.
8616 (error_at_rich_loc): Rename to...
8617 (error_at): ...this overload.
8618 (pedwarn_at_rich_loc): Rename to...
8619 (pedwarn): ...this overload.
8620 (permerror_at_rich_loc): Rename to...
8621 (permerror): ...this overload.
8622 (inform_at_rich_loc): Rename to...
8623 (inform): ...this overload.
8624 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8625 (diagnostic_n_impl_richloc): Rename to...
8626 (diagnostic_n_impl): ...this rich_location *-based decl.
8627 (inform_at_rich_loc): Rename to...
8628 (inform): ...this, and add an assertion.
8629 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8630 (warning_at_rich_loc): Rename to...
8631 (warning_at): ...this, and add an assertion.
8632 (warning_at_rich_loc_n): Rename to...
8633 (warning_n): ...this, and add an assertion.
8634 (warning_n): Update location_t-based implementation for removal of
8635 location_t-based diagnostic_n_impl.
8636 (pedwarn_at_rich_loc): Rename to...
8637 (pedwarn): ...this, and add an assertion.
8638 (permerror_at_rich_loc): Rename to...
8639 (permerror): ...this, and add an assertion.
8640 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8641 (error_at_rich_loc): Rename to...
8642 (error_at): ...this, and add an assertion.
8643 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8644 (driver::do_spec_on_infiles): Likewise.
8645 * substring-locations.c (format_warning_va): Update for renaming
8646 of inform_at_rich_loc.
8648 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8650 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8651 _Float<N>X built-in functions so that the variant without the
8652 "__builtin_" prefix is only enabled for the GNU C and Objective C
8653 languages when they are in non-strict ANSI/ISO mode.
8654 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8655 * target.def (floatn_builtin_p): Add a target hook to control
8656 whether _Float<N> and _Float<N>X built-in functions without the
8657 "__builtin_" prefix are enabled, and return true for C and
8658 Objective C in the default hook. Include langhooks.h in
8660 * targhooks.h (default_floatn_builtin_p): Likewise.
8661 * targhooks.c (default_floatn_builtin_p): Likewise.
8662 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8663 floatn_builtin_p target hook.
8664 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8666 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8667 Eric Botcazou <ebotcazou@adacore.com>
8669 PR rtl-optimization/81803
8670 * lra-constraints.c (curr_insn_transform): Also reload the whole
8671 register for a strict subreg no wider than a word if this is for
8672 a WORD_REGISTER_OPERATIONS target.
8674 2017-10-31 Jason Merrill <jason@redhat.com>
8676 * gdbinit.in: Skip over inlines from timevar.h.
8678 2017-10-31 Martin Liska <mliska@suse.cz>
8680 * doc/gcov.texi: Document new option.
8681 * gcov.c (print_usage): Likewise print it.
8682 (process_args): Support the argument.
8683 (format_count): New function.
8684 (format_gcov): Use the function.
8686 2017-10-31 Martin Liska <mliska@suse.cz>
8688 * gcov.c (struct name_map): do not use typedef.
8689 Define operator== and operator<.
8690 (name_search): Remove.
8691 (name_sort): Remove.
8692 (main): Do not allocate names.
8693 (process_file): Add vertical space.
8694 (generate_results): Use std::find.
8695 (release_structures): Do not release memory.
8696 (find_source): Use std::find.
8698 2017-10-31 Martin Liska <mliska@suse.cz>
8700 * gcov.c (struct line_info): Remove it's typedef.
8701 (line_info::line_info): Add proper ctor.
8702 (line_info::has_block): Do not use a typedef.
8703 (struct source_info): Do not use typedef.
8704 (circuit): Likewise.
8705 (get_cycles_count): Likewise.
8706 (output_intermediate_file): Iterate via vector iterator.
8707 (add_line_counts): Use std::vector methods.
8708 (accumulate_line_counts): Likewise.
8709 (output_lines): Likewise.
8711 2017-10-31 Martin Liska <mliska@suse.cz>
8713 * gcov.c (struct source_info): Remove typedef.
8714 (source_info::source_info): Add proper ctor.
8715 (accumulate_line_counts): Use struct, not it's typedef.
8716 (output_gcov_file): Likewise.
8717 (output_lines): Likewise.
8718 (main): Do not allocate an array.
8719 (output_intermediate_file): Use size of vector container.
8720 (process_file): Resize the vector.
8721 (generate_results): Do not preallocate, use newly added vector
8723 (release_structures): Do not release sources.
8724 (find_source): Use vector methods.
8725 (add_line_counts): Do not use typedef.
8727 2017-10-31 Martin Liska <mliska@suse.cz>
8729 * doc/gcov.texi: Document that.
8730 * gcov.c (add_line_counts): Mark lines with a non-executed
8732 (output_line_beginning): Handle such lines.
8733 (output_lines): Pass new argument.
8734 (output_intermediate_file): Print it in intermediate format.
8736 2017-10-31 Martin Liska <mliska@suse.cz>
8738 * color-macros.h: New file.
8739 * diagnostic-color.c: Factor out color related to macros to
8741 * doc/gcov.texi: Document -k option.
8742 * gcov.c (INCLUDE_STRING): Include string.h.
8743 (print_usage): Add -k option.
8744 (process_args): Parse it.
8745 (pad_count_string): New function.
8746 (output_line_beginning): Likewise.
8747 (DEFAULT_LINE_START): New macro.
8748 (output_lines): Support color output.
8750 2017-10-31 Martin Liska <mliska@suse.cz>
8752 PR gcov-profile/82633
8753 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8754 their interaction with GCOV infrastructure.
8755 * configure.ac: Add -fkeep-{inline,static}-functions to
8757 * configure: Regenerate.
8759 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8762 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8764 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8767 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8768 into a register if it does not fit into an immediate offset field.
8770 2017-10-31 Olivier Hainque <hainque@adacore.com>
8772 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8774 2017-10-31 Julia Koval <julia.koval@intel.com>
8776 * config.gcc: Add gfniintrin.h.
8777 * config/i386/gfniintrin.h: New.
8778 * config/i386/i386-builtin-types.def
8779 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8780 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8781 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8782 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8783 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8784 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8785 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8786 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8787 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8788 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8789 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8790 * config/i386/immintrin.h: Include gfniintrin.h.
8791 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8793 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
8795 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
8797 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8799 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
8800 (arm_ashldi3_1bit): Remove pattern.
8801 (ashrdi3): Remove shift by 1 expansion.
8802 (arm_ashrdi3_1bit): Remove pattern.
8803 (lshrdi3): Remove shift by 1 expansion.
8804 (arm_lshrdi3_1bit): Remove pattern.
8805 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
8806 cost of ashldi3 by 1.
8807 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
8808 (<shift>di3_neon): Likewise.
8810 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8812 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
8813 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
8814 and (*aarch64_simd_mov<VQ:mode>).
8815 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
8816 pattern alternative.
8817 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
8818 attributes to match pattern alternative.
8820 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
8822 * config.gcc (powerpc*-*-*): Add emmintrin.h.
8823 * config/rs6000/emmintrin.h: New file.
8824 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
8826 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8828 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
8829 * (movdi_vfp_cortexa8): Remove pattern.
8831 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8833 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
8835 (Specific, alpha*-dec-osf5.1): Remove.
8836 (Specific, mips-sgi-irix5): Remove.
8837 (Specific, mips-sgi-irix6): Remove.
8839 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8842 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
8843 arguments to clear_bit_region_be.
8845 2017-10-30 Jim Wilson <wilson@tuliptree.org>
8847 * gimplify.c: Include memmodel.h.
8849 2017-10-30 Martin Jambor <mjambor@suse.cz>
8851 * omp-grid.c (grid_attempt_target_gridification): Also insert a
8852 condition whether loop should be executed at all.
8854 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
8856 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
8857 gimple folding of vec_madd() intrinsics.
8858 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
8859 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
8860 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
8862 2017-10-30 Richard Biener <rguenther@suse.de>
8864 PR tree-optimization/82762
8866 2017-10-23 Richard Biener <rguenther@suse.de>
8868 PR tree-optimization/82129
8870 2017-08-01 Richard Biener <rguenther@suse.de>
8872 PR tree-optimization/81181
8873 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8874 (compute_antic): ... end of iteration here.
8876 2017-10-30 Joseph Myers <joseph@codesourcery.com>
8878 * doc/invoke.texi (C Dialect Options): Document -std=c17,
8879 -std=iso9899:2017 and -std=gnu17.
8880 * doc/standards.texi (C Language): Document C17 support.
8881 * doc/cpp.texi (Overview): Mention -std=c17.
8882 (Standard Predefined Macros): Document C11 and C17 values of
8883 __STDC_VERSION__. Do not refer to C99 support as incomplete.
8884 * doc/extend.texi (Inline): Do not list individual options for
8885 standards newer than C99.
8886 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
8888 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
8891 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
8893 * asan.c (asan_finish_file): Align asan globals array by shadow
8896 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8899 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
8900 (struct store_immediate_info): Add bitregion_start and bitregion_end
8902 (store_immediate_info::store_immediate_info): Add brs and bre
8903 arguments and initialize bitregion_{start,end} from those.
8904 (struct merged_store_group): Add bitregion_start, bitregion_end,
8905 align_base and mask fields. Drop unnecessary struct keyword from
8906 struct store_immediate_info. Add do_merge method.
8907 (clear_bit_region_be): Use memset instead of loop storing zeros.
8908 (merged_store_group::do_merge): New method.
8909 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
8910 stores as long as the surrounding bitregions have no gaps.
8911 (merged_store_group::merge_overlapping): Use do_merge.
8912 (merged_store_group::apply_stores): Test that bitregion_{start,end}
8913 is byte aligned, rather than requiring that start and width are
8914 byte aligned. Drop unnecessary struct keyword from
8915 struct store_immediate_info. Allocate and populate also mask array.
8916 Make start of the arrays relative to bitregion_start rather than
8917 start and size them according to bitregion_{end,start} difference.
8918 (struct imm_store_chain_info): Drop unnecessary struct keyword from
8919 struct store_immediate_info.
8920 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
8921 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
8922 struct keyword from struct store_immediate_info.
8923 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
8924 between stores as long as the surrounding bitregions have no gaps.
8926 (struct split_store): Add orig non-static data member.
8927 (split_store::split_store): Initialize orig to false.
8928 (find_constituent_stmts): Return store_immediate_info *, non-NULL
8929 if there is exactly a single original stmt. Change stmts argument
8930 to pointer from reference, if NULL, don't push anything to it. Add
8931 first argument, use it to optimize skipping over orig stmts that
8932 are known to be before bitpos already. Simplify.
8933 (split_group): Return unsigned int count how many stores are or
8934 would be needed rather than a bool. Add allow_unaligned argument.
8935 Change split_stores argument from reference to pointer, if NULL,
8936 only do a dry run computing how many stores would be produced.
8937 Rewritten algorithm to use both alignment and misalign if
8938 !allow_unaligned and handle bitfield stores with gaps.
8939 (imm_store_chain_info::output_merged_store): Set start_byte_pos
8940 from bitregion_start instead of start. Compute allow_unaligned
8941 here, if true, do 2 split_group dry runs to compute which one
8942 produces fewer stores and prefer aligned if equal. Punt if
8943 new count is bigger or equal than original before emitting any
8944 statements, rather than during that. Remove no longer needed
8945 new_ssa_names tracking. Replace num_stmts with
8946 split_stores.length (). Use 32-bit stack allocated entries
8947 in split_stores auto_vec. Try to reuse original store lhs/rhs1
8948 if possible. Handle bitfields with gaps.
8949 (pass_store_merging::execute): Ignore bitsize == 0 stores.
8950 Compute bitregion_{start,end} for the stores and construct
8951 store_immediate_info with that. Formatting fixes.
8953 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
8956 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
8957 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
8959 2017-10-29 Jim Wilson <wilson@tuliptree.org>
8961 * gimplify.c: Include tm_p.h.
8963 * common.opt (gcoff): Re-add as ignored option.
8964 (gcoff1, gcoff2, gcoff3): Likewise.
8966 * Makefile.in (OBJS): Delete sdbout.o.
8967 (GTFILES): Delete $(srcdir)/sdbout.c.
8968 * debug.h: Delete sdb_debug_hooks.
8969 * final.c: Delete sdbout.h include.
8970 (final_scan_insn): Delete SDB_DEBUG check.
8971 (rest_of_clean_state): Likewise.
8972 * output.h: Delete sdb_begin_function_line.
8975 * toplev.c: Delete sdbout.h include.
8976 (process_options): Delete SDB_DEBUG check.
8977 * tree-core.h (tree_type_common): Delete pointer field of
8979 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
8980 TYPE_SYMTAB_POINTER.
8981 * tree.h (TYPE_SYMTAB_POINTER): Delete.
8982 (TYPE_SYMTAB_IS_POINTER): Delete.
8983 (TYPE_SYMTAB_IS_DIE): Renumber.
8984 * xcoffout.c: Refer to former sdbout.c file.
8985 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
8987 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
8988 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
8989 (Debugging Options): Delete -gcoff.
8990 (-gstabs): Delete SDB reference.
8992 (-gcoff@var{level}): Delete.
8993 * doc/passes.texi (Debugging information output): Delete SDB and
8994 sdbout.c references.
8995 * doc/tm.texi: Regenerate.
8996 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
8997 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
8999 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9000 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9001 (SDB_DEBUGGING_INFO): Delete.
9002 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9003 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9004 * target.def (output_source_filename): Delete COFF reference.
9006 * common.opt (gcoff): Delete.
9007 (gxcoff+): Update Negative chain.
9008 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9010 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9011 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9012 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9014 (expand_function_start): Change sdb reference to past tense.
9015 (expand_function_end): Change sdb reference to past tense.
9016 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9017 * opts.c (debug_type_names): Delete coff entry.
9018 (common_handle_option): Delete OPT_gcoff case.
9019 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9021 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9022 * config/cris/cris.h: Delete SDB reference in comment.
9023 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9024 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9025 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9026 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9028 (ix86_expand_prologue): Likewise.
9029 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9030 * config/ia64/ia64.h: Likewise.
9031 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9032 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9034 * config/mmix/mmix.h: Likewise.
9035 * config/nds32/nds32.c: Likewise.
9036 * config/stormy/storym16.h: Likewise.
9037 * config/visium/visium.h: Likewise.
9038 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9040 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9042 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9043 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9044 FRAME_POINTER_REGNUM point at high end of local var area.
9046 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9048 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9049 Move comment around. Do not reset best_edge for a copiable
9050 destination if the copy would cause a partition change.
9051 (better_edge_p): Remove redundant check.
9053 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9055 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9057 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9059 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9060 for math functions that have _Float<N> and _Float<N>X variants.
9061 (mathfn_built_in_2): Add support for math functions that have
9062 _Float<N> and _Float<N>X variants.
9063 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9064 (expand_builtin_mathfn_ternary): Add support for fma with
9065 _Float<N> and _Float<N>X variants.
9066 (expand_builtin): Likewise.
9067 (fold_builtin_3): Likewise.
9068 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9069 create math function _Float<N> and _Float<N>X variants as external
9071 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9072 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9073 the __builtin_ prefix and if not strict ansi, without the prefix.
9074 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9075 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9076 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9077 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9078 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9079 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9080 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9081 function signatures for fma _Float<N> and _Float<N>X variants.
9082 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9083 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9084 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9085 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9086 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9087 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9088 * gencfn-macros.c (print_case_cfn): Add support for math functions
9089 that have _Float<N> and _Float<N>X variants.
9090 (print_define_operator_list): Likewise.
9091 (fltfn_suffixes): Likewise.
9093 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9094 for math functions that have _Float<N> and _Float<N>X variants.
9095 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9096 and _Float<N>X variants.
9097 (COPYSIGN): Likewise.
9100 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9101 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9103 (integer_valued_read_call_p): Likewise.
9104 * fold-const-call.c (fold_const_call_ss): Likewise.
9105 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9106 _Float<N> and _Float<N>X variants.
9107 (fold_const_call_ssss): Add support for fma _Float<N> and
9108 _Float<N>X variants.
9109 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9110 support for copysign and fma _Float<N> and _Float<N>X variants.
9111 (backprop::process_builtin_call_use): Likewise.
9112 * tree-call-cdce.c (can_test_argument_range); Add support for
9113 sqrt _Float<N> and _Float<N>X variants.
9114 (edom_only_function): Likewise.
9115 (get_no_error_domain): Likewise.
9116 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9117 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9118 copysign _Float<N> and _Float<N>X variants.
9119 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9120 handled by machine independent code.
9121 (FMAF128): Likewise.
9122 * doc/cpp.texi (Common Predefined Macros): Document defining
9123 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9124 fma _Float<N> and _Float<N>X variants.
9126 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9129 * config/i386/i386-modes.def (CCFPU): Remove definition.
9130 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9131 (ix86_cc_modes_compatible): Ditto.
9132 (ix86_expand_carry_flag_compare): Ditto.
9133 (ix86_expand_int_movcc): Ditto.
9134 (ix86_expand_int_addcc): Ditto.
9135 (ix86_reverse_condition): Ditto.
9136 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9137 Return true/false for unordered/ordered fp comparisons.
9138 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9139 (ix86_prepare_fp_compare_args): Update for rename.
9140 (ix86_expand_fp_compare): Update for rename. Generate unordered
9141 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9142 (ix86_expand_sse_compare_and_jump): Ditto.
9143 * config/i386/predicates.md (fcmov_comparison_operator):
9144 Remove CCFPU mode handling.
9145 (ix86_comparison_operator): Ditto.
9146 (ix86_carry_flag_operator): Ditto.
9147 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9148 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9149 (*cmpu<mode>_cc_i387): Ditto.
9150 (FPCMP): Remove mode iterator.
9151 (unord): Remove mode attribute.
9152 (unord_subst): New define_subst transformation
9153 (unord): New define_subst attribute.
9155 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9156 (*cmpi<unord>xf_i387): Ditto.
9157 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9158 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9159 using unord_subst transformation.
9160 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9161 (round_saeonly): Also handle CCFP mode.
9162 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9163 Remove UNSPEC_SAHF unspec handling.
9165 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9167 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9169 2017-10-27 Jeff Law <law@redhat.com>
9171 * gimple-ssa-sprintf.c: Include domwalk.h.
9172 (class sprintf_dom_walker): New class, derived from dom_walker.
9173 (sprintf_dom_walker::before_dom_children): New function.
9174 (struct call_info): Moved into sprintf_dom_walker class
9175 (compute_formath_length, handle_gimple_call): Likewise.
9176 (sprintf_length::execute): Call the dominator walker rather
9177 than walking the statements.
9179 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9180 gimple statement locations.
9181 (check_array_bounds): Corresponding changes. Get the statement's
9182 location directly from wi->stmt.
9184 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9187 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9189 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9191 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9192 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9194 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9197 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9198 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9199 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9200 maybe_get_pool_constant.
9201 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9204 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9206 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9207 2.26 caveat. Update gas and gld versions.
9208 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9211 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9213 * cgraph.h (set_malloc_flag): Declare.
9214 * cgraph.c (set_malloc_flag_1): New function.
9215 (set_malloc_flag): Likewise.
9216 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9217 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9219 (read_ipa_call_summary): Add support for reading is_return_callee.
9220 (write_ipa_call_summary): Stream is_return_callee.
9221 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9222 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9223 ipa-prop.h, ipa-fnsummary.h.
9224 (pure_const_names): Change to static.
9225 (malloc_state_e): Define.
9226 (malloc_state_names): Define.
9227 (funct_state_d): Add field malloc_state.
9228 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9229 (check_retval_uses): New function.
9230 (malloc_candidate_p): Likewise.
9231 (analyze_function): Add support for malloc attribute.
9232 (pure_const_write_summary): Stream malloc_state.
9233 (pure_const_read_summary): Add support for reading malloc_state.
9234 (dump_malloc_lattice): New function.
9235 (propagate_malloc): New function.
9236 (warn_function_malloc): New function.
9237 (ipa_pure_const::execute): Call propagate_malloc and
9238 ipa_free_fn_summary.
9239 (pass_local_pure_const::execute): Add support for malloc attribute.
9240 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9241 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9243 2017-10-27 Martin Liska <mliska@suse.cz>
9245 PR gcov-profile/82457
9246 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9247 for fork-like functions to be properly instrumented.
9249 2017-10-27 Richard Biener <rguenther@suse.de>
9252 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9253 info when we redirected EH.
9255 2017-10-26 Michael Collison <michael.collison@arm.com>
9257 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9259 (<optab>_trunchf<GPI:mode>2: New pattern.
9260 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9261 * config/aarch64/iterators.md (wv): New mode attribute.
9262 (vf, VF): New mode attributes.
9263 (vgp, VGP): New mode attributes.
9264 (s): Update attribute with SImode and DImode prefixes.
9266 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9268 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9269 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9270 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9271 (nios2_option_overide): Initialize it. Don't allow R0-relative
9272 addressing with PIC.
9273 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9274 (nios2_symbolic_constant_p): Likewise.
9275 (nios2_legitimate_address_p): Likewise.
9276 (nios2_r0rel_section_name_p): New.
9277 (nios2_symbol_ref_in_r0rel_data_p): New.
9278 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9279 (r0rel_constant_p): New.
9280 (nios2_print_operand_address): Handle r0rel_constant_p.
9281 (nios2_cdx_narrow_form_p): Likewise.
9282 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9283 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9284 (Nios II Options): Document -mr0rel-sec.
9286 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9288 * config/nios2/nios2.c: Include xregex.h.
9289 (nios2_gprel_sec_regex): New.
9290 (nios2_option_overide): Initialize it. Don't allow GP-relative
9291 addressing with PIC.
9292 (nios2_small_section_name_p): Check for regex match.
9293 * config/nios2/nios2.opt (mgprel-sec=): New option.
9294 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9295 (Nios II Options): Document -mgprel-sec.
9297 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9299 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9301 2017-10-26 Tom de Vries <tom@codesourcery.com>
9303 PR tree-optimization/82707
9304 * gimple.c (gimple_copy): Fix unsharing of
9305 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9307 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9309 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9310 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9311 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9312 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9313 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9314 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9315 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9316 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9317 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9318 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9319 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9320 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9321 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9322 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9323 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9324 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9325 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9327 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9329 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9331 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9332 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9333 warning to rs6000.c. Remove the Undocumented flag, since it has
9335 (-mabi=ibmlongdouble): Likewise.
9336 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9337 already set, set the default format for long double.
9338 (rs6000_debug_reg_global): Print whether long double is IBM or
9340 (rs6000_option_override_internal): Rework setting long double
9341 format. Only warn if the user is changing the long double default
9342 and they did not use -Wno-psabi.
9343 * doc/invoke.texi (PowerPC options): Update the documentation for
9344 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9346 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9347 Alan Hayward <alan.hayward@arm.com>
9348 David Sherwood <david.sherwood@arm.com>
9350 * rtl.h (wider_subreg_mode): New function.
9351 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9352 rather than an unsigned int *.
9353 * ira-color.c (regno_max_ref_width): Replace with...
9354 (regno_max_ref_mode): ...this new variable.
9355 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9356 Use wider_subreg_mode.
9357 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9358 rather than an unsigned int *.
9359 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9360 (process_alt_operands): Likewise.
9361 (invariant_p): Likewise.
9362 * lra-spills.c (assign_mem_slot): Likewise.
9363 (add_pseudo_to_slot): Likewise.
9364 * lra.c (collect_non_operand_hard_regs): Likewise.
9365 (add_regs_to_insn_regno_info): Likewise.
9366 * reload1.c (regno_max_ref_width): Replace with...
9367 (regno_max_ref_mode): ...this new variable.
9368 (reload): Update accordingly. Update call to
9369 ira_sort_regnos_for_alter_reg.
9370 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9371 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9372 (scan_paradoxical_subregs): Likewise.
9374 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9376 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9377 (aarch64_frame): Add emit_frame_chain boolean.
9378 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9379 Move eh_return case to aarch64_layout_frame.
9380 (aarch64_layout_frame): Initialize emit_frame_chain.
9381 (aarch64_expand_prologue): Use emit_frame_chain.
9383 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9385 * config/aarch64/aarch64.c (aarch64_layout_frame):
9386 Ensure LR is always stored at the bottom of the callee-saves.
9387 Remove rarely used frame layout which saves callee-saves at top of
9388 frame, so the store of LR can be used as a valid probe in all cases.
9390 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9392 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9393 Improve unaligned TImode/TFmode base/offset split.
9395 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9396 Alan Hayward <alan.hayward@arm.com>
9397 David Sherwood <david.sherwood@arm.com>
9399 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9400 * combine.c (find_single_use_1): Likewise.
9401 (expand_field_assignment): Likewise.
9402 (move_deaths): Likewise.
9403 * lra-constraints.c (simplify_operand_subreg): Likewise.
9404 (curr_insn_transform): Likewise.
9405 * lra.c (collect_non_operand_hard_regs): Likewise.
9406 (add_regs_to_insn_regno_info): Likewise.
9407 * rtlanal.c (reg_referenced_p): Likewise.
9408 (covers_regno_no_parallel_p): Likewise.
9410 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9412 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9413 Don't print any bits outside the precision of the value.
9414 * wide-int.cc (test_printing): Add some new tests.
9416 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9418 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9419 supports -xbrace_comment option.
9420 * configure: Regenerate.
9421 * config.in: Regenerate.
9422 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9423 (ASM_CPU_SPEC): Use it.
9425 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9427 * target.def (static_rtx_alignment): New hook.
9428 * targhooks.h (default_static_rtx_alignment): Declare.
9429 * targhooks.c (default_static_rtx_alignment): New function.
9430 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9431 * doc/tm.texi: Regenerate.
9432 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9433 instead of targetm.constant_alignment. Remove call to
9435 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9436 (cris_preferred_mininum_alignment): New function, split out from...
9437 (cris_constant_alignment): ...here.
9438 (cris_static_rtx_alignment): New function.
9439 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9441 (ix86_constant_alignment): ...here.
9442 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9443 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9444 (mmix_static_rtx_alignment): New function.
9445 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9446 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9448 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9451 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9452 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9454 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9456 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9457 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9458 mark as a sign-extending load.
9459 (local_pic_load_u): Define.
9461 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9464 * fold-const.c (operand_equal_for_comparison_p): Also return true
9465 if ARG0 is a simple variant of ARG1 with narrower precision.
9466 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9468 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9470 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9472 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9473 scatter_static, scatter_per_elt.
9474 * x86-tune-costs.h: Add new cost entries.
9476 2017-10-25 Richard Biener <rguenther@suse.de>
9478 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9479 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9480 class pass_fre): Move to ...
9481 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9482 class pass_fre): ... here and adjust for statistics.
9484 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9487 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9488 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9489 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9490 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9493 2017-10-25 Richard Biener <rguenther@suse.de>
9495 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9496 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9497 eliminate_push_avail, eliminate_insert): Move inside...
9498 (class eliminate_dom_walker): ... this class in preparation
9500 (fini_eliminate): Remove by merging with ...
9501 (eliminate): ... this function. Adjust for class changes.
9502 (pass_pre::execute): Remove fini_eliminate call.
9503 (pass_fre::execute): Likewise.
9505 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9508 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9509 (VPERMI2, VPERMI2I): New mode iterators.
9510 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9511 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9513 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9514 mode iterator. Remove 3 old define_insn patterns.
9515 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9516 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9517 VPERMI2 mode iterator, remove the other two expanders.
9518 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9519 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9520 instructions, remove the other two patterns.
9521 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9522 mode iterator, remove the other two patterns.
9523 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9524 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9525 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9526 and adjust argument order accordingly.
9527 (ix86_expand_vec_perm): Adjust caller.
9528 (expand_vec_perm_1): Likewise.
9529 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9530 (expand_vec_perm_vpermt2_vpshub2): ... this.
9531 (ix86_expand_vec_perm_const_1): Adjust caller.
9532 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9535 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9536 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9537 (vec_shl_<mode>): Remove unused expander.
9538 (avx512bw_<shift_insn><mode>3): New define_insn.
9539 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9540 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9542 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9545 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9548 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9550 PR rtl-optimization/82396
9551 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9552 (autopref_multipass_init): Simplify initialization.
9553 (autopref_rank_data): Simplify sort order.
9554 * gcc/sched-int.h (autopref_multipass_data_): Remove
9555 multi_mem_insn_p, min_offset and max_offset.
9557 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9560 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9561 Check special value of flag_omit_frame_pointer.
9562 (aarch64_can_eliminate): Likewise.
9563 (aarch64_override_options_after_change_1): Simplify handling of
9564 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9566 2017-10-24 Richard Biener <rguenther@suse.de>
9568 PR tree-optimization/82697
9569 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9570 zero for conditional load and unconditional store.
9572 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9574 * doc/install.texi: Document bootstrap-cet.
9576 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9579 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9580 ENDBR instruction at function entrance if function is only
9583 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9586 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9587 patterns to better describe from which operation the CF is computed.
9588 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9589 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9590 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9591 is 0, use _0 suffixed expanders instead of emitting a comparison
9594 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9596 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9597 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9599 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9602 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9603 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9604 * loop-iv.c (iv_get_reaching_def): Likewise.
9605 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9606 variable is promoted and the partition contains undefined values.
9608 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9610 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9612 (nios2_address_cost): Define.
9613 (nios2_legitimize_address): Recognize (exp + constant) directly.
9614 (TARGET_ADDRESS_COST): Define.
9616 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9618 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9619 (nios2_symbolic_memory_operand_p): Declare.
9620 (nios2_split_large_constant): Declare.
9621 (nios2_split_symbolic_memory_operand): Declare.
9622 * config/nios2/nios2.c: Adjust includes.
9623 (nios2_symbolic_constant_allowed): New.
9624 (nios2_symbolic_constant_p): New.
9625 (nios2_plus_symbolic_constant_p): New.
9626 (nios2_valid_addr_expr_p): Recognize addresses involving
9628 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9629 (nios2_symbolic_memory_operand_p): New.
9630 (nios2_large_constant_p): New.
9631 (nios2_split_large_constant): New.
9632 (nios2_split_plus_large_constant): New.
9633 (nios2_split_symbolic_memory_operand): New.
9634 (nios2_legitimize_address): Code refactoring. Handle addresses
9635 involving symbolic constants.
9636 (nios2_emit_move_sequence): Likewise.
9637 (nios2_print_operand): Improve error output.
9638 (nios2_print_operand_address): Handle LO_SUM.
9639 (nios2_cdx_narrow_form_p): Likewise.
9640 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9641 operands involving symbolic constants.
9642 (movhi_internal, movsi_internal): Likewise.
9643 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9644 (extendhisi2, extendqi<mode>2): Likewise.
9646 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9648 * tree-pass.h (PROP_rtl_split_insns): Define.
9649 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9651 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9653 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9655 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9658 * target.def (const_not_ok_for_debug_p): Default to
9659 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9660 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9661 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9662 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9663 which targetm.const_not_ok_for_debug_p returned true.
9664 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9666 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9668 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9669 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9670 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9672 (ix86_const_not_ok_for_debug_p): New function.
9673 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9674 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9676 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9679 * system.h: Conditionally include "unique-ptr.h" if
9680 INCLUDE_UNIQUE_PTR is defined.
9681 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9682 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9684 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9686 * config/rl78/rl78.md: New define_expand "subdi3".
9688 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9691 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9692 DF_REF_INSN if DF_REF_INSN_INFO is false.
9694 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9696 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9697 xmm_move instead of sse_move.
9698 (sse_store_index): New function.
9699 (ix86_register_move_cost): Be more sensible about mismatch stall;
9700 model AVX moves correctly; make difference between sse->integer and
9702 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9703 moves; make difference between SSE and AVX.
9704 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9705 and zmm_move. Increase size of sse load and store tables;
9706 add unaligned load and store tables; add ssemmx_to_integer.
9707 * x86-tune-costs.h: Update all entries according to real
9708 move latencies from Agner Fog's manual and chip documentation.
9710 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9713 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9714 * config/i386/constraints.md (Wf): New constraint.
9715 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9716 (cmp<dwi>_doubleword): Removed.
9717 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9718 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9719 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9720 expand with cmp<dwi>_doubleword. For LTU and GEU use
9721 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9723 * common.opt (gcolumn-info): Enable by default.
9724 * doc/invoke.texi (gcolumn-info): Document new default.
9726 2017-10-23 Richard Biener <rguenther@suse.de>
9728 PR tree-optimization/82672
9729 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9730 Fold the stmt if we propagated into it.
9732 2017-10-23 Richard Biener <rguenther@suse.de>
9734 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9735 (bitmap_remove_expr_from_set): ... this. All callers call this
9736 for non-constant values.
9737 (bitmap_set_subtract): Rename to...
9738 (bitmap_set_subtract_expressions): ... this. Adjust and
9740 (bitmap_set_contains_value): Remove superfluous check.
9741 (bitmap_set_replace_value): Inline into single caller ...
9742 (bitmap_value_replace_in_set): ... here and simplify.
9743 (dependent_clean): Merge into ...
9744 (clean): ... this using an overload. Adjust.
9745 (prune_clobbered_mems): Adjust.
9746 (compute_antic_aux): Likewise.
9747 (compute_partial_antic_aux): Likewise.
9749 2017-10-23 Richard Biener <rguenther@suse.de>
9751 PR tree-optimization/82129
9753 2017-08-01 Richard Biener <rguenther@suse.de>
9755 PR tree-optimization/81181
9756 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9757 (compute_antic): ... end of iteration here.
9759 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9761 * target.def (starting_frame_offset): New hook.
9762 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9763 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9764 * doc/tm.texi.in: Regenerate.
9765 * hooks.h (hook_hwi_void_0): Declare.
9766 * hooks.c (hook_hwi_void_0): New function.
9767 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9768 STARTING_FRAME_OFFSET.
9769 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9770 * reload1.c (reload): Likewise.
9771 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9772 instead of STARTING_FRAME_OFFSET.
9773 * function.c (try_fit_stack_local): Likewise.
9774 (assign_stack_local_1): Likewise
9775 (instantiate_virtual_regs): Likewise.
9776 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9777 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9778 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9779 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9780 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9781 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9782 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9783 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9784 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9785 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9786 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9787 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9788 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9789 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9790 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9791 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9792 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9793 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
9794 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
9795 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
9796 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
9797 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
9798 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
9799 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
9800 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
9801 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
9802 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
9803 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
9804 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
9805 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
9806 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
9807 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
9808 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
9809 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
9810 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
9811 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
9812 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
9813 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
9814 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
9815 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
9816 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
9817 * config/avr/avr.c (avr_starting_frame_offset): Make static and
9818 return a HOST_WIDE_INT.
9819 (avr_builtin_setjmp_frame_value): Use it instead of
9820 STARTING_FRAME_OFFSET.
9821 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9822 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
9823 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
9825 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9826 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
9827 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
9828 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9829 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
9830 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
9831 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9832 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
9833 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
9834 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9835 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
9836 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
9838 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9839 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
9840 * config/mips/mips.c (mips_compute_frame_info): Refer to
9841 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9842 (mips_starting_frame_offset): New function.
9843 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9844 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
9845 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
9846 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
9847 and return a HOST_WIDE_INT.
9848 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9849 (mmix_initial_elimination_offset): Refer to
9850 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9851 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
9852 * config/pa/pa.c (pa_starting_frame_offset): New function.
9853 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
9854 (pa_expand_prologue): Likewise.
9855 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9856 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
9857 !FRAME_GROWS_DOWNWARD handling to...
9858 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9859 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
9860 !FRAME_GROWS_DOWNWARD handling to...
9861 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9862 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
9863 !FRAME_GROWS_DOWNWARD handling to...
9864 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9865 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
9867 (rs6000_starting_frame_offset): New function.
9868 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
9869 !FRAME_GROWS_DOWNWARD handling to...
9870 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9871 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
9872 !FRAME_GROWS_DOWNWARD handling to...
9873 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9874 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
9875 !FRAME_GROWS_DOWNWARD handling to...
9876 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9877 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
9878 (rs6000_starting_frame_offset): New function.
9879 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
9880 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
9881 * config/vax/vax.c (vax_starting_frame_offset): New function.
9882 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
9883 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9884 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
9885 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
9886 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9887 * system.h (STARTING_FRAME_OFFSET): Poison.
9889 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9891 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9892 SCALAR_TYPE_MODE instead of TYPE_MODE.
9894 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9895 Alan Hayward <alan.hayward@arm.com>
9896 David Sherwood <david.sherwood@arm.com>
9898 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
9900 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9901 Alan Hayward <alan.hayward@arm.com>
9902 David Sherwood <david.sherwood@arm.com>
9904 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
9906 2017-10-23 Richard Biener <rguenther@suse.de>
9908 PR tree-optimization/82129
9909 * tree-ssa-pre.c (bitmap_set_and): Remove.
9910 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
9911 canonicalizing expressions in the set to those with lowest
9912 ID rather than taking that from the first edge.
9914 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9916 * combine.c (rtx_equal_for_field_assignment_p): Use
9917 byte_lowpart_offset.
9919 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9920 Alan Hayward <alan.hayward@arm.com>
9921 David Sherwood <david.sherwood@arm.com>
9923 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
9924 to a SUBREG_PROMOTED_VAR.
9926 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9927 Alan Hayward <alan.hayward@arm.com>
9928 David Sherwood <david.sherwood@arm.com>
9930 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
9931 (expand_debug_source_expr): Likewise.
9932 * combine.c (combine_simplify_rtx): Likewise.
9933 * cse.c (fold_rtx): Likewise.
9934 * optabs.c (expand_float): Likewise.
9935 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9936 (simplify_binary_operation_1): Likewise.
9938 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9939 Alan Hayward <alan.hayward@arm.com>
9940 David Sherwood <david.sherwood@arm.com>
9942 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
9943 (record_promoted_value): Likewise.
9944 * expr.c (expand_expr_real_2): Likewise.
9945 * ree.c (update_reg_equal_equiv_notes): Likewise.
9946 (combine_set_extension): Likewise.
9947 * rtlanal.c (low_bitmask_len): Likewise.
9948 * simplify-rtx.c (neg_const_int): Likewise.
9949 (simplify_binary_operation_1): Likewise.
9951 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9952 Alan Hayward <alan.hayward@arm.com>
9953 David Sherwood <david.sherwood@arm.com>
9955 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
9956 * regcprop.c (maybe_mode_change): Likewise.
9957 * reload1.c (alter_reg): Likewise.
9959 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9961 * inchash.h (inchash::hash::add_wide_int): New function.
9962 * lto-streamer-out.c (hash_tree): Use it.
9964 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9966 * inchash.h (inchash::hash::add_wide_int): Rename to...
9967 (inchash::hash::add_hwi): ...this.
9968 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
9969 (polymorphic_call_target_hasher::hash): Likewise.
9970 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
9971 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
9972 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
9973 * lto-streamer-out.c (hash_tree): Likewise.
9974 * optc-save-gen.awk: Likewise.
9975 * tree.c (add_expr): Likewise.
9977 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9980 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
9981 for ordered inequality comparisons even with TARGET_IEEE_FP.
9983 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9986 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
9987 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9988 Expand with cmp<dwi>_doubleword.
9990 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9992 * extend.texi: Add x86 specific to 'nocf_check' attribute.
9993 List CET intrinsics.
9994 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
9995 specific to -fcf-protection option.
9997 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9999 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10000 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10001 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10002 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10003 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10004 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10005 (extra_objs): Add cet.o for Linux/x86 targets.
10006 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10007 * config/i386/cet.c: New file.
10008 * config/i386/cetintrin.h: Likewise.
10009 * config/i386/t-cet: Likewise.
10010 * config/i386/cpuid.h (bit_SHSTK): New.
10011 (bit_IBT): Likewise.
10012 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10013 pass IBT and SHSTK bits.
10014 * config/i386/i386-builtin-types.def
10015 (VOID_FTYPE_UNSIGNED_PVOID): New.
10016 (VOID_FTYPE_UINT64_PVOID): Likewise.
10017 * config/i386/i386-builtin.def: Add CET intrinsics.
10018 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10019 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10020 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10021 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10023 * config/i386/i386.c (rest_of_insert_endbranch): New.
10024 (pass_data_insert_endbranch): Likewise.
10025 (pass_insert_endbranch): Likewise.
10026 (make_pass_insert_endbranch): Likewise.
10027 (ix86_notrack_prefixed_insn_p): Likewise.
10028 (ix86_target_string): Add -mibt, -mshstk flags.
10029 (ix86_option_override_internal): Add flag_cf_protection
10031 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10032 (ix86_print_operand): Add 'notrack' prefix output.
10033 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10034 (ix86_expand_builtin): Expand CET intrinsics.
10035 (x86_output_mi_thunk): Add 'endbranch' instruction.
10036 * config/i386/i386.h (TARGET_IBT): New.
10037 (TARGET_IBT_P): Likewise.
10038 (TARGET_SHSTK): Likewise.
10039 (TARGET_SHSTK_P): Likewise.
10040 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10041 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10042 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10043 (builtin_setjmp_setup): New pattern.
10044 (builtin_longjmp): Likewise.
10045 (rdssp<mode>): Likewise.
10046 (incssp<mode>): Likewise.
10047 (saveprevssp): Likewise.
10048 (rstorssp): Likewise.
10049 (wrss<mode>): Likewise.
10050 (wruss<mode>): Likewise.
10051 (setssbsy): Likewise.
10052 (clrssbsy): Likewise.
10053 (nop_endbr): Likewise.
10054 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10056 * config/i386/immintrin.h: Include <cetintrin.h>.
10057 * config/i386/linux-common.h
10058 (file_end_indicate_exec_stack_and_cet): New prototype.
10059 (TARGET_ASM_FILE_END): New.
10061 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10063 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10064 latencies instead of having separate table; make difference between
10065 integer and float costs.
10066 * i386.h (processor_costs): Remove scalar_stmt_cost,
10067 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10068 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10070 * x86-tune-costs.h: Remove entries which has been removed in
10071 procesor_costs from all tables; make cond_taken_branch_cost
10072 and cond_not_taken_branch_cost COST_N_INSNS based.
10074 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10076 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10078 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10080 * config/i386/i386.md (isa): Remove fma_avx512f.
10081 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10082 <avx512>_fmadd_<mode>_mask3<round_name>,
10083 <avx512>_fmsub_<mode>_mask<round_name>,
10084 <avx512>_fmsub_<mode>_mask3<round_name>,
10085 <avx512>_fnmadd_<mode>_mask<round_name>,
10086 <avx512>_fnmadd_<mode>_mask3<round_name>,
10087 <avx512>_fnmsub_<mode>_mask<round_name>,
10088 <avx512>_fnmsub_<mode>_mask3<round_name>,
10089 <avx512>_fmaddsub_<mode>_mask<round_name>,
10090 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10091 <avx512>_fmsubadd_<mode>_mask<round_name>,
10092 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10093 (*vec_widen_umult_even_v16si<mask_name>,
10094 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10095 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10097 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10099 * extend.texi: Add 'nocf_check' documentation.
10100 * gimple.texi: Add second parameter to
10101 gimple_build_call_from_tree.
10102 * invoke.texi: Add -fcf-protection documentation.
10103 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10105 2017-10-20 Richard Biener <rguenther@suse.de>
10107 PR tree-optimization/82473
10108 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10109 the largest input type.
10111 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10113 * c-attribs.c (handle_nocf_check_attribute): New function.
10114 (c_common_attribute_table): Add 'nocf_check' handling.
10115 * gimple-parser.c: Add second argument NULL to
10116 gimple_build_call_from_tree.
10117 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10118 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10120 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10121 * common.opt: Add fcf-protection flag.
10122 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10123 * flag-types.h: Add enum cf_protection_level.
10124 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10125 Add 'nocf_check' attribute propagation to gimple call.
10126 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10127 (gimple_build_call_from_tree): Update prototype.
10128 (gimple_call_nocf_check_p): New function.
10129 (gimple_call_set_nocf_check): Likewise.
10130 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10131 * ipa-icf.c: Add nocf_check attribute in statement hash.
10132 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10133 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10134 * toplev.c (process_options): Add flag_cf_protection handling.
10136 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10138 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10140 2017-10-20 Richard Biener <rguenther@suse.de>
10142 PR tree-optimization/82603
10143 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10144 remove false predicated stores.
10146 2017-10-20 Richard Biener <rguenther@suse.de>
10148 * graphite-isl-ast-to-gimple.c
10149 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10150 Remove return value and simplify, dump copied stmt after lhs
10152 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10153 Reduce dump verbosity.
10154 (gsi_insert_earliest): Likewise.
10155 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10156 * graphite.c (print_global_statistics): Adjust dumping.
10157 (print_graphite_scop_statistics): Likewise.
10158 (print_graphite_statistics): Do not dump loops here.
10159 (graphite_transform_loops): But here.
10161 2017-10-20 Nicolas Roche <roche@adacore.com>
10163 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10164 * configure: Regenerate.
10166 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10169 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10170 functions when optimizing replace GIMPLE_RETURN stmts with
10171 calls to __builtin_unreachable ().
10174 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10175 for -fsanitize=thread link of executables.
10176 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10177 link of executables.
10180 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10181 New mode iterators.
10182 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10183 define_insns for logical vector shifts to use VI248_AVX512BW
10184 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10185 condition, useless isa and prefix attributes. Change the first
10186 2 of these define_insns to ...
10187 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10188 define_insn for avx512vl.
10189 (<shift_insn><mode>3): ... and this, new define_insn without
10190 masking for non-avx512vl.
10193 * config/i386/sse.md (*andnot<mode>3,
10194 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10195 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10196 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10197 not applied use empty suffix even for TARGET_AVX512VL.
10198 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10199 is applied, supply evex,evex or evex,evex,evex instead of just
10202 2017-10-20 Julia Koval <julia.koval@intel.com>
10204 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10205 (OPTION_MASK_ISA_GFNI_UNSET): New.
10206 (ix86_handle_option): Handle OPT_mgfni.
10207 * config/i386/cpuid.h (bit_GFNI): New.
10208 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10209 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10210 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10211 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10212 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10213 * config/i386/i386.opt: Add mgfni.
10215 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10217 * config/msp430/msp430.c (msp430_option_override): Disable
10218 -fdelete-null-pointer-checks.
10219 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10221 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10223 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10224 of x87 and SSE instructions.
10226 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10228 * asan.c (create_cond_insert_point): Do not update edge count.
10229 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10230 (afdo_propagate_circuit): Likewise.
10231 (afdo_calculate_branch_prob): Likewise.
10232 (afdo_annotate_cfg): Likewise.
10233 * basic-block.h (struct edge_def): Remove count.
10234 (edge_def::count): New accessor.
10235 * bb-reorder.c (rotate_loop): Update.
10236 (find_traces_1_round): Update.
10237 (connect_traces): Update.
10238 (sanitize_hot_paths): Update.
10239 * cfg.c (unchecked_make_edge): Update.
10240 (make_single_succ_edge): Update.
10241 (check_bb_profile): Update.
10242 (dump_edge_info): Update.
10243 (update_bb_profile_for_threading): Update.
10244 (scale_bbs_frequencies_int): Update.
10245 (scale_bbs_frequencies_gcov_type): Update.
10246 (scale_bbs_frequencies_profile_count): Update.
10247 (scale_bbs_frequencies): Update.
10248 * cfganal.c (connect_infinite_loops_to_exit): Update.
10249 * cfgbuild.c (compute_outgoing_frequencies): Update.
10250 (find_many_sub_basic_blocks): Update.
10251 * cfgcleanup.c (try_forward_edges): Update.
10252 (try_crossjump_to_edge): Update
10253 * cfgexpand.c (expand_gimple_cond): Update
10254 (expand_gimple_tailcall): Update
10255 (construct_exit_block): Update
10256 * cfghooks.c (verify_flow_info): Update
10257 (redirect_edge_succ_nodup): Update
10258 (split_edge): Update
10259 (make_forwarder_block): Update
10260 (duplicate_block): Update
10261 (account_profile_record): Update
10262 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10263 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10264 * cfgloopmanip.c (scale_loop_profile): Update.
10266 (lv_adjust_loop_entry_edge): Update.
10267 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10268 (force_nonfallthru_and_redirect): Update.
10269 (purge_dead_edges): Update.
10270 (rtl_flow_call_edges_add): Update.
10271 * cgraphunit.c (init_lowered_empty_function): Update.
10272 (cgraph_node::expand_thunk): Update.
10273 * gimple-pretty-print.c (dump_probability): Update.
10274 (dump_edge_probability): Update.
10275 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10276 * haifa-sched.c (sched_create_recovery_edges): Update.
10277 * hsa-gen.c (convert_switch_statements): Update.
10278 * ifcvt.c (dead_or_predicable): Update.
10279 * ipa-inline-transform.c (inline_transform): Update.
10280 * ipa-split.c (split_function): Update.
10281 * ipa-utils.c (ipa_merge_profiles): Update.
10282 * loop-doloop.c (add_test): Update.
10283 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10284 * lto-streamer-in.c (input_cfg): Update.
10285 (input_function): Update.
10286 * lto-streamer-out.c (output_cfg): Update.
10287 * modulo-sched.c (sms_schedule): Update.
10288 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10289 * predict.c (maybe_hot_edge_p): Update.
10290 (unlikely_executed_edge_p): Update.
10291 (probably_never_executed_edge_p): Update.
10292 (dump_prediction): Update.
10293 (drop_profile): Update.
10294 (propagate_unlikely_bbs_forward): Update.
10295 (determine_unlikely_bbs): Update.
10296 (force_edge_cold): Update.
10297 * profile.c (compute_branch_probabilities): Update.
10298 * reg-stack.c (better_edge): Update.
10299 * shrink-wrap.c (handle_simple_exit): Update.
10300 * tracer.c (better_p): Update.
10301 * trans-mem.c (expand_transaction): Update.
10302 (split_bb_make_tm_edge): Update.
10303 * tree-call-cdce.c: Update.
10304 * tree-cfg.c (gimple_find_sub_bbs): Update.
10305 (gimple_split_edge): Update.
10306 (gimple_duplicate_sese_region): Update.
10307 (gimple_duplicate_sese_tail): Update.
10308 (gimple_flow_call_edges_add): Update.
10309 (insert_cond_bb): Update.
10310 (execute_fixup_cfg): Update.
10311 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10312 * tree-complex.c (expand_complex_div_wide): Update.
10313 * tree-eh.c (lower_resx): Update.
10314 (unsplit_eh): Update.
10315 (cleanup_empty_eh_move_lp): Update.
10316 * tree-inline.c (copy_edges_for_bb): Update.
10317 (freqs_to_counts): Update.
10318 (copy_cfg_body): Update.
10319 * tree-ssa-dce.c (remove_dead_stmt): Update.
10320 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10321 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10322 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10323 (unloop_loops): Update.
10324 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10325 * tree-ssa-loop-split.c (connect_loops): Update.
10326 (split_loop): Update.
10327 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10328 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10329 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10330 * tree-ssa-reassoc.c (branch_fixup): Update.
10331 * tree-ssa-tail-merge.c (replace_block_by): Update.
10332 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10333 (compute_path_counts): Update.
10334 (update_profile): Update.
10335 (recompute_probabilities): Update.
10336 (update_joiner_offpath_counts): Update.
10337 (estimated_freqs_path): Update.
10338 (freqs_to_counts_path): Update.
10339 (clear_counts_path): Update.
10340 (ssa_fix_duplicate_block_edges): Update.
10341 (duplicate_thread_path): Update.
10342 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10343 (case_bit_test_cmp): Update.
10344 (collect_switch_conv_info): Update.
10345 (gen_inbound_check): Update.
10346 (do_jump_if_equal): Update.
10347 (emit_cmp_and_jump_insns): Update.
10348 * tree-tailcall.c (decrease_profile): Update.
10349 (eliminate_tail_call): Update.
10350 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10351 (vect_do_peeling): Update.
10352 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10353 * ubsan.c (ubsan_expand_null_ifn): Update.
10354 (ubsan_expand_ptr_ifn): Update.
10355 * value-prof.c (gimple_divmod_fixed_value): Update.
10356 (gimple_mod_pow2): Update.
10357 (gimple_mod_subtract): Update.
10358 (gimple_ic): Update.
10359 (gimple_stringop_fixed_value): Update.
10361 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10364 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10366 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10368 PR rtl-optimization/82395
10369 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10370 based on non_spilled_static_chain_regno_p.
10372 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10374 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10375 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10376 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10378 2017-10-19 Martin Sebor <msebor@redhat.com>
10380 PR tree-optimization/82596
10381 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10383 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10385 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10386 (get_mem_refs_of_builtin_call): Likewise.
10387 * builtins.c (expand_builtin_apply): Adjust call to
10388 allocate_dynamic_stack_space.
10389 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10390 the third argument to allocate_dynamic_stack_space, otherwise -1.
10391 (expand_builtin): Deal with all alloca variants.
10392 (is_inexpensive_builtin): Likewise.
10393 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10394 * calls.c (special_function_p): Deal with all alloca variants.
10395 (initialize_argument_information): Adjust call to
10396 allocate_dynamic_stack_space.
10397 (expand_call): Likewise.
10398 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10399 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10400 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10401 use it for the stack usage computation.
10402 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10403 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10404 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10405 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10406 (in_loop_p): Remove first argument and useless check.
10407 (pass_walloca::execute): Remove useless test and adjust call to above.
10408 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10409 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10410 (gimplify_call_expr): Deal with all alloca variants.
10411 * hsa-gen.c (gen_hsa_alloca): Likewise.
10412 (gen_hsa_insns_for_call): Likewise.
10413 * ipa-pure-const.c (special_builtin_state): Likewise.
10414 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10415 * tree-object-size.c (alloc_object_size): Likewise.
10416 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10417 (call_may_clobber_ref_p_1): Likewise.
10418 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10419 (ccp_fold_stmt): Likewise.
10420 (optimize_stack_restore): Likewise.
10421 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10422 (mark_all_reaching_defs_necessary_1): Likewise.
10423 (propagate_necessity): Likewise.
10424 (eliminate_unnecessary_stmts): Likewise.
10425 * tree.c (build_common_builtin_nodes): Build
10426 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10427 (build_alloca_call_expr): New function.
10428 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10429 (CASE_BUILT_IN_ALLOCA): Likewise.
10430 (build_alloca_call_expr): Declare.
10431 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10433 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10436 * dwarf2out.c (new_die_raw): New static inline function.
10437 (new_die): Use it to create the DIE.
10438 (add_AT_external_die_ref): Likewise.
10439 (clone_die): Likewise.
10440 (clone_as_declaration): Likewise.
10441 (dwarf2out_vms_debug_main_pointer): Likewise.
10442 (base_type_die): Likewise. Remove early return for corner cases.
10443 Do not call add_pubtype on the DIE here.
10444 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10445 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10446 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10447 native order exists for base types, attach the DIE manually and call
10448 add_pubtype on it. Do not equate a reverse order DIE to the type.
10450 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10452 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10453 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10455 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10457 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10459 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10460 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10461 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10462 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10463 * opts.c (sanitizer_opts): Add builtin.
10464 * ubsan.c (instrument_builtin): New function.
10465 (pass_ubsan::execute): Call it.
10466 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10467 * doc/invoke.texi: Document -fsanitize=builtin.
10469 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10470 builtins, store max (log2 (align), 0) into uchar field instead of
10471 align into uptr field.
10472 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10473 store uchar 0 field instead of uptr 0 field.
10474 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10475 instead of passing one address of struct with 2 locations pass
10476 two addresses of structs with 1 location each.
10477 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10478 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10479 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10480 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10481 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10482 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10483 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10484 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10486 2017-10-19 Martin Liska <mliska@suse.cz>
10489 * file-find.c (remove_prefix): Remove.
10490 * file-find.h (remove_prefix): Likewise.
10491 * gcc-ar.c: Remove smartness of lookup.
10493 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10495 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10496 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10497 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10499 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10502 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10503 (setcc + and to xor + setcc): New peephole2.
10505 2017-10-19 Tom de Vries <tom@codesourcery.com>
10507 * doc/sourcebuild.texi (Test Directives, Variants of
10508 dg-require-support): Add dg-require-stack-size.
10510 2017-10-19 Martin Liska <mliska@suse.cz>
10513 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10514 that have a large alignment.
10515 (gimplify_target_expr): Likewise.
10517 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10519 PR rtl-optimization/82602
10520 * ira.c (rtx_moveable_p): Return false for volatile asm.
10522 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10525 * config/i386/i386-modes.def (CCGZ): New CC mode.
10526 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10527 * config/i386/predicates.md (ix86_comparison_operator):
10529 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10530 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10531 with double-word subtraction.
10532 (put_condition_code): Handle CCGZmode.
10534 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10536 * wide-int.cc (debug (const wide_int &)): New.
10537 (debug (const wide_int *)): New.
10538 (debug (const widest_int &)): New.
10539 (debug (const widest_int *)): New.
10541 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10543 PR middle-end/82556
10544 * lra-constraints.c (curr_insn_transform): Use non-input operand
10545 instead of output one for matched reload.
10547 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10549 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10550 (tree-ssa-loop-ivopts.h): New header file.
10551 (struct builtin_info): New fields.
10552 (classify_builtin_1): Compute and record base and offset parts for
10553 memset builtin partition by calling strip_offset.
10554 (offset_cmp, fuse_memset_builtins): New functions.
10555 (finalize_partitions): Fuse adjacent memset partitions by calling
10557 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10558 Expose the interface.
10559 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10561 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10563 PR tree-optimization/82574
10564 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10565 that data reference must be executed exactly once per iteration
10566 against the outermost loop in nest.
10567 (classify_partition): Update call to above function.
10569 2017-10-18 Richard Biener <rguenther@suse.de>
10571 PR tree-optimization/82591
10572 * graphite.c (graphite_transform_loops): Move code gen message
10574 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10575 Here. Handle scop_to_isl_ast failing.
10576 (scop_to_isl_ast): Limit the number of ISL operations.
10578 2017-10-18 Richard Biener <rguenther@suse.de>
10580 * graphite-isl-ast-to-gimple.c
10581 (translate_isl_ast_to_gimple::set_rename): Simplify.
10582 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10583 (graphite_copy_stmts_from_block): ... here.
10584 (copy_bb_and_scalar_dependences): Simplify.
10585 (add_parameters_to_ivs_params): Canonicalize.
10586 (generate_entry_out_of_ssa_copies): Simplify.
10587 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10588 by passing in ISL dimension.
10589 (parameter_index_in_region_1): Rename to ...
10590 (parameter_index_in_region): ... this.
10591 (extract_affine): Adjust assert, pass down parameter index.
10592 (add_param_constraints): Use range-info when available.
10593 (build_scop_context): Adjust.
10594 * sese.c (new_sese_info): Adjust.
10595 (free_sese_info): Likewise.
10596 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10597 Remove unused typedefs.
10598 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10600 2017-10-18 Martin Liska <mliska@suse.cz>
10602 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10604 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10606 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10607 (s390_sched_init): Do not reset s390_sched_state if we entered the
10608 current basic block via a fallthru edge and all others are unlikely.
10610 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10612 * config/s390/s390.c (NUM_SIDES): New variable.
10613 (LONGRUNNING_THRESHOLD): New variable.
10614 (LATENCY_FACTOR): New variable.
10615 (s390_sched_score): Decrease score for long-running instructions on
10617 (s390_sched_variable_issue): Perform bookkeeping for long-running
10620 2017-10-18 Richard Biener <rguenther@suse.de>
10622 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10623 Simplify with removal of the parameter rename map.
10624 (set_rename): Likewise.
10625 (should_copy_to_new_region): Likewise.
10626 (graphite_copy_stmts_from_block): Likewise.
10627 (copy_bb_and_scalar_dependences): Remove initialization of
10628 unused copied_bb_map.
10629 (copy_def): Remove.
10630 (copy_internal_parameters): Likewise.
10631 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10632 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10633 Use INTEGRAL_TYPE_P.
10634 (parameter_index_in_region_1): Rename to ...
10635 (assign_parameter_index_in_region): ... this. Assert we have
10636 a parameter we handle.
10637 (scan_tree_for_params): Adjust.
10638 * sese.h (parameter_rename_map_t): Remove.
10639 (struct sese_info_t): Remove unused parameter_rename_map and
10640 copied_bb_map members.
10641 * sese.c (new_sese_info): Adjust.
10642 (free_sese_info): Likewise.
10644 2017-10-18 Martin Liska <mliska@suse.cz>
10647 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10648 on an abnormal edge.
10650 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10652 * doc/invoke.texi (ffunction-sections and fdata-sections):
10655 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10657 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10658 the use statement can throw internally.
10660 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10662 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10663 any RTX present on the RHS of a SET.
10664 * compare-elim.c (try_eliminate_compare): Restore comment.
10666 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10668 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10669 may be also called on tcc_type nodes.
10670 * langhooks.c (lhd_tree_size): Likewise.
10672 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10674 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10675 format_warning_at_substring.
10676 (maybe_warn): Convert source_range * param to a location_t. Pass
10677 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10678 (format_directive): Remove code to extract source_ranges and
10679 source_range * in favor of just a location_t.
10680 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10682 * substring-locations.c (format_warning_va): Convert
10683 source_range * param to a location_t.
10684 (format_warning_at_substring): Likewise.
10685 * substring-locations.h (format_warning_va): Likewise.
10686 (format_warning_at_substring): Likewise.
10688 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10690 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10692 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10693 and scatter/gather ops.
10695 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10696 vec_gather_load and vec_scatter_store.
10697 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10698 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10699 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10700 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10701 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10702 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10704 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10706 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10707 Detect FTST instruction and handle its register pops. Only pop
10708 second operand if can_pop_second_op is true.
10709 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10710 set can_pop_second_op to false in the compare_for_stack_reg call.
10712 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10713 output_fp_compare for stack register operands.
10714 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10715 instructions here. Do not emit stack register pops here. Assert
10716 that FCOMPP pops next to top stack register. Rewrite function.
10718 2017-10-17 Nathan Sidwell <nathan@acm.org>
10720 PR middle-end/82577
10721 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10722 use DECL_ASSEMBLER_NAME_RAW.
10724 PR middle-end/82546
10725 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10728 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10729 Wilco Dijkstra <wilco.dijkstra@arm.com>
10731 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10732 converstion to Pmode from the buf_addr.
10734 2017-10-17 Richard Biener <rguenther@suse.de>
10736 * graphite-dependences.c (scop_get_reads_and_writes): Change
10737 output parameters to references.
10739 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10741 PR 71026/tree-optimization
10742 * fold-const.c (distribute_real_division): Removed.
10743 (fold_binary_loc): Remove calls to distribute_real_divison.
10745 2017-10-17 Richard Biener <rguenther@suse.de>
10747 * graphite-scop-detection.c
10748 (scop_detection::stmt_has_simple_data_refs_p): Always use
10749 the full nest as region.
10750 (try_generate_gimple_bb): Likewise.
10751 * sese.c (scalar_evolution_in_region): Simplify now that
10752 SCEV can handle instantiation in regions.
10753 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10754 in the non-loop part of a function if requested.
10756 2017-10-17 Richard Biener <rguenther@suse.de>
10758 PR tree-optimization/82563
10759 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10761 (graphite_regenerate_ast_isl): Call it.
10762 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10764 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10766 PR tree-optimization/82549
10767 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10768 Formatting fixes. Instead of calling make_bit_field_ref with negative
10771 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10773 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10774 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10775 _mm_maskz_reduce_ss): New.
10776 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10777 __builtin_ia32_reducess_mask): Ditto..
10778 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10779 * config/i386/sse.md (reduces<mode>): Renamed to ...
10780 (reduces<mode><mask_scalar_name>): ... this.
10781 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10783 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10784 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10786 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10788 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10789 * selftest-run-tests.c (selftest::run_tests): Call
10790 selftest::unique_ptr_tests_cc_tests.
10791 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10792 * unique-ptr-tests.cc: New file.
10794 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
10797 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10799 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10800 (make_hard_regno_born, make_hard_regno_dead): Update
10801 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
10803 2017-10-16 Jeff Law <law@redhat.com>
10805 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10807 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
10809 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
10811 2017-10-16 Olivier Hainque <hainque@adacore.com>
10813 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
10814 with_cpu if we were configured for an e500v2 target cpu name.
10816 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
10818 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
10819 * doc/invoke.texi: Document +nodsp as a valid extension for
10822 2017-10-16 Martin Liska <mliska@suse.cz>
10824 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
10825 (test_set_range): Likewise.
10826 (test_range_functions): Rename to ...
10827 (test_bit_in_range): ... this.
10828 (sbitmap_c_tests): Add new test.
10830 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10832 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
10834 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
10835 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
10837 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10839 * config/aarch64/aarch64-builtins.c
10840 (aarch64_types_quadopu_lane_qualifiers): New.
10841 (TYPES_QUADOPU_LANE): New.
10842 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
10843 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
10844 (aarch64_<sur>dot_laneq<vsi2qi>): New.
10845 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
10846 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
10847 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
10848 (Vdottype, DOTPROD): New.
10849 (sur): Add SDOT and UDOT.
10851 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10853 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
10854 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
10855 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
10856 Add TARGET_DOTPROD.
10857 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
10858 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
10859 Enable TARGET_DOTPROD.
10860 (cortex-a75.cortex-a55): Likewise.
10861 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
10863 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10865 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
10866 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
10868 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
10870 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
10871 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
10872 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
10873 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
10874 * config/arm/types.md (neon_dot, neon_dot_q): New.
10875 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
10877 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10879 * config/arm/arm.h (TARGET_DOTPROD): New.
10880 * config/arm/arm.c (arm_arch_dotprod): New.
10881 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
10882 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
10883 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
10884 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
10885 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
10886 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
10887 * doc/invoke.texi (armv8.2-a): Document dotprod
10889 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10891 * i386.c (ix86_vec_cost): New function.
10892 (ix86_rtx_costs): Handle vector operations better.
10893 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
10894 * x86-tune-costs.h: Add new costs to all tables.
10896 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10898 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
10900 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
10901 divsd, sqrtss and sqrtsd
10902 * x86-tune-costs.h: Add new entries to all costs.
10903 (znver1_cost): Fix to match real instruction latencies.
10905 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10906 Michael Collison <michael.collison@arm.com>
10908 * compare-elim.c: Include emit-rtl.h.
10909 (can_merge_compare_into_arith): New function.
10910 (try_validate_parallel): Likewise.
10911 (try_merge_compare): Likewise.
10912 (try_eliminate_compare): Call the above when no previous clobber
10914 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
10917 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10919 PR middle-end/62263
10920 PR middle-end/82498
10921 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
10922 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
10924 PR middle-end/62263
10925 PR middle-end/82498
10926 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
10927 to be any operand_equal_p operands. For & (B - 1) require
10928 B to be power of 2. Recognize
10929 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
10931 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10934 * optabs.c (expand_memory_blockage): Fix call of
10935 targetm.have_memory_blockage.
10937 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10940 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
10941 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
10942 objects to extra_objs instead of overwriting it.
10944 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10946 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
10947 Use any_fp_register_operand as operand[3] predicate. Simplify
10948 equality test for operands[2] and operands[4] memory location.
10949 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
10950 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
10951 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
10952 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
10953 any_fp_register_operand as operand[1] predicate. Simplify
10954 equality test for operands[0] and operands[3] memory location.
10955 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
10956 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
10957 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
10959 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10961 * target-insns.def: Add memory_blockage.
10962 * optabs.c (expand_memory_blockage): New function.
10963 (expand_asm_memory_barrier): Rename ...
10964 (expand_asm_memory_blockage): ... to this.
10965 (expand_mem_thread_fence): Call expand_memory_blockage
10966 instead of expand_asm_memory_barrier.
10967 (expand_mem_singnal_fence): Ditto.
10968 (expand_atomic_load): Ditto.
10969 (expand_atomic_store): Ditto.
10970 * doc/md.texi (Standard Pattern Names For Generation):
10971 Document memory_blockage instruction pattern.
10973 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
10975 * config/rl78/rl78.c (rl78_emit_libcall): New function.
10976 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
10977 * config/rl78/rl78.md: New define_expand "adddi3".
10979 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10981 * cfghooks.c (verify_flow_info): Disable check that all probabilities
10984 2017-10-13 Jeff Law <law@redhat.com>
10986 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
10988 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10991 * internal-fn.c (expand_mul_overflow): If both operands have
10992 the same highpart of -1 or 0 and the topmost bit of lowpart
10993 is different, overflow is if res <= 0 rather than res < 0.
10995 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
10997 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
10998 TARGET_P9_VECTOR code for unaligned_load case.
11000 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11002 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11004 2017-10-13 Nathan Sidwell <nathan@acm.org>
11006 * tree-core.h (tree_contains_struct): Make bool.
11007 * tree.c (tree_contains_struct): Likewise.
11008 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11009 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11010 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11011 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11012 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11014 2017-10-13 Richard Biener <rguenther@suse.de>
11016 * graphite-isl-ast-to-gimple.c
11017 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11018 parameters and dominance check.
11019 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11020 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11021 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11022 Do not update SSA form here or do intermediate IL verification.
11023 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11024 (graphite_initialize): Remove check on the number of loops in
11025 the function and inline into graphite_transform_loops.
11026 (graphite_finalize): Inline into graphite_transform_loops.
11027 (graphite_transform_loops): Perform SSA update and IL verification
11029 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11031 2017-10-13 Richard Biener <rguenther@suse.de>
11033 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11034 graphite_expression_type_precision): Avoid global constructor
11036 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11037 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11038 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11039 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11040 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11041 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11043 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11046 * config/i386/i386.h (ix86_red_zone_size): New.
11047 * config/i386/i386.md (push peephole2s): Replace
11048 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11050 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11051 Alan Hayward <alan.hayward@arm.com>
11052 David Sherwood <david.sherwood@arm.com>
11054 * combine.c (can_change_dest_mode): Reject changes in
11055 REGMODE_NATURAL_SIZE.
11057 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11058 Alan Hayward <alan.hayward@arm.com>
11059 David Sherwood <david.sherwood@arm.com>
11061 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11062 (expand_debug_source_expr): Likewise.
11063 * combine.c (combine_simplify_rtx): Likewise.
11064 * cse.c (fold_rtx): Likewise.
11065 * fwprop.c (canonicalize_address): Likewise.
11066 * targhooks.c (default_shift_truncation_mask): Likewise.
11068 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11069 Alan Hayward <alan.hayward@arm.com>
11070 David Sherwood <david.sherwood@arm.com>
11072 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11073 (widened_mode): Likewise.
11074 (expand_unop): Likewise.
11075 * ree.c (transform_ifelse): Likewise.
11076 (merge_def_and_ext): Likewise.
11077 (combine_reaching_defs): Likewise.
11078 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11080 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11081 Alan Hayward <alan.hayward@arm.com>
11082 David Sherwood <david.sherwood@arm.com>
11084 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11085 * combine.c (gen_lowpart_for_combine): Likewise.
11086 * dwarf2out.c (rtl_for_decl_location): Likewise.
11087 * final.c (alter_subreg): Likewise.
11088 * rtlhooks.c (gen_lowpart_general): Likewise.
11089 (gen_lowpart_if_possible): Likewise.
11091 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11092 Alan Hayward <alan.hayward@arm.com>
11093 David Sherwood <david.sherwood@arm.com>
11095 * calls.c (expand_call): Use subreg_lowpart_offset.
11096 * cse.c (cse_insn): Likewise.
11097 * regcprop.c (copy_value): Likewise.
11098 (copyprop_hardreg_forward_1): Likewise.
11100 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11103 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11104 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11105 =Q constraints to +Q and into insn condition add check
11106 that operands[0] and operands[1] are equal.
11107 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11108 =Q constraints to +Q and into insn condition add check
11109 that operands[0] is equal to either operands[1] or operands[2].
11112 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11113 instead of handling MINUS_EXPR twice (once for each argument),
11114 canonicalize operand order and handle just once, use rtype where
11115 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11118 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11119 any values of __C while still being pattern recognizable as a simple
11120 rotate instruction.
11122 2017-10-13 Richard Biener <rguenther@suse.de>
11124 PR tree-optimization/82451
11126 2017-10-02 Richard Biener <rguenther@suse.de>
11128 PR tree-optimization/82355
11129 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11130 a mapping for the enclosing loop but avoid generating one for
11131 the loop tree root.
11132 (copy_bb_and_scalar_dependences): Remove premature codegen
11133 error on PHIs in blocks duplicated into multiple places.
11134 * graphite-scop-detection.c
11135 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11136 in the region use it as loop and nest to analyze the DR in.
11137 (try_generate_gimple_bb): Likewise.
11138 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11139 (add_loop_constraints): For blocks in a loop not in the region
11140 create a dimension with a single iteration.
11141 * sese.h (gbb_loop_at_index): Remove assert.
11143 * cfgloop.c (loop_preheader_edge): For the loop tree root
11144 return the single successor of the entry block.
11145 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11146 Reset the SCEV hashtable and niters.
11147 * graphite-scop-detection.c
11148 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11149 assert that we only have POLYNOMIAL_CHREC that vary in loops
11150 contained in the region.
11151 (scop_detection::graphite_can_represent_expr): Adjust.
11152 (scop_detection::stmt_has_simple_data_refs_p): For loops
11153 not in the region set loop to NULL. The nest is now the
11154 entry edge to the region.
11155 (try_generate_gimple_bb): Likewise.
11156 * sese.c (scalar_evolution_in_region): Adjust for
11157 instantiate_scev change.
11158 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11159 Make nest parameter the edge into the region.
11160 (create_data_ref): Likewise.
11161 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11162 entry edge into a region and adjust instantiate_scev calls.
11163 (create_data_ref): Likewise.
11164 (graphite_find_data_references_in_stmt): Likewise.
11165 (find_data_references_in_stmt): Pass the loop preheader edge
11166 from the nest argument.
11167 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11168 parameter the edge into the region.
11169 (instantiate_parameters): Use the loop preheader edge as entry.
11170 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11172 (get_instantiated_value_entry): Make instantiate_below parameter
11173 the edge into the region.
11174 (instantiate_scev_name): Likewise. Adjust dominance checks,
11175 when we cannot use loop-based instantiation instantiate by
11176 walking use-def chains.
11177 (instantiate_scev_poly): Adjust.
11178 (instantiate_scev_binary): Likewise.
11179 (instantiate_scev_convert): Likewise.
11180 (instantiate_scev_not): Likewise.
11181 (instantiate_array_ref): Remove.
11182 (instantiate_scev_3): Likewise.
11183 (instantiate_scev_2): Likewise.
11184 (instantiate_scev_1): Likewise.
11185 (instantiate_scev_r): Do not blindly handle N-operand trees.
11186 Do not instantiate array-refs. Handle all constants and invariants.
11187 (instantiate_scev): Make instantiate_below parameter
11188 the edge into the region.
11189 (resolve_mixers): Use the loop preheader edge for the region
11190 parameter to instantiate_scev_r.
11191 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11193 2017-10-13 Richard Biener <rguenther@suse.de>
11195 PR tree-optimization/82525
11196 * graphite-isl-ast-to-gimple.c
11197 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11199 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11200 Fail code generation when we cannot represent the isl integer.
11201 (binary_op_to_tree): Elide modulo operations that are no-ops
11202 in the type we code generate. Remove now superfluous code
11205 2017-10-13 Richard Biener <rguenther@suse.de>
11207 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11208 (scop_detection::harmful_loop_in_region): Remove premature
11209 IV type restriction.
11210 (scop_detection::graphite_can_represent_scev): We can handle
11211 pointer IVs just fine.
11213 2017-10-13 Alan Modra <amodra@gmail.com>
11215 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11216 "Clobbers and Scratch Registers". Add paragraph on
11217 alternative to clobbers for scratch registers and OpenBLAS
11220 2017-10-13 Alan Modra <amodra@gmail.com>
11222 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11223 example of a memory input for a string of known length. Move
11224 commentary out of table. Add a number of new examples
11225 covering array memory inputs.
11227 2017-10-12 Martin Liska <mliska@suse.cz>
11229 PR tree-optimization/82493
11230 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11231 (test_range_functions): New function.
11232 (sbitmap_c_tests): Likewise.
11233 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11234 * selftest.h (sbitmap_c_tests): New function.
11236 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11238 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11240 * config/rs6000/amo.h: Fix spacing issue.
11242 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11245 * config/i386/i386.md (*ashl<mode>3_mask_1,
11246 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11247 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11250 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11252 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11253 (profile_probability): Set max_probability
11254 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11256 * profile-count.c (profile_probability::differs_from_p): Do not
11257 rely on max_probaiblity == 10000
11259 2017-10-12 Jeff Law <law@redhat.com>
11261 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11264 2017-10-12 Martin Sebor <msebor@redhat.com>
11268 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11269 (handle_alias_pairs): Call it.
11270 * common.opt (-Wattribute-alias): New option.
11271 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11272 * doc/invoke.texi (-Wattribute-alias): Document.
11274 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11277 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11279 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11281 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11282 (make_hard_regno_born, make_hard_regno_dead): Update
11283 bb_killed_pseudos and bb_gen_pseudos.
11285 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11287 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11289 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11291 * config/alpha/alpha.c (alpha_split_conditional_move):
11292 Use std::swap instead of manually swapping.
11293 (alpha_stdarg_optimize_hook): Ditto.
11294 (alpha_canonicalize_comparison): Ditto.
11296 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11298 * tree-loop-distribution.c (struct builtin_info): New struct.
11299 (struct partition): Refactor fields into struct builtin_info.
11300 (partition_free): Free struct builtin_info.
11301 (build_size_arg_loc, build_addr_arg_loc): Delete.
11302 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11303 information from struct builtin_info.
11304 (find_single_drs): New function refactored from classify_partition.
11305 Also moved builtin validity checks to this function.
11306 (compute_access_range, alloc_builtin): New functions.
11307 (classify_builtin_st, classify_builtin_ldst): New functions.
11308 (classify_partition): Refactor code into functions find_single_drs,
11309 classify_builtin_st and classify_builtin_ldst.
11310 (distribute_loop): Don't do runtime alias check when distributing
11312 (find_seed_stmts_for_distribution): New function.
11313 (pass_loop_distribution::execute): Refactor code finding seed
11314 stmts into above function. Support distribution for the innermost
11315 two-level loop nest. Adjust dump information.
11317 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11319 * tree-loop-distribution.c: Adjust the general comment.
11320 (NUM_PARTITION_THRESHOLD): New macro.
11321 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11322 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11323 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11324 in call to build_partition_graph.
11325 (finalize_partitions): New parameter. Make loop distribution more
11326 conservative by fusing more partitions.
11327 (distribute_loop): Don't do runtime alias check in case of loop nest
11329 (find_seed_stmts_for_distribution): New function.
11330 (prepare_perfect_loop_nest): New function.
11331 (pass_loop_distribution::execute): Refactor code finding seed stmts
11332 and loop nest into above functions. Support loop nest distribution.
11333 Adjust dump information accordingly.
11335 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11337 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11338 and set PTYPE_SEQUENTIAL for merged partition.
11340 2017-10-12 Richard Biener <rguenther@suse.de>
11342 PR tree-optimization/69728
11344 2017-09-19 Richard Biener <rguenther@suse.de>
11346 PR tree-optimization/69728
11347 * graphite-sese-to-poly.c (schedule_error): New global.
11348 (add_loop_schedule): Handle empty domain by failing the
11350 (build_original_schedule): Handle schedule_error.
11352 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11353 domain by returning an unchanged schedule.
11355 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11357 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11358 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11360 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11362 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11365 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11368 * expr.c (store_field): Don't optimize away bitsize == 0 store
11369 from CALL_EXPR with addressable return type.
11371 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11373 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11374 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11375 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11376 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11377 TARGET_ISEL instead of TARGET_ISEL<sel>.
11379 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11381 * config/rs6000/rs6000.c
11382 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11384 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11386 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11388 (reg_or_cint_operand): Fix comment.
11389 (reg_or_zero_operand): New predicate.
11390 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11391 * config/rs6000/rs6000.c (output_isel): Delete.
11392 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11393 instead of reg_or_cint_operand. Output instruction directly (not via
11395 (isel_unsigned_<mode>): Ditto.
11396 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11397 gpc_reg_operand. Add an instruction alternative for this. Output
11398 instruction directly.
11399 (*isel_reversed_unsigned_<mode>): Ditto.
11401 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11403 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11404 (TARGET_CANONICALIZE_COMPARISON): Define.
11406 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11409 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11410 Check whether the dest is REG before adding REG_EQUIV note.
11412 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11415 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11417 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11418 (make_hard_regno_born, make_hard_regno_dead): Update
11419 bb_killed_pseudos and bb_gen_pseudos.
11421 2017-10-11 Nathan Sidwell <nathan@acm.org>
11423 * incpath.h (enum incpath_kind): Name enum, prefix values.
11424 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11425 * incpath.c (heads, tails): Use INC_MAX.
11426 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11427 (merge_include_chains, split_quote_chain,
11428 register_include_chains): Update incpath_kind names.
11429 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11430 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11432 (add_framework_path, darwin_register_objc_includes): Likewise.
11433 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11435 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11437 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11438 Do not use float_operator operator predicate.
11439 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11440 * config/i386/predicates.md (float_operator): Remove predicate.
11442 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11444 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11445 (*jccxf_i387): Ditto.
11446 (*jcc<mode>_i387): Ditto.
11447 (*jccu<mode>_i387): Ditto.
11448 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11449 (*jcc_*_i387 splitters): Remove.
11450 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11451 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11452 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11455 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11457 * profile-count.h (slow_safe_scale_64bit): New function.
11458 (safe_scale_64bit): New inline.
11459 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11460 * profile-count.c: Include wide-int.h
11461 (slow_safe_scale_64bit): New.
11463 2017-10-11 Nathan Sidwell <nathan@acm.org>
11465 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11466 HAS_DECL_ASSEMBLER_NAME_P.
11467 * gimple-expr.c (gimple_decl_printable_name: Check
11468 HAS_DECL_ASSEMBLER_NAME_P too.
11469 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11470 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11471 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11472 * passes.c (rest_of_decl_compilation): Check
11473 HAS_DECL_ASSEMBLER_NAME_P too.
11474 * recog.c (verify_changes): Likewise.
11475 * tree-pretty-print.c (dump_decl_name): Likewise.
11476 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11478 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11479 (SET_DECL_ASSEMBLER_NAME): Use it.
11480 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11481 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11482 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11484 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11486 * config.gcc (i386, x86_64): Add extra objects.
11487 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11488 (ix86_min_insn_size): Declare.
11489 (ix86_issue_rate): Declare.
11490 (ix86_adjust_cost): Declare.
11491 (ia32_multipass_dfa_lookahead): Declare.
11492 (ix86_macro_fusion_p): Declare.
11493 (ix86_macro_fusion_pair_p): Declare.
11494 (ix86_bd_has_dispatch): Declare.
11495 (ix86_bd_do_dispatch): Declare.
11496 (ix86_core2i7_init_hooks): Declare.
11497 (ix86_atom_sched_reorder): Declare.
11498 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11499 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11500 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11501 (rip_relative_addr_p): Rename to ...
11502 (ix86_rip_relative_addr_p): ... this one; export.
11503 (memory_address_length): Update.
11504 (ix86_issue_rate): Move to x86-tune-sched.c.
11505 (ix86_flags_dependent): Move to x86-tune-sched.c.
11506 (ix86_agi_dependent): Move to x86-tune-sched.c.
11507 (exact_dependency_1): Move to x86-tune-sched.c.
11508 (exact_store_load_dependency): Move to x86-tune-sched.c.
11509 (ix86_adjust_cost): Move to x86-tune-sched.c.
11510 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11511 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11512 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11513 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11514 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11515 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11516 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11517 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11518 (min_insn_size): Rename to ...
11519 (ix86_min_insn_size): ... this one; export.
11520 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11521 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11522 (core2i7_first_cycle_multipass_backtrack): Move to
11523 x86-tune-sched-core.c.
11524 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11525 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11526 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11527 (ix86_avoid_jump_mispredicts): Update.
11528 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11529 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11530 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11531 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11532 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11533 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11534 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11535 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11536 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11537 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11538 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11539 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11540 (BIG): Move to ix86-tune-sched-bd.c.
11541 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11542 (enum insn_path): Move to ix86-tune-sched-bd.c.
11543 (get_mem_group): Move to ix86-tune-sched-bd.c.
11544 (is_cmp): Move to ix86-tune-sched-bd.c.
11545 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11546 (is_branch): Move to ix86-tune-sched-bd.c.
11547 (is_prefetch): Move to ix86-tune-sched-bd.c.
11548 (init_window): Move to ix86-tune-sched-bd.c.
11549 (allocate_window): Move to ix86-tune-sched-bd.c.
11550 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11551 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11552 (process_end_window): Move to ix86-tune-sched-bd.c.
11553 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11554 (find_constant): Move to ix86-tune-sched-bd.c.
11555 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11556 (has_immediate): Move to ix86-tune-sched-bd.c.
11557 (get_insn_path): Move to ix86-tune-sched-bd.c.
11558 (get_insn_group): Move to ix86-tune-sched-bd.c.
11559 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11560 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11561 (add_insn_window): Move to ix86-tune-sched-bd.c.
11562 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11563 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11564 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11565 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11566 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11567 (do_dispatch): Move to ix86-tune-sched-bd.c.
11568 (has_dispatch): Move to ix86-tune-sched-bd.c.
11569 * config/i386/t-i386: Add new object files.
11570 * config/i386/x86-tune-costs.h: New file.
11571 * config/i386/x86-tune-sched-atom.c: New file.
11572 * config/i386/x86-tune-sched-bd.c: New file.
11573 * config/i386/x86-tune-sched-core.c: New file.
11574 * config/i386/x86-tune-sched.c: New file.
11576 2017-10-11 Liu Hao <lh_mouse@126.com>
11578 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11579 the generic version below instead.
11580 (should_colorize): Recognize Windows consoles as terminals
11582 * pretty-print.c [__MINGW32__] (write_all): New function.
11583 [__MINGW32__] (find_esc_head): Likewise.
11584 [__MINGW32__] (find_esc_terminator): Likewise.
11585 [__MINGW32__] (eat_esc_sequence): Likewise.
11586 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11588 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11591 2017-10-11 Richard Biener <rguenther@suse.de>
11593 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11594 Properly call analyze_scalar_evolution with the loop of the stmt.
11596 2017-10-11 Richard Biener <rguenther@suse.de>
11598 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11599 * tree-core.h (tree_base): Add chrec_var union member.
11600 * tree.h (CHREC_VAR): Remove.
11601 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11602 * tree-chrec.h (build_polynomial_chrec): Adjust.
11603 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11604 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11606 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11608 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11609 * match.pd: ... here.
11610 ((T) X == (T) Y): Relax condition.
11612 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11614 PR tree-optimization/82472
11615 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11617 (break_alias_scc_partitions): Update postorder number.
11619 2017-10-11 Martin Liska <mliska@suse.cz>
11622 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11624 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11626 2017-10-11 Martin Liska <mliska@suse.cz>
11628 * print-rtl.c (print_insn): Move declaration of idbuf
11629 to same scope as name.
11631 2017-10-11 Martin Liska <mliska@suse.cz>
11636 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11637 volatile arguments.
11639 2017-10-11 Martin Liska <mliska@suse.cz>
11642 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11643 volatile arguments.
11645 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11647 * config.gcc (default_gnu_indirect_function): Default to yes for
11648 arm*-*-linux* with glibc.
11650 2017-10-11 Richard Biener <rguenther@suse.de>
11652 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11653 default-defs and types we do not want to analyze.
11654 (interpret_loop_phi): Replace unreachable code with an assert.
11655 (compute_scalar_evolution_in_loop): Remove and inline ...
11656 (analyze_scalar_evolution_1): ... here, replacing condition with
11657 what makes the intent clearer. Remove handling of cases
11658 get_scalar_evolution now handles.
11660 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11662 PR rtl-optimization/81434
11663 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11664 comment for main loop. In sched_group_found if, also add checks for
11665 pass and min_cost_group.
11667 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11669 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11670 (rs6000_insn_cost): New function.
11671 * config/rs6000/rs6000.md (cost): New attribute.
11673 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11674 H.J. Lu <hongjiu.lu@intel.com>
11678 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11679 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11680 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11681 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11682 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11683 OPTION_MASK_ISA_MMX and some other bit set require both
11684 mmx and the ISAs without the mmx bit.
11685 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11686 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11687 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11688 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11689 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11690 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11691 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11692 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11693 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11694 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11695 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11696 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11697 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11698 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11699 Add OPTION_MASK_ISA_MMX.
11701 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11703 * config.gcc (armv7*-*-freebsd*): New target.
11704 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11706 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11708 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11709 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11712 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11714 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11715 a template parameter.
11716 (WIDE_INT_REF_FOR): Update accordingly.
11717 * tree.h (wi::int_traits <const_tree>): Delete.
11718 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11719 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11720 (wi::tree_to_wide_ref): New typedef.
11721 (wi::to_wide): New function.
11722 * calls.c (get_size_range): Use wi::to_wide when operating on
11723 trees as wide_ints.
11724 * cgraph.c (cgraph_node::create_thunk): Likewise.
11725 * config/i386/i386.c (ix86_data_alignment): Likewise.
11726 (ix86_local_alignment): Likewise.
11727 * dbxout.c (stabstr_O): Likewise.
11728 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11729 * expr.c (const_vector_from_tree): Likewise.
11730 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11731 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11732 (fold_negate_expr_1, int_const_binop_1, const_binop)
11733 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11734 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11735 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11736 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11737 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11738 (fold_not_const, round_up_loc): Likewise.
11739 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11740 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11741 (alloca_call_type): Likewise.
11742 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11743 * godump.c (go_output_typedef): Likewise.
11744 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11745 * internal-fn.c (get_min_precision): Likewise.
11746 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11747 * ipa-polymorphic-call.c
11748 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11749 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11750 (ipa_modify_call_arguments): Likewise.
11751 * match.pd: Likewise.
11752 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11753 * print-tree.c (print_node_brief, print_node): Likewise.
11754 * stmt.c (expand_case): Likewise.
11755 * stor-layout.c (layout_type): Likewise.
11756 * tree-affine.c (tree_to_aff_combination): Likewise.
11757 * tree-cfg.c (group_case_labels_stmt): Likewise.
11758 * tree-data-ref.c (dr_analyze_indices): Likewise.
11759 (prune_runtime_alias_test_list): Likewise.
11760 * tree-dump.c (dequeue_and_dump): Likewise.
11761 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11762 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11763 * tree-pretty-print.c (dump_generic_node): Likewise.
11764 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11765 (simple_iv_with_niters): Likewise.
11766 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11767 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11768 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11769 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11770 (refine_value_range_using_guard, number_of_iterations_ne_max)
11771 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11772 (get_cst_init_from_scev, record_nonwrapping_iv)
11773 (scev_var_range_cant_overflow): Likewise.
11774 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11775 * tree-ssa-pre.c (compute_avail): Likewise.
11776 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11777 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11778 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11779 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11780 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11781 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11782 (dump_case_nodes, try_switch_expansion): Likewise.
11783 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11784 (vect_do_peeling): Likewise.
11785 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11786 * tree-vect-stmts.c (vectorizable_load): Likewise.
11787 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11788 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11789 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11790 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11791 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11792 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11793 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
11794 (evrp_dom_walker::before_dom_children): Likewise.
11795 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
11796 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
11797 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
11798 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
11799 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
11800 (get_range_pos_neg): Likewise.
11801 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
11802 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
11803 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
11804 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
11805 * config/avr/avr.c (avr_fold_builtin): Likewise.
11806 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
11807 * config/msp430/msp430.c (msp430_attr): Likewise.
11808 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
11809 * config/powerpcspe/powerpcspe-c.c
11810 (altivec_resolve_overloaded_builtin): Likewise.
11811 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
11812 (rs6000_expand_ternop_builtin): Likewise.
11813 * config/rs6000/rs6000-c.c
11814 (altivec_resolve_overloaded_builtin): Likewise.
11815 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
11816 (rs6000_expand_ternop_builtin): Likewise.
11817 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
11819 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11821 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
11822 when copying loop nest with only one inner loop.
11824 2017-10-10 Richard Biener <rguenther@suse.de>
11826 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
11827 blocks if SCEV is active.
11828 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
11830 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
11831 (scev_initialize): Assert we are not yet initialized.
11833 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11835 * tree-loop-distribution.c (generate_loops_for_partition): Remove
11836 inner loop's exit stmt by making it always exit the loop, otherwise
11837 we would generate an infinite empty loop.
11839 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11841 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
11842 renaming variables in new preheader if it's deleted.
11844 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11846 * tree-loop-distribution.c (struct partition): Remove unused field
11847 loops of the structure.
11848 (partition_alloc, partition_free): Ditto.
11849 (build_rdg_partition_for_vertex): Ditto.
11851 2017-10-09 Jeff Law <law@redhat.com>
11853 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
11854 return type to match prototype and documentation.
11856 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11858 * config/rs6000/rs6000.c (processor_costs): Move to ...
11859 * config/rs6000/rs6000.h: ... here.
11860 (rs6000_cost): Declare.
11862 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
11864 * except.c (setjmp_fn): New global variable.
11865 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
11866 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
11867 if DONT_USE_BUILTIN_SETJMP is defined.
11869 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11871 * target.def (insn_cost): New hook.
11872 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
11873 * doc/tm.texi: Regenerate.
11874 * rtlanal.c (insn_cost): Use the new hook.
11876 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11878 * combine.c (combine_validate_cost): Compute the new insn_cost,
11879 not just pattern_cost.
11880 (try_combine): Adjust comment.
11882 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11884 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
11886 * combine.c (uid_insn_cost): Adjust comment.
11887 (combine_validate_cost): Adjust comment. Use pattern_cost instead
11889 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
11890 * dse.c (find_shift_sequence): Ditto.
11891 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
11892 (bb_valid_for_noce_process_p): Use pattern_cost.
11893 * rtl.h (insn_rtx_cost): Delete.
11894 (pattern_cost): New prototype.
11895 (insn_cost): New prototype.
11896 * rtlanal.c (insn_rtx_cost): Rename to...
11897 (pattern_cost): ... this.
11900 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
11902 * config/i386/i386.md (*jcc_2): Remove insn pattern.
11903 (*jcc<mode>_0_r_i387): Ditto.
11904 (*jccxf_r_i387): Ditto.
11905 (*jcc<mode>_r_i387): Ditto.
11906 (*jccu<mode>_r_i387): Ditto.
11907 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
11908 (*jcc): Rename from *jcc_1.
11910 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11912 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
11913 deferred rescans after the lvx/stvx recombination pre-pass.
11915 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
11917 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
11918 memory operation instruction support.
11919 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
11920 (rs6000-ibm-aix[789]*): Likewise.
11921 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
11922 Document new functions.
11924 2017-10-09 Richard Biener <rguenther@suse.de>
11926 PR tree-optimization/82397
11927 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
11928 equality only for semantically equal trees.
11930 2017-10-09 Richard Biener <rguenther@suse.de>
11932 PR tree-optimization/82449
11933 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
11934 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
11935 allow constant addresses.
11936 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
11939 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11941 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
11944 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11947 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
11950 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11953 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
11955 2017-10-09 Jakub Jelinek <jakub@redhat.com>
11958 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
11959 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
11961 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
11963 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
11964 (WI_BINARY_PREDICATE_RESULT): Likewise.
11965 (wi::binary_traits::operator_result): New type.
11966 (wi::binary_traits::predicate_result): Likewise.
11967 (generic_wide_int::operator~, unary generic_wide_int::operator-)
11968 (generic_wide_int::operator==, generic_wide_int::operator!=)
11969 (generic_wide_int::operator&, generic_wide_int::and_not)
11970 (generic_wide_int::operator|, generic_wide_int::or_not)
11971 (generic_wide_int::operator^, generic_wide_int::operator+
11972 (binary generic_wide_int::operator-, generic_wide_int::operator*):
11974 (operator~, unary operator-, operator==, operator!=, operator&)
11975 (operator|, operator^, operator+, binary operator-, operator*): New
11977 * expr.c (get_inner_reference): Use wi::bit_and_not.
11978 * fold-const.c (fold_binary_loc): Likewise.
11979 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
11980 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
11981 (bit_value_binop): Likewise.
11982 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
11983 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
11984 (extract_range_from_binary_expr_1): Likewise.
11985 (masked_increment): Likewise.
11986 (simplify_bit_ops_using_ranges): Likewise.
11988 2017-10-09 Martin Jambor <mjambor@suse.cz>
11991 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
11992 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
11993 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
11995 (hsa_fixup_mov_insn_type): New function.
11996 (hsa_op_with_type::get_in_type): Use it.
11997 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
11998 immediates in an assert.
11999 (hsa_op_with_type::extend_int_to_32bit): New method.
12000 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12001 types. Convert to dest type if necessary.
12002 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12003 (reg_for_gimple_ssa): Pass false as min32int to
12004 hsa_type_for_scalar_tree_type.
12005 (gen_hsa_addr): Fixup type when creating addresable temporary.
12006 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12007 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12008 necessary. Call hsa_fixup_mov_insn_type.
12009 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12010 extend operands and convert to dest type if necessary.
12011 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12012 to dest type if necessary.
12013 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12014 if conversion nt necessary and size matches.
12015 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12016 to dest type if necessary.
12017 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12018 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12020 (gen_hsa_clrsb): Likewise.
12021 (gen_hsa_ffs): Likewise.
12022 (gen_hsa_divmod): Extend operands and convert to dest type if
12024 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12026 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12028 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12029 Remove empty default arguments. Use a brace block as output
12031 (conditional return): Ditto.
12033 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12034 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12036 (group_ending_nop): Ditto.
12037 (doloop_end): Ditto.
12038 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12039 (splitters for those): Ditto.
12041 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12043 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12044 a conditional jump (and the compare for it) so that pc_rtx is the
12046 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12047 for the deleted and renamed ctr<mode>_internal[234] patterns.
12048 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12049 Delete second conditional return pattern.
12050 (ctr<mode>_internal2): Delete this second bdnz pattern.
12051 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12052 (ctr<mode>_internal4): Delete this second bdz pattern.
12054 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12056 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12057 (always_initialized_rtx_for_ssa_name_p): New predicate.
12058 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12059 (finish_out_of_ssa): Free new field of SA.
12060 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12061 * tree-ssa-coalesce.c: Include tree-ssa.h.
12062 (get_parm_default_def_partitions): Remove extern keyword.
12063 (get_undefined_value_partitions): New function.
12064 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12065 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12066 uninitialized bits.
12067 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12069 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12071 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12073 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12075 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12076 for targets that preffer 128bit.
12078 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12080 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12082 2017-10-08 Olivier Hainque <hainque@adacore.com>
12084 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12085 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12087 (thumb_set_return_address): Likewise.
12089 2017-10-08 Olivier Hainque <hainque@adacore.com>
12091 * common/config/arm/arm-common.c (arm_except_unwind_info):
12092 Handle DWARF2_UNWIND_INFO.
12094 2017-10-07 Michael Collison <michael.collison@arm.com>
12096 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12099 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12101 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12102 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12103 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12104 defined, force the creation of a new block for a dispatch label.
12106 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12108 * invoke.texi (Wsuggest-attribute=cold): Document.
12109 * common.opt (Wsuggest-attribute=cold): New
12110 * ipa-pure-const.c (warn_function_cold): New function.
12111 * predict.c (compute_function_frequency): Use it.
12112 * predict.h (warn_function_cold): Declare.
12114 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12116 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12119 2017-10-06 Martin Liska <mliska@suse.cz>
12121 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12122 keyword for member functions.
12123 (struct sanopt_tree_couple): New struct.
12124 (struct sanopt_tree_couple_hash): New function.
12125 (struct sanopt_ctx): Add new hash_map.
12126 (has_dominating_ubsan_ptr_check): New function.
12127 (record_ubsan_ptr_check_stmt): Likewise.
12128 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12129 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12130 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12132 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12135 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12136 aarch64_simd_valid_immediate on CONST_VECTORs.
12137 (aarch64_reg_or_bic_imm): Likewise.
12139 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12141 PR rtl-optimization/82396
12142 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12144 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12146 * graphite-dependences.c (scop_get_reads): Move code to...
12147 (scop_get_must_writes): Move code to...
12148 (scop_get_may_writes): Move code to...
12149 (scop_get_reads_and_writes): ... here.
12150 (scop_get_dependences): Call scop_get_reads_and_writes.
12152 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12154 PR tree-optimization/82434
12155 * fold-const.h (can_native_encode_type_p,
12156 can_native_encode_string_p): Remove.
12157 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12158 don't encode anything, just return what would be otherwise returned.
12159 (native_encode_fixed, native_encode_complex, native_encode_vector):
12161 (native_encode_string): Likewise. Inline by hand
12162 can_native_encode_string_p.
12163 (can_native_encode_type_p): Remove.
12164 (can_native_encode_string_p): Remove.
12165 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12166 STRING_CSTs using can_native_encode_string_p, test all
12167 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12168 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12169 argument from native_encode_expr.
12170 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12171 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12174 2017-10-06 Richard Biener <rguenther@suse.de>
12176 PR tree-optimization/82397
12177 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12178 operand_equal_p but rely on data_ref_compare_tree for detecting
12180 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12181 to match up with dr_group_sort_cmp.
12183 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12186 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12188 * config/s390/s390-builtin-types.def: Regenerate.
12190 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12193 * config/s390/s390-builtin-types.def: Regenerate.
12194 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12195 Change flag from B_VXE to B_VX.
12196 (s390_vec_min_dbl): Remove B_VXE flag.
12198 2017-10-06 Richard Biener <rguenther@suse.de>
12200 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12201 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12202 translate_isl_ast_to_gimple::is_valid_rename,
12203 translate_isl_ast_to_gimple::get_rename,
12204 translate_isl_ast_to_gimple::get_def_bb_for_const,
12205 translate_isl_ast_to_gimple::get_new_name,
12206 translate_isl_ast_to_gimple::collect_all_ssa_names,
12207 translate_isl_ast_to_gimple::copy_loop_phi_args,
12208 translate_isl_ast_to_gimple::collect_all_ssa_names,
12209 translate_isl_ast_to_gimple::copy_loop_phi_args,
12210 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12211 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12212 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12213 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12214 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12215 translate_isl_ast_to_gimple::copy_cond_phi_args,
12216 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12217 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12218 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12219 translate_isl_ast_to_gimple::rename_uses,
12220 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12221 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12222 (set_rename_for_each_def): Likewise.
12223 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12224 here. Handle rewriting SCEV analyzable uses here.
12225 (copy_bb_and_scalar_dependences): Generate code for PHI
12227 (graphite_regenerate_ast_isl): Adjust.
12228 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12229 (add_write, add_read): New functions.
12230 (build_cross_bb_scalars_def): Use it and simplify.
12231 (build_cross_bb_scalars_use): Likewise.
12232 (graphite_find_cross_bb_scalar_vars): Inline into...
12233 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12234 simulating out-of-SSA. Compute liveout and add dependencies.
12235 (build_scops): Force an empty entry block.
12236 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12238 (sese_build_liveouts): Declare.
12239 (sese_trivially_empty_bb_p): Likewise.
12240 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12241 compute liveout and debug_liveout.
12242 (sese_bad_liveouts_use): Remove.
12243 (sese_reset_debug_liveouts_bb): Likewise.
12244 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12245 (sese_build_liveouts): Build liveout and debug_liveout and store
12247 (new_sese_info): Adjust.
12248 (free_sese_info): Likewise.
12249 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12250 do not build liveout here.
12251 (move_sese_in_condition): Adjust region entry.
12252 (scev_analyzable_p): Match up with chrec_apply requirements.
12253 (sese_trivially_empty_bb_p): New.
12254 * tree-into-ssa.c (get_reaching_def): Properly support generating
12255 default-defs for incremental rewrite of anonymous names.
12257 2017-10-06 Richard Biener <rguenther@suse.de>
12259 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12260 precision do not perform modulo reduction.
12262 2017-10-06 Richard Biener <rguenther@suse.de>
12264 PR tree-optimization/82436
12265 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12266 conservatively choose the vectorization factor when checking
12267 whether we can perform the required load permutation.
12268 (vect_transform_slp_perm_load): Assert when we may not fail.
12270 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12272 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12273 message for incompatible -msdata=* and -mcall-* options.
12275 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12277 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12278 rate for post-reload scheduling.
12280 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12282 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12284 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12286 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12287 to improve monte carlo in scimark.
12289 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12291 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12292 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12293 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12294 pentium4_cost, nocona_cost): Set reassociation width to 1.
12295 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12296 width to 2 for fp operations and 1 otherwise.
12297 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12299 (atom_cost): Set reassociation width to 2.
12300 (slm_cost, generic_cost): Set fp reassociation width
12301 to 2 and 1 otherwise.
12302 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12303 (core_cost): Set fp reassociation width to 4 and vector to 2.
12304 (ix86_reassociation_width): Rewrite using cost table; special case
12305 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12306 and TARGET_AVX128_OPTIMAL.
12307 * config/i386/i386.h (processor_costs): Add
12308 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12309 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12310 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12311 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12312 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12313 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12315 2017-10-05 Nathan Sidwell <nathan@acm.org>
12317 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12319 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12321 * config/arm/arm.c (arm_test_fpu_data): New.
12322 (arm_run_selftests): Call arm_test_fpu_data.
12324 2017-10-04 Nathan Sidwell <nathan@acm.org>
12326 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12328 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12330 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12332 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12333 check type for aarch64_simd_valid_immediate.
12334 (aarch64_output_simd_mov_immediate): Update prototype.
12335 (aarch64_simd_valid_immediate): Update prototype.
12336 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12337 support for ORR-immediate.
12338 (and<mode>3): modified pattern to add support for BIC-immediate.
12339 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12340 now checks for valid immediate for BIC and ORR based on new enum
12342 (aarch64_output_simd_mov_immediate): Function now used to output
12343 BIC/ORR imm as well based on new enum argument.
12344 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12346 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12347 (aarch64_reg_or_bic_imm): Likewise.
12349 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12351 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12352 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12354 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12358 PR rtl-optimization/82396
12359 * haifa-sched.c (autopref_multipass_init): Simplify
12361 (autopref_rank_data): Simplify sort order.
12362 * sched-int.h (autopref_multipass_data_): Remove
12363 multi_mem_insn_p, min_offset and max_offset.
12365 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12367 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12369 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12371 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12372 vect_doubleint_cvt.
12374 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12376 * doc/sourcebuild.texi: Document vect_long_mult.
12378 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12380 PR tree-optimization/82413
12381 * fold-const.c (build_range_check): Use widest_int when comparing
12382 the maximum ETYPE value with HIGH.
12384 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12386 PR rtl-optimization/82396
12387 * haifa-sched.c (autopref_multipass_init): Simplify
12389 (autopref_rank_data): Simplify sort order.
12390 * sched-int.h (autopref_multipass_data_): Remove
12391 multi_mem_insn_p, min_offset and max_offset.
12393 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12395 PR tree-optimization/82381
12396 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12397 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12400 PR tree-optimization/82374
12401 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12402 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12403 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12404 current_function_decl to the new decl.
12406 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12408 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12409 helper macro for IEEE float128 hardware built-in functions.
12410 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12412 (TRUNCF128_ODD): Likewise.
12413 (ADDF128_ODD): Likewise.
12414 (SUBF128_ODD): Likewise.
12415 (MULF128_ODD): Likewise.
12416 (DIVF128_ODD): Likewise.
12417 (FMAF128_ODD): Likewise.
12418 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12419 UNSPEC_TRUNC_ROUND_TO_ODD.
12420 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12421 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12422 floating point round to odd instructions.
12423 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12424 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12425 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12426 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12427 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12428 (trunc<mode>sf2_hw): Change the truncate with round to odd
12429 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12430 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12431 to odd hardware instructions.
12432 (sub<mode>3_odd): Likewise.
12433 (mul<mode>3_odd): Likewise.
12434 (div<mode>3_odd): Likewise.
12435 (sqrt<mode>2_odd): Likewise.
12436 (fma<mode>4_odd): Likewise.
12437 (fms<mode>4_odd): Likewise.
12438 (nfma<mode>4_odd): Likewise.
12439 (nfms<mode>4_odd): Likewise.
12440 (trunc<mode>df2_odd): Change the truncate with round to odd
12441 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12443 * doc/extend.texi (PowerPC built-in functions): Update documentation
12444 for existing IEEE float128-bit built-in functions. Add built-in
12445 functions that generate the IEEE 128-bit floating point round to
12448 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12450 PR rtl-optimization/77729
12451 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12452 to (X&(C1&~C2))|C2 transformations.
12454 2017-10-03 Martin Jambor <mjambor@suse.cz>
12456 PR tree-optimization/82363
12457 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12458 mismatch, mark lacc written regardless of racc.
12460 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12462 PR tree-optimization/82381
12463 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12464 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12465 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12466 fallthrough into reassoc_stmt_dominates_stmt_p.
12469 * combine.c (combine_instructions): Don't combine in unreachable
12472 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12475 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12476 function to not use the have_cpu variable. Do not set cpu_index,
12477 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12478 or the default cpu.
12479 (rs6000_valid_attribute_p): Remove duplicate initializations of
12480 old_optimize and func_optimize.
12481 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12482 (rs6000_activate_target_options): Make global.
12483 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12486 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12488 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12489 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12490 Set *poffset to 0 and *psize and *pmax_size to -1 if
12491 *poffset + *psize overflows in HOST_WIDE_INT.
12493 PR tree-optimization/82387
12494 PR tree-optimization/82388
12495 PR tree-optimization/82389
12496 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12497 instead of live_bytes non-NULL.
12499 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12502 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12505 2017-10-02 Richard Biener <rguenther@suse.de>
12507 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12508 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12509 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12511 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12513 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12514 requested precision matches the type's.
12515 * calls.c (alloc_max_size): Calculate the new candidate size as
12516 a widest_int and use wi::to_widest when comparing it with the
12517 current candidate size.
12518 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12519 zero rather than integer_zero_node.
12520 * match.pd: Check for a no-op conversion before using wi::add
12521 rather than after. Use tree_to_uhwi when summing small shift
12522 counts into an unsigned int.
12524 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12525 Alan Hayward <alan.hayward@arm.com>
12526 David Sherwood <david.sherwood@arm.com>
12529 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12530 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12531 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12532 POINTER_AND_FP_REGS.
12534 2017-10-02 Richard Biener <rguenther@suse.de>
12536 PR tree-optimization/82355
12537 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12538 a mapping for the enclosing loop but avoid generating one for
12539 the loop tree root.
12540 (copy_bb_and_scalar_dependences): Remove premature codegen
12541 error on PHIs in blocks duplicated into multiple places.
12542 * graphite-scop-detection.c
12543 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12544 in the region use it as loop and nest to analyze the DR in.
12545 (try_generate_gimple_bb): Likewise.
12546 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12547 (add_loop_constraints): For blocks in a loop not in the region
12548 create a dimension with a single iteration.
12549 * sese.h (gbb_loop_at_index): Remove assert.
12551 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12553 * omp-expand.c (adjust_context_scope): New function.
12554 (expand_parallel_call): Call adjust_context_scope.
12556 2017-10-01 Jeff Law <law@redhat.com>
12558 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12559 dom_opt_dom_walker class with direct access to private members.
12560 Add comments. Call test_for_singularity.
12561 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12562 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12563 m_dummy_cond anymore.
12564 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12566 (pass_dominator:execute): Build the dummy_cond here and pass it
12567 to the dom_opt_dom_walker ctor.
12568 (test_for_singularity): New function.
12570 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12571 Maya Rashish <coypu@sdf.org>
12573 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12574 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12575 (alpha*-*-netbsd*) Use nbsd_tm_file.
12576 (arm*-*-netbsdelf*) Likewise.
12577 (i[34567]86-*-netbsdelf*) Likewise.
12578 (x86_64-*-netbsd*) Likewise.
12579 (mips*-*-netbsd*) Likewise.
12580 (powerpc-*-netbsd*) Likewise.
12581 (sh*-*-netbsd*) Likewise.
12582 (sparc-*-netbsdelf*) Likewise.
12583 (sparc64-*-netbsd*) Likewise.
12584 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12586 (vax-*-netbsdelf*) Likewise.
12587 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12588 (UINT_FAST8_TYPE) Likewise.
12589 (INT_FAST16_TYPE) Check CHAR_FAST16.
12590 (UINT_FAST16_TYPE) Likewise.
12592 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12595 * config/i386/i386.md
12596 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12597 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12598 *divmodsi4_zext_2): New define_insn_and_split.
12599 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12600 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12601 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12602 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12603 New define_insn_and_split.
12604 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12605 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12606 operands[1] having DImode when mode is SImode.
12608 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12609 always SImode for DIV and MOD in REG_EQUAL notes.
12611 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12613 PR middle-end/82319
12614 * match.pd: Fix handling of NaNs in pattern.
12616 2017-09-29 Jeff Law <law@redhat.com>
12618 * sbitmap.c (bitmap_bit_in_range_p): New function.
12619 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12620 * tree-ssa-dse.c (live_bytes_read): New function.
12621 (dse_classify_store): Ignore reads of dead bytes.
12623 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12624 typos and whitespace errors.
12625 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12626 * config/s390/s390.c (s390_emit_prologue): Likewise.
12628 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12631 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12632 with a symbol for LRA.
12634 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12636 PR rtl-optimization/82338
12637 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12639 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12641 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12642 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12643 (qsort_chk): Declare.
12644 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12645 (qsort_chk): New function.
12647 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12649 PR tree-optimization/82337
12650 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12651 phi definition if the PHI result appears in an abnormal PHI.
12652 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12653 basis appears in an abnormal PHI.
12655 2017-09-29 Richard Biener <rguenther@suse.de>
12657 * graphite-isl-ast-to-gimple.c
12658 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12659 (binary_op_to_tree): Use it.
12660 (get_rename_from_scev): Likewise.
12661 (copy_loop_phi_nodes): Likewise.
12662 (copy_bb_and_scalar_dependences): Likewise.
12663 (translate_pending_phi_nodes): Likewise.
12665 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12668 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12669 for movabsq $(i32 << shift), r64.
12671 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12673 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12674 index when encoding %esp as %rsp to avoid 0x67 prefix.
12676 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12678 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12679 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12681 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12683 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12684 Extensions with more than 16 double VFP registers.
12685 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12686 to_clear_mask and all code related to it. Replace the remaining
12687 entry by a sbitmap and adapt code accordingly.
12689 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12691 * brig-builtins.def: Change pure attributes to const.
12693 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12695 * config.gcc (default_gnu_indirect_function): Default to yes for
12696 sparc*-*-linux* with glibc.
12698 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12700 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12701 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12702 when creating .init_array and .fini_array sections with priority
12705 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12708 * config/aarch64/aarch64.c
12709 (aarch64_builtin_support_vector_misalignment): Always return false
12710 when misalignment is unknown.
12712 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12714 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12715 this function to return false if the definition used by the swap
12716 instruction is artificial, or if the memory address from which the
12717 constant value is loaded is not represented by a base address held
12718 in a register or if the base address register is a frame or stack
12719 pointer. Additionally, return false if the base address of the
12720 loaded constant is a SYMBOL_REF but is not considered to be a
12722 (replace_swapped_load_constant): New function.
12723 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12724 loaded constant vector with a load of a swapped constant vector.
12726 2017-09-27 Carl Love <cel@us.ibm.com>
12728 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12729 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12730 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12733 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12735 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12736 first, always call autopref_rank_data otherwise.
12738 2017-09-27 Richard Biener <rguenther@suse.de>
12740 * graphite-scop-detection.c (find_scop_parameters): Move
12741 loop bound handling ...
12742 (gather_bbs::before_dom_children): ... here, avoiding the need
12743 to build scop_info->loop_nest.
12744 (record_loop_in_sese): Remove.
12745 * sese.h (sese_info_t::loop_nest): Remove.
12746 * sese.c (new_sese_info): Do not allocate loop_nest.
12747 (free_sese_info): Do not free loop_nest.
12749 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12752 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12753 lhs from calls if the lhs has addressable type.
12755 2017-09-27 Richard Biener <rguenther@suse.de>
12757 * graphite.h (scop::max_alias_set): New member.
12758 * graphite-scop-detection.c: Remove references to non-existing
12759 --param in comments.
12760 (build_alias_sets): Record the maximum alias set used for drs.
12761 (build_scops): Support zero as unlimited for
12762 --param graphite-max-arrays-per-scop.
12763 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12764 and inline into ...
12765 (build_poly_sr_1): ... here. Compute alias set based on the
12766 maximum alias set used for drs rather than
12767 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12769 2017-09-27 Richard Biener <rguenther@suse.de>
12771 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12772 --param loop-block-tile-size=0 to disable tiling.
12774 2017-09-27 Richard Biener <rguenther@suse.de>
12776 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12777 (graphite-max-nb-scop-params): Document special value zero.
12778 * domwalk.h (dom_walker::STOP): New symbolical constant.
12779 (dom_walker::dom_walker): Add optional parameter for bb to
12781 (dom_walker::~dom_walker): Declare.
12782 (dom_walker::before_dom_children): Document STOP return value.
12783 (dom_walker::m_user_bb_to_rpo): New member.
12784 (dom_walker::m_bb_to_rpo): Likewise.
12785 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12786 mapping here if not provided by the user.
12787 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12788 provided by the user.
12789 (dom_walker::STOP): Define.
12790 (dom_walker::walk): Do not compute bb to RPO mapping here.
12791 Support STOP return value from before_dom_children to stop
12793 * graphite-optimize-isl.c (optimize_isl): If the schedule
12794 is the same still generate code if -fgraphite-identity
12795 or -floop-parallelize-all are given.
12796 * graphite-scop-detection.c: Include cfganal.h.
12797 (gather_bbs::gather_bbs): Get and pass through bb to RPO
12799 (gather_bbs::before_dom_children): Return STOP for BBs
12801 (build_scops): Compute bb to RPO mapping and pass it to
12802 the domwalk. Treat --param graphite-max-nb-scop-params=0
12803 as not limiting the number of params.
12804 * graphite.c (graphite_initialize): Remove limit on the
12805 number of basic-blocks in a function.
12806 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
12807 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
12808 default value of 10.
12810 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12812 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
12813 Adjust code to eliminate needing to do the shift right 32-bits
12814 operation after XSCVDPSPN.
12816 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12818 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
12819 ((X / Y) != 0 -> X >= Y): Likewise.
12821 2017-09-26 Carl Love <cel@us.ibm.com>
12823 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
12824 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
12825 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
12826 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
12827 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
12828 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
12829 definitions and overloading.
12830 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
12831 statement for P9V_BUILTIN_XST_LEN_R.
12832 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
12833 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
12834 define_expand and define_insn for the instructions and builtins.
12835 * doc/extend.texi: Update the built-in documentation file for the new
12836 built-in functions.
12837 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
12838 define_insn for the instructions
12840 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
12843 * gcc/config/netbsd-protos.h: New file.
12844 * gcc/config/netbsd.c: New file.
12845 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
12846 * gcc/config/t-netbsd: New file.
12847 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
12848 (tmake_file) Add t-netbsd.
12849 (extra_objs) Add netbsd.o.
12851 2017-09-26 Janus Weil <janus@gcc.gnu.org>
12855 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
12857 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12859 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
12860 sign extension from a vector register to a GPR by doing a 32-bit
12861 direct move and then an EXTSW.
12862 (extendsi<mode>2 splitter): Likewise.
12863 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
12864 right or vector extract after doing XSCVDPSPN. Use
12865 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
12867 (movdi_from_sf_zero_ext): Likewise.
12868 (reload_gpr_from_vsxsf): Likewise.
12869 (p8_mfvsrd_4_disf): Delete, no longer used.
12870 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
12871 then needing to move the SFmode to a GPR to use the XSCVDPSP
12872 instruction instead of FRSP and XSCVDPSPN.
12873 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
12874 it is adjacent to the other XSCVSPDP insns.
12875 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
12876 SFmode to be in traditional Altivec registers.
12877 (vsx_xscvdpspn): Eliminate useless alternative constraint.
12878 (vsx_xscvspdpn): Likewise.
12879 (vsx_xscvspdpn_scalar): Likewise.
12881 2017-09-26 Martin Jambor <mjambor@suse.cz>
12883 * tree-sra.c (compare_access_positions): Put integral types first,
12884 stabilize sorting of integral types, remove conditions putting
12885 non-full-precision integers last.
12886 (sort_and_splice_var_accesses): Disable scalarization if a
12887 non-integert would be represented by a non-full-precision integer.
12889 2017-09-26 Joseph Myers <joseph@codesourcery.com>
12891 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
12892 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
12893 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
12894 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
12895 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
12896 conditionals inside the function instead of around it. Call
12897 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
12898 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
12900 2017-09-26 Richard Biener <rguenther@suse.de>
12902 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
12904 (scop_detection::build_scop_breadth): ... this. Removed.
12905 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
12906 (scop_detection::harmful_stmt_in_bb): Likewise.
12907 (scop_detection::graphite_can_represent_stmt): Likewise.
12908 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
12909 (scop_detection::can_represent_loop): Remove recursion, fold in ...
12910 (scop_detection::can_represent_loop_1): ... this. Removed.
12911 (scop_detection::harmful_loop_in_region): Simplify after inlining
12912 the above and remove more quadraticness.
12913 (build_scops): Adjust.
12914 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
12917 2017-09-26 Jakub Jelinek <jakub@redhat.com>
12920 * config/i386/i386.c (ix86_print_operand_address_as): Only test
12921 REGNO (base) == SP_REG if base is a REG.
12923 PR middle-end/35691
12924 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
12925 if it is different SSA_NAME.
12926 (optimize_range_tests_cmp_bitwise): New function.
12927 (optimize_range_tests): Call it.
12929 2017-09-26 Richard Biener <rguenther@suse.de>
12931 PR tree-optimization/82321
12932 * graphite.c (canonicalize_loop_closed_ssa): Properly check
12933 for the def being inside the loop.
12935 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12937 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
12939 * config/s390/s390-builtins.def: Fix constraint on op4.
12941 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12943 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
12944 independent expanders.
12945 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
12946 ("vec_ordered", "vec_unordered"): New expanders.
12948 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12950 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
12953 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12955 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
12956 vec_unpacks_lo_v16qi.
12957 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
12959 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12961 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
12962 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
12963 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
12965 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12967 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
12969 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
12971 ("*vec_slb<mode>"): New insn pattern.
12972 ("vec_shr_<mode>"): New expander.
12973 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
12974 and force the shift count operand to V16QImode.
12975 ("vec_srb<mode>"): Set shift count mode to V16QI.
12977 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12979 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
12980 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
12981 ("vec_widen_smult_hi_<mode>"): New expander definitions.
12983 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
12986 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
12988 2017-09-26 Richard Biener <rguenther@suse.de>
12990 PR tree-optimization/82320
12991 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
12994 2017-09-25 Jeff Law <law@redhat.com>
12996 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
12997 prototype for new argument.
12998 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
12999 mostly extracted from rs6000_emit_allocate_stack.
13000 (rs6000_emit_probe_stack_range_stack_clash): New function.
13001 (rs6000_emit_allocate_stack): Call
13002 rs6000_emit_probe_stack_range_stack_clash as needed.
13003 (rs6000_emit_probe_stack_range): Add additional argument
13004 to call to gen_probe_stack_range{si,di}.
13005 (output_probe_stack_range): New.
13006 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13007 (output_probe_stack_range_stack_clash): New.
13008 (rs6000_emit_prologue): Emit notes into dump file as requested.
13009 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13010 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13011 Add additional operand and pass it to output_probe_stack_range.
13013 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13015 PR tree-optimization/82163
13016 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13017 (checking_verify_loop_closed_ssa): New parameter.
13018 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13019 (check_loop_closed_ssa_stmt): Delete.
13020 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13021 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13022 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13025 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13027 * brig-builtins.def: Treat HSAIL barrier builtins as
13028 setjmp/longjump style functions.
13030 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13032 * target.def (constant_alignment): New hook.
13033 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13034 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13035 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13036 * doc/tm.texi: Regenerate.
13037 * targhooks.h (default_constant_alignment): Declare.
13038 (constant_alignment_word_strings): Likewise.
13039 * targhooks.c (default_constant_alignment): New function.
13040 (constant_alignment_word_strings): Likewise.
13041 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13042 instead of CONSTANT_ALIGNMENT.
13043 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13044 (force_const_mem): Likewise.
13045 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13046 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13047 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13048 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13049 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13051 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13052 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13053 constant_alignment_word_strings.
13054 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13055 (CONSTANT_ALIGNMENT): Likewise.
13056 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13057 (arm_constant_alignment): New function.
13058 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13059 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13060 constant_alignment_word_strings.
13061 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13062 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13063 constant_alignment_word_strings.
13064 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13065 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13066 (cris_constant_alignment): New function.
13067 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13068 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13069 (epiphany_constant_alignment): New function.
13070 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13071 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13072 constant_alignment_word_strings.
13073 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13074 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13075 constant_alignment_word_strings.
13076 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13077 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13078 constant_alignment_word_strings.
13079 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13080 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13081 * config/i386/i386.c (ix86_constant_alignment): Make static.
13082 Use the same interface as the target hook.
13083 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13084 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13085 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13086 constant_alignment_word_strings.
13087 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13088 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13089 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13090 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13091 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13092 constant_alignment_word_strings.
13093 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13094 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13095 constant_alignment_word_strings.
13096 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13097 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13098 constant_alignment_word_strings.
13099 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13100 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13102 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13103 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13104 * config/mips/mips.c (mips_constant_alignment): New function.
13105 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13106 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13107 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13108 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13109 (mmix_constant_alignment): Make static. Use the same interface
13110 as the target hook.
13111 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13112 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13113 constant_alignment_word_strings.
13114 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13115 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13116 constant_alignment_word_strings.
13117 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13118 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13119 constant_alignment_word_strings.
13120 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13121 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13122 (rs6000_constant_alignment): New function.
13123 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13124 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13125 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13126 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13127 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13128 (rs6000_constant_alignment): New function.
13129 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13130 * config/s390/s390.c (s390_constant_alignment): New function.
13131 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13132 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13133 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13134 constant_alignment_word_strings.
13135 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13136 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13137 (sparc_constant_alignment): New function.
13138 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13139 * config/spu/spu.c (spu_constant_alignment): New function.
13140 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13141 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13142 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13143 constant_alignment_word_strings.
13144 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13145 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13146 constant_alignment_word_strings.
13147 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13148 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13149 constant_alignment_word_strings.
13150 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13151 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13152 (visium_constant_alignment): New function.
13153 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13154 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13155 (xtensa_constant_alignment): New function.
13156 * system.h (CONSTANT_ALIGNMENT): Poison.
13158 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13160 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13161 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13162 (rs6000_builtin_valid_without_lhs): New helper function.
13163 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13164 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13166 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13168 * target.h (vec_perm_indices): Use unsigned short rather than
13170 (auto_vec_perm_indices): Likewise.
13171 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13172 Use unsigned int rather than unsigned char.
13173 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13175 2017-09-25 Richard Biener <rguenther@suse.de>
13177 * cfgloop.h (sort_sibling_loops): Declare.
13178 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13179 (sort_sibling_loops): New function sorting the sibling loop list
13181 * graphite.c (graphite_transform_loops): Sort sibling loops.
13183 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13185 * target.def (vec_perm_const_ok): Change sel parameter to
13187 * optabs-query.c (can_vec_perm_p): Update accordingly.
13188 * doc/tm.texi: Regenerate.
13189 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13190 auto_vec_perm_indices and remove separate nelt field.
13191 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13192 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13193 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13194 (aarch64_expand_vec_perm_const): Update accordingly.
13195 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13196 to vec_perm_indices.
13197 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13198 auto_vec_perm_indices and remove separate nelt field.
13199 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13200 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13201 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13203 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13204 to vec_perm_indices.
13205 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13206 sel to vec_perm_indices.
13207 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13208 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13209 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13211 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13214 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13217 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13218 on the FUNCTION_DECL function context if it has a DIE that is a
13221 2017-09-25 Richard Biener <rguenther@suse.de>
13223 PR tree-optimization/82285
13224 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13227 2017-09-25 Tom de Vries <tom@codesourcery.com>
13231 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13234 2017-09-25 Richard Biener <rguenther@suse.de>
13236 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13237 ISL errors other than isl_error_quota happen. Dump if the
13238 schedule is the same.
13239 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13240 errors instead of aborting inside ISL.
13242 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13245 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13246 of libgcc_eh for m64.
13247 * config/i386/darwin64.h: Likewise.
13249 2017-09-25 Richard Biener <rguenther@suse.de>
13251 PR middle-end/82144
13252 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13253 attribute for incomplete types nor twice for complete ones.
13255 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13258 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13259 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13262 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13265 * config/i386/i386.opt (mprefer-avx256): Use
13266 ix86_target_flags variable.
13267 * config/i386/i386.c (ix86_target_string): Move
13268 -mprefer-avx256 to flag2_opts.
13270 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13272 PR middle-end/35691
13273 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13274 and x != -1 | y != -1 into (x & y) != -1.
13276 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13278 * config.gcc: Add new case statement to set
13279 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13280 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13281 s390x-*-linux* case statements. Added aarch64 to the list of
13282 supported architectures.
13284 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13286 PR tree-optimization/82289
13287 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13288 STMT_VINFO_RELEVANT_P.
13290 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13291 Alan Hayward <alan.hayward@arm.com>
13292 David Sherwood <david.sherwood@arm.com>
13294 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13295 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13296 (extract_range_from_binary_expr_1): Don't call
13297 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13299 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13300 Alan Hayward <alan.hayward@arm.com>
13301 David Sherwood <david.sherwood@arm.com>
13303 * target.def (preferred_vector_alignment): New hook.
13304 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13306 * doc/tm.texi: Regenerate.
13307 * targhooks.h (default_preferred_vector_alignment): Declare.
13308 * targhooks.c (default_preferred_vector_alignment): New function.
13309 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13311 (DR_TARGET_ALIGNMENT): New macro.
13312 (aligned_access_p): Update commentary.
13313 (vect_known_alignment_in_bytes): New function.
13314 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13316 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13317 Calculate the misalignment based on the target alignment rather than
13319 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13320 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13321 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13322 the target alignment, rather than masking the element misalignment
13323 with the number of elements in a vector. Also use the target
13324 alignment when calculating the maximum number of peels.
13325 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13326 instead of TYPE_ALIGN_UNIT.
13327 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13328 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13329 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13330 (vect_create_data_ref_ptr): Likewise.
13331 (vect_setup_realignment): Realign by ANDing with
13332 -DR_TARGET_MISALIGNMENT.
13333 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13334 the number of peels based on DR_TARGET_ALIGNMENT.
13335 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13336 with the guaranteed alignment boundary when deciding whether
13338 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13339 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13340 (ensure_base_align): Remove stmt_info parameter. Get the
13341 target base alignment from DR_TARGET_ALIGNMENT.
13342 (vectorizable_store): Update call accordingly. Interpret
13343 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13345 (vectorizable_load): Likewise.
13347 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13348 Alan Hayward <alan.hayward@arm.com>
13349 David Sherwood <david.sherwood@arm.com>
13351 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13352 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13353 (vect_enhance_data_refs_alignment): Likewise.
13355 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13357 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13358 error. Only quit immediately if parsing is complete.
13359 (BEGIN): Initialize fatal_err and parse_done.
13360 (begin fpu, end fpu): Check number of arguments.
13361 (begin arch, end arch): Likewise.
13362 (begin cpu, end cpu): Likewise.
13363 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13364 (optalias): Likewise.
13366 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13368 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13369 * config/arm/arm-isa.h: Delete. Move definitions to ...
13370 * arm-cpus.in: ... here. Use new feature and fgroup values.
13371 * config/arm/arm.c (arm_option_override): Use lower case for feature
13373 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13374 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13375 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13377 (print_isa_bits_for): New function.
13378 (gen_isa): New function.
13379 (gen_comm_data): Use print_isa_bits_for.
13380 (define feature): New keyword.
13381 (define fgroup): New keyword.
13382 * config/arm/t-arm (TM_H): Remove.
13383 (GTM_H): Add arm-isa.h.
13384 (arm-isa.h): Add rule to generate file.
13385 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13386 case for feature bit names.
13388 2017-09-22 Richard Biener <rguenther@suse.de>
13390 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13392 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13393 print of no dependency loops ...
13394 * graphite.c (graphite_transform_loops): ... here.
13395 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13397 (same_close_phi_node, remove_duplicate_close_phi,
13398 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13399 (canonicalize_loop_closed_ssa): ... here and simplify.
13400 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13401 (optimize_isl): Use dump_printf_loc to tell when we stopped
13402 optimizing because of an ISL timeout.
13404 2017-09-22 Richard Biener <rguenther@suse.de>
13406 PR tree-optimization/82291
13407 * tree-if-conv.c (predicate_mem_writes): Make sure to
13408 remove writes in blocks predicated with false.
13410 2017-09-22 Richard Biener <rguenther@suse.de>
13412 * sese.c: Include cfganal.h.
13413 (if_region_set_false_region): Remove.
13414 (create_if_region_on_edge): Likewise.
13415 (move_sese_in_condition): Re-implement without destroying
13418 2017-09-22 Richard Biener <rguenther@suse.de>
13420 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13421 Verify both BBs contain loop PHI nodes before dispatching to
13422 copy_loop_phi_args.
13423 (graphite_regenerate_ast_isl): Do not recompute dominators,
13424 do not verify three times. Restructure for clarity.
13425 * graphite-scop-detection.c (same_close_phi_node,
13426 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13427 defined_in_loop_p, canonicalize_loop_closed_ssa,
13428 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13429 checking and SSA rewrite, move to ...
13430 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13431 (graphite_initialize): Do not pass in ctx, do not reset the
13432 SCEV cache, compute only dominators.
13433 (graphite_transform_loops): Allocate ISL ctx after
13434 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13435 Maintain post-dominators only around build_scops.
13436 * sese.c (if_region_set_false_region): Make static. Free
13437 and recompute dominators.
13438 (move_sese_in_condition): Assert we don't get called with
13439 post-dominators computed.
13440 * sese.h (if_region_set_false_region): Remove.
13442 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13444 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13445 mode attribute for TARGET_AVX512VL.
13447 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13449 * config/i386/i386.opt (mprefer-avx256): New option.
13450 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13452 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13453 for TARGET_PREFER_AVX256.
13454 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13456 2017-09-21 Jeff Law <law@redhat.com>
13458 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13459 Fix dump output if the only stack space is for pushed registers.
13461 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13463 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13466 2017-09-21 Martin Sebor <msebor@redhat.com>
13469 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13470 code (in C++) or code that triggers warnings.
13472 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13474 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13476 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13478 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13479 * sched-rgn.c: Ditto.
13480 * sel-sched-ir.c: Ditto.
13482 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13484 * toplev.h (set_random_seed): Adjust return type.
13485 * toplev.c (init_local_tick): Move eager initialization of random_seed
13486 to get_random_seed. Adjust comment.
13487 (init_random_seed): Inline to get_random_seed, delete.
13488 (get_random_seed): Initialize random_seed lazily.
13489 (set_random_seed): Do not return previous value.
13490 (print_switch_value): Do not call get_random_seed.
13492 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13494 * cgraph.c (delete_function_version): New, broken out from...
13495 (cgraph_node::delete_function_version): ...here. Rename to
13496 cgraph_node::delete_function_version_by_decl. Update all uses.
13497 (cgraph_node::remove): Call delete_function_version.
13499 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13502 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13503 VAR_DECLs to which addressable non-volatile parameters are mapped
13504 and for id->retvar after the return value assignment. Clear
13505 id->retval and id->retbnd after inlining.
13507 2017-09-21 Richard Biener <rguenther@suse.de>
13509 PR tree-optimization/82276
13510 PR tree-optimization/82244
13511 * tree-vrp.c (build_assert_expr_for): Set
13512 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13514 (remove_range_assertions): Revert earlier change.
13516 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13519 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13521 2017-09-21 Richard Biener <rguenther@suse.de>
13523 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13524 Restore valid IL after code generation errors.
13525 * graphite.c (graphite_transform_loops): Diagnose code
13526 generation issues as MSG_MISSED_OPTIMIZATION and continue
13527 with processing SCOPs.
13529 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13530 Alan Hayward <alan.hayward@arm.com>
13531 David Sherwood <david.sherwood@arm.com>
13533 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13534 rather than choosing between plus_constant and gen_rtx_<CODE>.
13535 * expr.c (emit_push_insn): Likewise.
13536 (expand_expr_real_2): Likewise.
13538 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13539 Alan Hayward <alan.hayward@arm.com>
13540 David Sherwood <david.sherwood@arm.com>
13542 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13544 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13545 Alan Hayward <alan.hayward@arm.com>
13546 David Sherwood <david.sherwood@arm.com>
13548 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13549 calling tree_to_uhwi.
13551 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13552 Alan Hayward <alan.hayward@arm.com>
13553 David Sherwood <david.sherwood@arm.com>
13555 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13556 INTEGER_CST rather than a negative test for ADDR_EXPR.
13558 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13559 Alan Hayward <alan.hayward@arm.com>
13560 David Sherwood <david.sherwood@arm.com>
13562 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13563 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13565 2017-09-21 Richard Biener <rguenther@suse.de>
13567 PR tree-optimization/71351
13568 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13569 graphite_create_new_loop_guard): Remove, fold remaining parts
13571 (translate_isl_ast_node_for): ... here and simplify.
13573 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13576 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13577 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13578 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13579 alternative always use QI mode, for -Os imov (=R,R) alternative
13580 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13583 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13584 Jeff Law <law@redhat.com>
13586 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13587 (allocate_stack_space): New function, partially extracted from
13588 s390_emit_prologue.
13589 (s390_emit_prologue): Track offset to most recent stack probe.
13590 Code to allocate space moved into allocate_stack_space.
13591 Dump actions when no stack is allocated.
13592 (s390_prologue_plus_offset): New function.
13593 (s390_emit_stack_probe): Likewise.
13595 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13597 * common.opt (Wa, Wl, Wp, g, gz=): Add
13599 (gno-column-info): Remove.
13600 (gcolumn-info): Drop RejectNegative.
13601 (gno-): New prefix.
13602 (gno-record-gcc-switches): Remove.
13603 (grecord-gcc-switches): Drop RejectNegative.
13604 (gno-split-dwarf): Remove.
13605 (gsplit-dwarf): Drop RejectNegative.
13606 (gno-strict-dwarf): Remove.
13607 (gstrict-dwarf): Drop RejectNegative.
13608 * config/darwin.opt (gfull, gused): Add RejectNegative.
13609 * dwarf2out.c (gen_producer_string): Drop
13610 gno-record-gcc-switches handler.
13611 * optc-gen.awk: Add g to prefixes with negative forms.
13612 * opts-common.c (remapping_prefix_p): New.
13613 (find_opt): Check it.
13614 (generate_canonical_option): Test g prefix.
13615 (option_map): Add -gno- mapping.
13616 (add_misspelling_candidates): Check remapping_prefix_p.
13618 2017-09-20 Jeff Law <law@redhat.com>
13620 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13621 thinko in stack clash protection support.
13623 * explow.c (compute_stack_clash_protection_loop_data): Use
13624 CONST_INT_P instead of explicit test. Verify object is a
13625 CONST_INT_P before looking at INTVAL.
13626 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13627 instead of explicit test.
13629 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13632 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13633 address instead of to r1 and r11.
13635 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13637 * config.gcc: Support "knm".
13638 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13639 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13641 * config/i386/i386.c (m_KNM): Define.
13642 (processor_target_table): Add "knm".
13644 (ix86_option_override_internal): Add "knm".
13645 (ix86_issue_rate): Add PROCESSOR_KNM.
13646 (ix86_adjust_cost): Ditto.
13647 (ia32_multipass_dfa_lookahead): Ditto.
13648 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13649 (fold_builtin_cpu): Add M_INTEL_KNM.
13650 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13651 (processor_type): Add PROCESSOR_KNM.
13652 * config/i386/x86-tune.def: Add m_KNM.
13653 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13655 2017-09-20 Richard Biener <rguenther@suse.de>
13657 PR tree-optimization/80213
13658 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13659 are allowed in empty BBs as well.
13660 (canonicalize_loop_closed_ssa): Also look for other complex
13662 (scop_detection::get_sese): Include the loop-closed PHI block
13664 (scop_detection::merge_sese): Remove code adding extra blocks.
13665 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13666 (build_scops): Assert the final returned scop is invalid.
13668 2017-09-20 Richard Biener <rguenther@suse.de>
13670 PR tree-optimization/82264
13671 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13673 (vn_phi_lookup): Likewise.
13674 (vn_phi_insert): Likewise.
13676 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13678 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13679 that fits into uhwi or shwi, add DW_AT_const_value regardless
13680 of early_dwarf without going through RTL, using add_AT_unsigned
13683 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13684 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13685 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13686 (DEBUG_MACRO_SECTION): Likewise.
13687 (DEBUG_LTO_MACRO_SECTION): Likewise.
13688 (DEBUG_STR_DWO_SECTION): Likewise.
13689 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13690 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13691 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13692 (DEBUG_LTO_LINE_STR_SECTION): Define.
13693 (init_sections_and_labels): Initialize debug_line_str_section
13694 variable. Initialize debug_loc_section for -gdwarf-5 to
13695 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13697 2017-09-20 Richard Biener <rguenther@suse.de>
13699 * graphite-sese-to-poly.c (extract_affine): Properly handle
13700 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13702 2017-09-20 Richard Biener <rguenther@suse.de>
13704 PR tree-optimization/81373
13705 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13706 Force SESE live-out defs to be handled even if they are
13709 2017-09-19 Jeff Law <law@redhat.com>
13711 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13712 nothing for stack adjustments with REG_STACK_CHECK.
13713 * sched-deps.c (parse_add_or_inc): Reject insns with
13714 REG_STACK_CHECK from dependency breaking.
13715 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13716 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13717 * reg-notes.def (STACK_CHECK): New note.
13719 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13720 (ix86_expand_prologue): Dump stack clash info as needed.
13721 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13723 * function.c (dump_stack_clash_frame_info): New function.
13724 * function.h (dump_stack_clash_frame_info): Prototype.
13725 (enum stack_clash_probes): New enum.
13727 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13728 flag_stack_clash_protection.
13729 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13730 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13731 (arm_frame_pointer_required): Likewise.
13732 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13733 (ia64_expand_prologue): Likewise.
13734 * config/mips/mips.c (mips_expand_prologue): Likewise.
13735 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13736 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13737 (sparc_flat_expand_prologue): Likewise.
13738 * config/spu/spu.c (spu_expand_prologue): Likewise.
13740 * explow.c: Include "params.h".
13741 (anti_adjust_stack_and_probe_stack_clash): New function.
13742 (get_stack_check_protect): Likewise.
13743 (compute_stack_clash_protection_loop_data): Likewise.
13744 (emit_stack_clash_protection_loop_start): Likewise.
13745 (emit_stack_clash_protection_loop_end): Likewise.
13746 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13747 Use anti_adjust_stack_and_probe_stack_clash.
13748 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13749 (emit_stack_clash_protection_loop_start): Likewise.
13750 (emit_stack_clash_protection_loop_end): Likewise.
13751 * rtl.h (get_stack_check_protect): Prototype.
13752 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13753 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13754 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13756 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13758 * doc/tm.texi: Rebuilt.
13759 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13760 get_stack_check_protect.
13761 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13762 * config/arm/arm.c (arm_expand_prologue): Likewise.
13763 (arm_frame_pointer_required): Likewise.
13764 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13765 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13766 * config/mips/mips.c (mips_expand_prologue): Likewise.
13767 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13768 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13769 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13770 (sparc_flat_expand_prologue): Likewise.
13772 * common.opt (-fstack-clash-protection): New option.
13773 * flag-types.h (enum stack_check_type): Note difference between
13774 -fstack-check= and -fstack-clash-protection.
13775 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13776 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13777 * toplev.c (process_options): Issue warnings/errors for cases
13778 not handled with -fstack-clash-protection.
13779 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13780 (-fstack-check): Note additional problem with -fstack-check=generic.
13781 Note that -fstack-check is primarily for Ada and refer users
13782 to -fstack-clash-protection for stack-clash-protection.
13783 Document new params for stack clash protection.
13785 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13787 * config/i386/i386.c (ix86_split_long_move): Do not handle
13788 address used for LEA in a special way.
13790 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13792 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
13794 2017-09-19 Martin Sebor <msebor@redhat.com>
13797 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
13798 of incompatible types.
13800 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
13802 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13803 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
13804 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13805 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
13807 2017-09-19 Richard Biener <rguenther@suse.de>
13809 PR tree-optimization/82244
13810 * tree-vrp.c (remove_range_assertions): Do not propagate
13811 a constant to abnormals but replace the assert with a copy.
13813 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
13815 PR rtl-optimization/57878
13816 PR rtl-optimization/68988
13817 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
13818 avoidance test involving non_reload_pseudos. Move frequency test
13819 below the general fragmentation avoidance test.
13821 2017-09-19 Richard Biener <rguenther@suse.de>
13823 PR tree-optimization/69728
13824 * graphite-sese-to-poly.c (schedule_error): New global.
13825 (add_loop_schedule): Handle empty domain by failing the
13827 (build_original_schedule): Handle schedule_error.
13829 2017-09-19 Richard Biener <rguenther@suse.de>
13831 * graphite-scop-detection.c (scop_detection::can_represent_loop):
13832 Do not iterate to sibling loops but only to siblings of inner
13835 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
13838 * config/m68k/m68k.md (moveq feeding equality comparison): Check
13839 that the registers are different.
13841 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
13843 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
13844 to processor_model and "amdfam17h" to arch_names_table.
13845 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
13847 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13850 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
13852 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13853 Alan Hayward <alan.hayward@arm.com>
13854 David Sherwood <david.sherwood@arm.com>
13856 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
13858 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
13860 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
13861 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
13862 vect_schedule_slp_instance.
13863 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
13864 Update call to vect_slp_analyze_node_operations. Simplify return
13866 (vect_slp_analyze_bb_1): Update call accordingly.
13867 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
13868 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
13869 (vect_schedule_slp): Update call accordingly.
13871 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13872 Alan Hayward <alan.hayward@arm.com>
13873 David Sherwood <david.sherwood@arm.com>
13875 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
13876 with types that aren't in fact scalar.
13878 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13880 * tree-vect-slp.c (vect_record_max_nunits): New function,
13882 (vect_build_slp_tree_1): ...here.
13883 (vect_build_slp_tree_2): Call it for phis too.
13885 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13887 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
13888 to vect_get_vec_def_for_operand when getting the mask operand.
13890 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13891 Alan Hayward <alan.hayward@arm.com>
13892 David Sherwood <david.sherwood@arm.com>
13894 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
13897 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13898 Alan Hayward <alan.hayward@arm.com>
13899 David Sherwood <david.sherwood@arm.com>
13901 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
13902 calculation for vector booleans.
13904 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13905 Alan Hayward <alan.hayward@arm.com>
13906 David Sherwood <david.sherwood@arm.com>
13908 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
13910 (vect_transform_stmt): ...here.
13911 (vect_analyze_stmt): Use it instead of calling
13912 vectorizable_live_operation directly.
13914 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
13916 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
13917 non-SIMT targets in acc vector loops.
13919 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
13921 * configure.ac: Add arc and check if assembler supports gdwarf2.
13922 * configure: Regenerate.
13924 2017-09-18 Richard Biener <rguenther@suse.de>
13926 PR tree-optimization/82220
13927 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
13928 epilogue niters from the min_profitable_iters compute.
13930 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13933 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
13934 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
13935 (ix86_init_pic_reg): Revert 2017-09-01 changes.
13937 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
13940 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
13941 switching to a new text section.
13943 2017-09-18 Richard Biener <rguenther@suse.de>
13945 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
13947 (build_alias_set): Reject aliases with no access function.
13949 2017-09-18 Richard Biener <rguenther@suse.de>
13951 PR tree-optimization/79622
13952 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
13954 (build_cross_bb_scalars_use): Likewise.
13956 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
13958 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
13960 2017-09-18 Alan Modra <amodra@gmail.com>
13963 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
13964 stack_pointer_rtx for count 0. Update comments. Break up
13965 large rtl expression.
13967 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
13969 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
13970 Increase to 20 bytes.
13971 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
13972 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
13973 or avx version of the stub.
13975 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
13978 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
13979 compute the minimum stack alignment. Also update preferred stack
13980 boundary for leaf functions.
13982 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
13984 PR tree-optimization/82228
13985 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
13988 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13990 * common/config/nds32/nds32-common.c
13991 (nds32_option_optimization_table): Refine formatting.
13992 (nds32_option_optimization_table): Use -fsched-pressure and
13993 -fomit-frame-pointer for specific optimization level.
13995 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13997 * config/nds32/nds32.c: Refine formatting and comments.
13998 * config/nds32/nds32.h: Likewise.
13999 * config/nds32/nds32.md: Likewise.
14000 * config/nds32/nds32-cost.c: Likewise.
14001 * config/nds32/nds32-isr.c: Likewise.
14002 * config/nds32/nds32-md-auxiliary.c: Likewise.
14003 * config/nds32/nds32-multiple.md: Likewise.
14004 * config/nds32/nds32-predicates.c: Likewise.
14006 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14007 Jakub Jelinek <jakub@redhat.com>
14009 Add support for -std=c++2a.
14010 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14012 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14014 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14017 * doc/extend.texi (Common Function Attributes): Add
14018 references to ARM, AArch64, and S/390 specific attributes.
14019 (Function Specific Option Pragmas): Add AArch64 and S/390
14020 to list of back ends that support the target pragma.
14022 2017-09-15 Nathan Sidwell <nathan@acm.org>
14024 * doc/standards.texi: Fix C++17 description. Update URLs for
14027 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14029 * common.opt (Wcast-align=strict): New warning option.
14030 * doc/invoke.texi: Document -Wcast-align=strict.
14032 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14034 * cgraph.h (cgraph_thunk_info): Add comments.
14035 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14036 assert for VIRTUAL_* arguments stricter.
14038 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14040 PR tree-optimization/71026
14041 * match.pd: Move RDIV patterns from fold-const.c
14042 * fold-const.c (distribute_real_division): Removed.
14043 (fold_binary_loc): Remove calls to distribute_real_divison.
14045 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14047 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14048 c++1z and gnu++1z as deprecated. Change other references to
14049 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14050 Change -Wc++1z-compat to -Wc++17-compat.
14051 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14052 * dwarf2out.c (highest_c_language): Handle C++17.
14053 (gen_compile_unit_die): Likewise.
14055 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14057 PR rtl-optimization/82192
14058 * combine.c (make_extraction): Don't look through non-paradoxical
14059 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14062 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14063 Alan Hayward <alan.hayward@arm.com>
14064 David Sherwood <david.sherwood@arm.com>
14066 * target.def (function_arg_offset): New hook.
14067 * targhooks.h (default_function_arg_offset): Declare.
14068 * targhooks.c (default_function_arg_offset): New function.
14069 * function.c (locate_and_pad_parm): Use
14070 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14071 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14072 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14073 * doc/tm.texi: Regenerate.
14074 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14075 * config/spu/spu.c (spu_function_arg_offset): New function.
14076 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14077 * system.h (FUNCTION_ARG_OFFSET): Poison.
14079 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14080 Alan Hayard <alan.hayward@arm.com>
14081 David Sherwood <david.sherwood@arm.com>
14083 * target.def (truly_noop_truncation): New hook.
14084 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14085 than TRULY_NOOP_TRUNCATION.
14086 * hooks.h (hook_bool_uint_uint_true): Declare.
14087 * hooks.c (hook_bool_uint_uint_true): New function.
14088 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14089 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14090 * doc/tm.texi: Regenerate.
14091 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14092 rather than TRULY_NOOP_TRUNCATION in comments.
14093 (simplify_comparison): Likewise.
14094 (record_truncated_value): Likewise.
14095 * expmed.c (extract_bit_field_1): Likewise.
14096 (extract_split_bit_field): Likewise.
14097 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14098 instead of TRULY_NOOP_TRUNCATION.
14099 * function.c (assign_parm_setup_block): Likewise.
14100 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14101 * rtlhooks.c: Include target.h.
14102 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14103 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14104 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14105 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14106 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14107 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14108 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14109 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14110 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14111 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14112 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14113 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14114 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14115 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14116 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14117 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14118 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14119 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14120 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14121 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14122 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14123 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14124 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14125 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14126 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14127 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14128 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14129 rather than TRULY_NOOP_TRUNCATION in comments.
14130 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14131 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14132 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14133 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14134 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14135 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14136 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14137 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14138 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14139 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14140 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14141 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14142 rather than TRULY_NOOP_TRUNCATION in comments.
14143 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14144 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14145 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14146 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14147 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14148 TRULY_NOOP_TRUNCATION condition.
14149 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14150 (TRULY_NOOP_TRUNCATION): Delete.
14151 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14152 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14153 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14154 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14155 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14156 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14157 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14158 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14159 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14160 rather than TRULY_NOOP_TRUNCATION in comments.
14161 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14162 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14163 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14164 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14165 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14166 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14168 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14171 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14172 (*cmp_ior): Likewise.
14173 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14174 (*ior_scc_scc_cmp): Likewise.
14175 (*and_scc_scc): Likewise.
14176 (*and_scc_scc_cmp): Likewise.
14178 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14179 Alan Hayard <alan.hayward@arm.com>
14180 David Sherwood <david.sherwood@arm.com>
14182 * target.def (can_change_mode_class): New hook.
14183 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14184 (hard_regno_nregs): Likewise.
14185 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14186 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14187 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14188 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14189 (LOAD_EXTEND_OP): Update accordingly.
14190 * doc/tm.texi: Regenerate.
14191 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14192 CANNOT_CHANGE_MODE_CLASS.
14193 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14194 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14195 * combine.c (simplify_set): Update accordingly.
14196 * emit-rtl.c (validate_subreg): Likewise.
14197 * recog.c (general_operand): Likewise.
14198 * regcprop.c (mode_change_ok): Likewise.
14199 * reload1.c (choose_reload_regs): Likewise.
14200 (inherit_piecemeal_p): Likewise.
14201 * rtlanal.c (simplify_subreg_regno): Likewise.
14202 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14203 instead of CANNOT_CHANGE_MODE_CLASS.
14204 (reload_cse_simplify_operands): Likewise.
14205 * reload.c (push_reload): Use targetm.can_change_mode_class
14206 instead of CANNOT_CHANGE_MODE_CLASS.
14207 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14208 REG_CANNOT_CHANGE_MODE_P.
14209 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14210 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14211 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14212 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14213 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14214 (arm_can_change_mode_class): New function.
14215 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14216 than CANNOT_CHANGE_MODE_CLASS in comments.
14217 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14218 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14219 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14220 (ix86_can_change_mode_class): ...this new function, inverting the
14221 sense of the return value.
14222 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14223 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14224 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14225 (ia64_can_change_mode_class): New function.
14226 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14227 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14228 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14229 (m32c_can_change_mode_class): ...this new function, inverting the
14230 sense of the return value.
14231 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14232 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14233 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14234 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14235 (mips_can_change_mode_class): ...this new function, inverting the
14236 sense of the return value.
14237 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14238 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14239 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14240 (msp430_can_change_mode_class): New function.
14241 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14242 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14243 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14244 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14245 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14246 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14247 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14248 (pa_cannot_change_mode_class): Replace with...
14249 (pa_can_change_mode_class): ...this new function, inverting the
14250 sense of the return value.
14251 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14252 than CANNOT_CHANGE_MODE_CLASS in comments.
14253 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14254 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14255 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14256 (pdp11_cannot_change_mode_class): Replace with...
14257 (pdp11_can_change_mode_class): ...this new function, inverting the
14258 sense of the return value.
14259 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14260 * config/powerpcspe/powerpcspe-protos.h
14261 (rs6000_cannot_change_mode_class_ptr): Delete.
14262 * config/powerpcspe/powerpcspe.c
14263 (rs6000_cannot_change_mode_class_ptr): Delete.
14264 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14265 (rs6000_option_override_internal): Assign to
14266 targetm.can_change_mode_class instead of
14267 rs6000_cannot_change_mode_class_ptr.
14268 (rs6000_cannot_change_mode_class): Replace with...
14269 (rs6000_can_change_mode_class): ...this new function, inverting the
14270 sense of the return value.
14271 (rs6000_debug_cannot_change_mode_class): Replace with...
14272 (rs6000_debug_can_change_mode_class): ...this new function.
14273 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14274 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14275 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14276 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14277 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14279 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14280 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14281 (rs6000_option_override_internal): Assign to
14282 targetm.can_change_mode_class instead of
14283 rs6000_cannot_change_mode_class_ptr.
14284 (rs6000_cannot_change_mode_class): Replace with...
14285 (rs6000_can_change_mode_class): ...this new function, inverting the
14286 sense of the return value.
14287 (rs6000_debug_cannot_change_mode_class): Replace with...
14288 (rs6000_debug_can_change_mode_class): ...this new function.
14289 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14290 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14291 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14292 (s390_can_change_mode_class): ...this new function, inverting the
14293 sense of the return value.
14294 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14295 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14296 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14297 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14298 (sh_cannot_change_mode_class): Replace with...
14299 (sh_can_change_mode_class): ...this new function, inverting the
14300 sense of the return value.
14301 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14302 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14303 (sparc_can_change_mode_class): New function.
14304 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14305 * config/spu/spu.c (spu_can_change_mode_class): New function.
14306 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14307 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14308 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14309 (visium_can_change_mode_class): New function.
14310 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14312 2017-09-15 Richard Biener <rguenther@suse.de>
14314 PR tree-optimization/82217
14315 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14316 but not undefined case.
14318 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14321 * postreload.c (reload_cse_simplify_operands): Skip
14322 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14324 2017-09-15 Richard Biener <rguenther@suse.de>
14326 PR tree-optimization/68823
14327 * graphite-scop-detection.c (build_alias_set): If we have a
14328 possible dependence check whether we can handle them by just
14329 looking at the DRs DR_ACCESS_FNs.
14330 (build_scops): If build_alias_set fails, fail the SCOP.
14332 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14334 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14335 to support float128 built-in functions that require the ISA 3.0
14337 (BU_FLOAT128_3_HW): Likewise.
14338 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14339 built-in functions.
14340 (FMAF128): Likewise.
14342 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14343 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14344 floating point instructions.
14345 (rs6000_invalid_builtin): Likewise.
14346 (rs6000_builtin_mask_names): Likewise.
14347 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14348 (RS6000_BTM_FLOAT128_HW): Likewise.
14349 (RS6000_BTM_COMMON): Likewise.
14350 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14352 * doc/extend.texi (RS/6000 built-in functions): Document the
14353 IEEE 128-bit floating point square root and fused multiply-add
14354 built-in functions.
14356 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14358 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14359 reg (r2) isn't in the set of registers defined in the prologue.
14361 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14362 Alan Hayward <alan.hayward@arm.com>
14363 David Sherwood <david.sherwood@arm.com>
14365 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14366 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14367 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14368 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14369 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14371 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14372 max_vectorization_factor.
14373 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14375 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14376 Alan Hayward <alan.hayward@arm.com>
14377 David Sherwood <david.sherwood@arm.com>
14379 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14380 (vect_worthwhile_without_simd_p): Declare.
14381 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14382 (vectorizable_reduction): Use it.
14383 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14384 (vectorizable_operation): Likewise.
14386 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14387 Alan Hayward <alan.hayward@arm.com>
14388 David Sherwood <david.sherwood@arm.com>
14390 * tree-vectorizer.h (vect_get_num_copies): New function.
14391 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14392 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14393 (vectorizable_induction): Likewise.
14394 (vectorizable_live_operation): Likewise.
14395 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14396 (vectorizable_bswap): Likewise.
14397 (vectorizable_call): Likewise.
14398 (vectorizable_conversion): Likewise.
14399 (vectorizable_assignment): Likewise.
14400 (vectorizable_shift): Likewise.
14401 (vectorizable_operation): Likewise.
14402 (vectorizable_store): Likewise.
14403 (vectorizable_load): Likewise.
14404 (vectorizable_condition): Likewise.
14405 (vectorizable_comparison): Likewise.
14406 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14408 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14409 Alan Hayward <alan.hayward@arm.com>
14410 David Sherwood <david.sherwood@arm.com>
14412 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14413 of vect_init_vector.
14415 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14416 Alan Hayward <alan.hayward@arm.com>
14417 David Sherwood <david.sherwood@arm.com>
14419 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14420 an inline wrapper that provides a location.
14421 (gimple_build_vector): Likewise.
14422 * gimple-fold.c (gimple_build_vector_from_val): New function.
14423 (gimple_build_vector): Likewise.
14424 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14425 functions to build the initial value. Always return a gimple value.
14426 (get_initial_defs_for_reduction): Likewise. Only compute
14428 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14429 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14430 (vectorizable_induction): Use gimple_build_vector rather than
14433 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14434 Alan Hayward <alan.hayward@arm.com>
14435 David Sherwood <david.sherwood@arm.com>
14437 * target.h (vec_perm_indices): New typedef.
14438 (auto_vec_perm_indices): Likewise.
14439 * optabs-query.h: Include target.h
14440 (can_vec_perm_p): Take a vec_perm_indices *.
14441 * optabs-query.c (can_vec_perm_p): Likewise.
14442 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14443 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14444 * tree-vect-generic.c (lower_vec_perm): Likewise.
14445 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14446 (vect_grouped_load_supported): Likewise.
14447 (vect_shift_permute_load_chain): Likewise.
14448 (vect_permute_store_chain): Use auto_vec_perm_indices.
14449 (vect_permute_load_chain): Likewise.
14450 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14451 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14452 Update uses of can_vec_perm_p.
14453 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14454 mode with a number of elements. Take a vec_perm_indices *.
14455 (vect_create_epilog_for_reduction): Update accordingly.
14456 Use auto_vec_perm_indices.
14457 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14458 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14459 (vect_transform_slp_perm_load): Likewise.
14460 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14461 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14462 (vect_gen_perm_mask_checked): Likewise.
14463 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14464 (vect_gen_perm_mask_checked): Likewise.
14465 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14466 (vectorizable_store): Likewise.
14467 (vectorizable_load): Likewise.
14468 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14469 (vectorizable_bswap): Likewise.
14471 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14472 Alan Hayward <alan.hayward@arm.com>
14473 David Sherwood <david.sherwood@arm.com>
14475 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14476 * tree.c (build_vector): Likewise.
14477 (build_vector_from_ctor): Update accordingly.
14478 (build_vector_from_val): Likewise.
14479 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14480 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14481 * tree-vect-generic.c (add_rshift): Likewise.
14482 (expand_vector_divmod): Likewise.
14483 (optimize_vector_constructor): Likewise.
14484 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14485 (vect_transform_slp_perm_load): Likewise.
14486 (vect_schedule_slp_instance): Likewise.
14487 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14488 (vectorizable_call): Likewise.
14489 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14490 * expmed.c (make_tree): Likewise.
14491 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14492 a vector passed to build_vector.
14493 (fold_convert_const): Likewise.
14494 (exact_inverse): Likewise.
14495 (fold_ternary_loc): Likewise.
14496 (fold_relational_const): Likewise.
14497 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14498 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14499 (const_unop): Likewise. Store the reduction accumulator in a
14500 variable rather than an array.
14501 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14502 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14503 the new vector, rather than constructing it after the input arrays.
14504 (native_interpret_vector): Use auto_vec<tree> when building
14505 a vector passed to build_vector. Add elements in order.
14506 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14507 auto_vec<tree> when building a vector passed to build_vector.
14508 (vect_create_epilog_for_reduction): Likewise.
14509 (vectorizable_induction): Likewise.
14510 (get_initial_def_for_reduction): Likewise. Fix indentation of
14512 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14514 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14515 passed to build_vector.
14517 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14518 Alan Hayward <alan.hayward@arm.com>
14519 David Sherwood <david.sherwood@arm.com>
14521 * tree-core.h (tree_base::u): Add an "nelts" field.
14522 (tree_vector): Use VECTOR_CST_NELTS as the length.
14523 * tree.c (tree_size): Likewise.
14524 (make_vector): Initialize VECTOR_CST_NELTS.
14525 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14526 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14527 TYPE_VECTOR_SUBPARTS.
14528 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14529 as the number of units, setting it from VECTOR_CST_NELTS.
14530 (const_vector_from_tree): Likewise.
14531 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14532 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14533 (fold_negate_expr_1): Likewise.
14534 (fold_convert_const): Likewise.
14535 (const_binop): Likewise. Differentiate the number of output and
14537 (const_unop): Likewise.
14538 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14539 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14540 in cases that did the opposite.
14542 2017-09-14 Richard Biener <rguenther@suse.de>
14544 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14547 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14549 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14551 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14554 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14555 if and where to split a bb, except for splitting before debug insn
14556 sequences followed by non-label real insn. Delete debug insns
14557 in between basic blocks.
14559 * combine.c (make_compound_operation_int): Formatting fixes.
14561 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14562 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14563 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14564 * config/sol2.h (LINK_EH_SPEC): Likewise.
14565 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14566 * config/s390/linux.h (LINK_SPEC): Likewise.
14567 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14568 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14569 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14570 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14571 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14572 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14573 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14574 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14576 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14578 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14580 (ENDFILE_LINUX_SPEC): Likewise.
14581 (LINK_EH_SPEC): Likewise.
14582 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14583 (LINK_OS_LINUX_SPEC32): Likewise.
14584 (LINK_OS_LINUX_SPEC64): Likewise.
14585 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14586 (LINK_OS_LINUX_SPEC): Likewise.
14588 2017-09-13 Martin Liska <mliska@suse.cz>
14590 PR middle-end/82154
14591 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14592 CASE_HIGH is NULL_TREE.
14594 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14595 Alan Hayward <alan.hayward@arm.com>
14596 David Sherwood <david.sherwood@arm.com>
14598 * target.def (secondary_memory_needed): New hook.
14599 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14600 instead of SECONDARY_MEMORY_NEEDED.
14601 (secondary_memory_needed_mode): Likewise.
14602 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14603 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14604 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14605 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14606 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14607 * doc/tm.texi: Regenerate.
14608 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14609 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14610 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14611 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14612 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14613 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14614 mode argument first and change the reg_class arguments to reg_class_t.
14615 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14616 Make static. Update the call to inline_secondary_memory_needed.
14617 (ix86_register_move_cost): Update the call to
14618 inline_secondary_memory_needed.
14619 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14620 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14622 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14623 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14625 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14626 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14627 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14628 and match hook interface. Add comment from mips.h.
14629 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14630 * config/mmix/mmix.md (truncdfsf2): Refer to
14631 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14633 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14634 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14635 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14636 (pa_secondary_memory_needed): New function.
14637 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14638 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14639 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14640 (pdp11_secondary_memory_needed): Make static and match hook interface.
14641 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14642 * config/powerpcspe/powerpcspe-protos.h
14643 (rs6000_secondary_memory_needed_ptr): Delete.
14644 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14646 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14647 (rs6000_option_override_internal): Assign to
14648 targetm.secondary_memory_needed rather than
14649 rs6000_secondary_memory_needed_ptr.
14650 (rs6000_secondary_memory_needed): Match hook interface.
14651 (rs6000_debug_secondary_memory_needed): Likewise.
14652 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14653 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14654 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14655 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14656 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14657 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14659 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14660 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14661 (rs6000_option_override_internal): Assign to
14662 targetm.secondary_memory_needed rather than
14663 rs6000_secondary_memory_needed_ptr.
14664 (rs6000_secondary_memory_needed): Match hook interface.
14665 (rs6000_debug_secondary_memory_needed): Likewise.
14666 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14667 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14668 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14669 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14670 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14671 (sparc_secondary_memory_needed): New function.
14672 * lra-constraints.c (check_and_process_move): Refer to
14673 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14675 (curr_insn_transform): Likewise.
14676 (process_alt_operands): Use targetm.secondary_memory_needed
14677 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14678 (check_secondary_memory_needed_p): Likewise.
14679 (choose_split_class): Likewise.
14680 * reload.c: Unconditionally include code that was previously
14681 conditional on SECONDARY_MEMORY_NEEDED.
14682 (push_secondary_reload): Use targetm.secondary_memory_needed
14683 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14684 (push_reload): Likewise.
14685 * reload1.c: Unconditionally include code that was previously
14686 conditional on SECONDARY_MEMORY_NEEDED.
14687 (choose_reload_regs): Use targetm.secondary_memory_needed
14688 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14689 (gen_reload): Likewise.
14690 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14692 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14693 Alan Hayward <alan.hayward@arm.com>
14694 David Sherwood <david.sherwood@arm.com>
14696 * target.def (secondary_memory_needed_mode): New hook:
14697 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14698 * targhooks.h (default_secondary_memory_needed_mode): New function.
14699 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14700 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14701 * doc/tm.texi: Regenerate.
14702 * lra-constraints.c (check_and_process_move): Use
14703 targetm.secondary_memory_needed_mode instead of
14704 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14705 (curr_insn_transform): Likewise.
14706 * reload.c (get_secondary_mem): Likewise.
14707 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14708 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14710 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14711 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14712 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14713 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14714 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14716 * config/powerpcspe/powerpcspe-protos.h
14717 (rs6000_secondary_memory_needed_mode): Delete.
14718 * config/powerpcspe/powerpcspe.c
14719 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14720 (rs6000_secondary_memory_needed_mode): Make static.
14721 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14722 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14724 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14726 (rs6000_secondary_memory_needed_mode): Make static.
14727 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14728 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14729 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14730 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14731 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14733 (sparc_secondary_memory_needed_mode): New function.
14734 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14736 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14738 * config/aarch64/constraints.md (Umq): New constraint.
14739 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14741 (mov<mode>): Update condition.
14743 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14745 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14746 when bitposition is the same.
14748 2017-09-13 Richard Biener <rguenther@suse.de>
14750 * dwarf2out.c (output_die_symbol): Remove.
14751 (output_die): Do not output a DIEs symbol.
14753 2017-09-13 Richard Biener <rguenther@suse.de>
14755 PR middle-end/82128
14756 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14757 default-def to avoid breaking iterator update with the weird
14758 interaction with cgraph_update_edges_for_call_stmt_node.
14760 2017-09-13 Richard Biener <rguenther@suse.de>
14762 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14763 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14764 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14765 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14767 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14769 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14770 Disable pc relative literal load irrespective of
14771 TARGET_FIX_ERR_A53_84341 for default.
14773 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14775 * config/sparc/sparc.c (output_return): Output the source location of
14776 the insn in the delay slot, if any.
14777 (output_sibcall): Likewise.
14779 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14782 * common.opt (-static-pie): New alias.
14783 (shared): Negate static-pie.
14784 (-no-pie): Update help text.
14786 (static-pie): New option.
14787 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14788 -static-pie support.
14789 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14790 (LINK_EH_SPEC): Likewise.
14791 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14792 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14793 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14794 * gcc.c (LINK_COMMAND_SPEC): Likewise.
14795 (init_gcc_specs): Likewise.
14796 (init_spec): Likewise.
14797 (display_help): Update help message for -pie.
14798 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
14801 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
14803 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
14804 (movdi_aarch64): Likewise.
14805 (movti_aarch64): Likewise.
14807 2017-09-12 Simon Wright <simon@pushface.org>
14810 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
14811 calculation of the minor version, always output as 0.
14813 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14816 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14817 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
14818 on it early, rather than manual conversion late. For
14819 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
14820 instead of performing manual conversion.
14822 2017-09-12 Carl Love <cel@us.ibm.com>
14824 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
14825 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
14827 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14828 VMULOSW): Add definitions.
14829 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14830 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14831 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14832 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14833 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14835 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14837 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
14839 (movti_aarch64): Likewise.
14840 (movdf_aarch64): Likewise.
14841 (movtf_aarch64): Likewise.
14842 (load_pairdi): Likewise.
14843 (store_pairdi): Likewise.
14844 (load_pairdf): Likewise.
14845 (store_pairdf): Likewise.
14846 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
14847 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
14848 (ldr_got_small_<mode>): Likewise.
14849 (ldr_got_small_28k_<mode>): Likewise.
14850 (ldr_got_tiny): Likewise.
14851 * config/aarch64/iterators.md (ldst_sz): New.
14852 (ldpstp_sz): Likewise.
14853 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
14855 (thunderx_load): Split load_8 to load_16.
14856 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
14858 (thunderx2t99_storepair_basic): Split store_8 to store_16.
14859 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
14860 (xgene1_store_pair): Split store_8 to store_16.
14861 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
14862 (falkor_st_0_st_sd): Split store_8 to store_16.
14864 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14866 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
14867 and store1/2/3/4 to store_4/8/12/16.
14868 * config/aarch64/aarch64.md: Update for rename.
14869 * config/arm/arm.md: Likewise.: Likewise.
14870 * config/arm/arm.c: Likewise.
14871 * config/arm/thumb1.md: Likewise.
14872 * config/arm/thumb2.md: Likewise.
14873 * config/arm/vfp.md: Likewise.
14874 * config/arm/arm-generic.md: Likewise.
14875 * config/arm/arm1020e.md: Likewise.
14876 * config/arm/arm1026ejs.md: Likewise.
14877 * config/arm/arm1136jfs.md: Likewise.
14878 * config/arm/arm926ejs.md: Likewise.
14879 * config/arm/cortex-a15.md: Likewise.
14880 * config/arm/cortex-a17.md: Likewise.
14881 * config/arm/cortex-a5.md: Likewise.
14882 * config/arm/cortex-a53.md: Likewise.
14883 * config/arm/cortex-a57.md: Likewise.
14884 * config/arm/cortex-a7.md: Likewise.
14885 * config/arm/cortex-a8.md: Likewise.
14886 * config/arm/cortex-a9.md: Likewise.
14887 * config/arm/cortex-m4.md: Likewise.
14888 * config/arm/cortex-m7.md: Likewise.
14889 * config/arm/cortex-r4.md: Likewise.
14890 * config/arm/exynos-m1.md: Likewise.
14891 * config/arm/fa526.md: Likewise.
14892 * config/arm/fa606te.md: Likewise.
14893 * config/arm/fa626te.md: Likewise.
14894 * config/arm/fa726te.md: Likewise.
14895 * config/arm/fmp626.md: Likewise.
14896 * config/arm/iwmmxt.md: Likewise.
14897 * config/arm/ldmstm.md: Likewise.
14898 * config/arm/marvell-pj4.md: Likewise.
14899 * config/arm/xgene1.md: Likewise.
14900 * config/aarch64/thunderx.md: Likewise.
14901 * config/aarch64/thunderx2t99.md: Likewise.
14902 * config/aarch64/falkor.md: Likewise.
14904 2017-09-12 Martin Liska <mliska@suse.cz>
14906 * attribs.c (private_lookup_attribute): New function.
14907 * attribs.h (private_lookup_attribute): Declared here.
14908 (lookup_attribute): Called from this place.
14910 2017-09-12 Richard Biener <rguenther@suse.de>
14912 PR tree-optimization/82157
14913 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
14914 stmts with side-effects.
14916 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14917 Alan Hayward <alan.hayward@arm.com>
14918 David Sherwood <david.sherwood@arm.com>
14920 * target.def (hard_regno_nregs): New hook.
14921 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
14922 * targhooks.h (default_hard_regno_nregs): Declare.
14923 * targhooks.c (default_hard_regno_nregs): New function.
14924 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
14925 (TARGET_HARD_REGNO_NREGS): ...this hook.
14926 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
14927 (CLASS_MAX_NREGS): Likewise.
14928 * doc/tm.texi: Regenerate.
14929 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
14930 instead of HARD_REGNO_NREGS.
14931 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
14932 HARD_REGNO_NREGS in the comment.
14933 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
14934 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
14935 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
14936 Return an unsigned int.
14937 (TARGET_HARD_REGNO_NREGS): Redefine.
14938 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
14939 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
14940 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14941 (arc_hard_regno_nregs): New function.
14942 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
14943 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
14944 (arm_hard_regno_nregs): New function.
14945 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
14946 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
14947 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
14948 (TARGET_HARD_REGNO_NREGS): Redefine.
14949 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
14950 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
14951 (HARD_REGNO_NREGS): Delete.
14952 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
14953 (cr16_hard_regno_nregs): New function.
14954 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
14955 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
14956 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
14957 (cris_hard_regno_nregs): New function.
14958 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
14959 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
14960 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14961 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
14962 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
14963 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
14964 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
14965 (frv_hard_regno_nregs): Make static. Take and return an
14967 (frv_class_max_nregs): Remove outdated copy of documentation.
14968 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
14969 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
14970 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
14971 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
14972 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
14973 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
14974 (TARGET_HARD_REGNO_NREGS): Redefine.
14975 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
14976 (CLASS_MAX_NREGS): Update comment.
14977 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
14978 (ia64_hard_regno_nregs): New function.
14979 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
14980 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
14981 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
14982 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
14983 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
14985 (m32c_hard_regno_nregs): Likewise. Make static.
14986 (TARGET_HARD_REGNO_NREGS): Redefine.
14987 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
14988 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
14989 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
14990 (m68k_hard_regno_nregs): New function.
14991 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
14992 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
14993 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
14994 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
14995 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
14996 Take and return an unsigned int.
14997 (TARGET_HARD_REGNO_NREGS): Redefine.
14998 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
14999 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15000 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15001 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15002 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15003 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15004 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15005 (msp430_hard_regno_nregs): Make static. Take and return an
15007 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15008 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15009 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15010 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15011 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15012 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15013 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15014 (TARGET_HARD_REGNO_NREGS): Redefine.
15015 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15016 (PA_HARD_REGNO_NREGS): ...this.
15017 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15018 (PA_HARD_REGNO_NREGS): ...this.
15019 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15020 (pa_hard_regno_nregs): New function.
15021 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15022 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15023 (pdp11_hard_regno_nregs): New function.
15024 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15025 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15026 (rs6000_hard_regno_nregs_hook): New function.
15027 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15028 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15029 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15030 Take and return an unsigned int. Move earlier in file.
15031 (TARGET_HARD_REGNO_NREGS): Redefine.
15032 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15033 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15034 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15035 (rl78_hard_regno_nregs): Make static. Take and return an
15037 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15038 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15039 (rs6000_hard_regno_nregs_hook): New function.
15040 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15041 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15042 (TARGET_HARD_REGNO_NREGS): Redefine.
15043 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15044 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15045 instead of HARD_REGNO_NREGS.
15046 (s390_hard_regno_nregs): New function.
15047 (s390_hard_regno_mode_ok): Add comment from s390.h.
15048 (TARGET_HARD_REGNO_NREGS): Redefine.
15049 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15050 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15051 (sh_hard_regno_nregs): New function.
15052 (sh_pass_in_reg_p): Use it.
15053 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15054 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15055 (sparc_hard_regno_nregs): New function.
15056 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15057 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15058 (spu_function_arg_advance): Use it, supplying a valid register number.
15059 (TARGET_HARD_REGNO_NREGS): Redefine.
15060 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15061 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15062 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15063 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15064 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15065 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15066 (CLASS_MAX_NREGS): Remove copy of old documentation.
15067 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15068 (visium_hard_regno_nregs): New function.
15069 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15070 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15071 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15072 xtensa_hard_regno_nregs): New function.
15073 * system.h (HARD_REGNO_NREGS): Poison.
15075 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15077 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15078 hard_regno_nregs instead of HARD_REGNO_NREGS.
15079 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15080 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15081 (c6x_expand_epilogue): Likewise.
15082 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15083 (frv_read_iacc_argument): Likewise.
15084 * config/sh/sh.c: Include regs.h.
15085 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15086 (regs_used): Likewise.
15087 (output_stack_adjust): Likewise.
15088 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15089 * expmed.c: Include regs.h.
15090 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15091 * ree.c: Include regs.h.
15092 (combine_reaching_defs): Use hard_regno_nregs instead of
15094 (add_removable_extension): Likewise.
15096 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15098 * regs.h (hard_regno_nregs): Turn into a function.
15099 (end_hard_regno): Update accordingly.
15100 * caller-save.c (setup_save_areas): Likewise.
15101 (save_call_clobbered_regs): Likewise.
15102 (replace_reg_with_saved_mem): Likewise.
15103 (insert_restore): Likewise.
15104 (insert_save): Likewise.
15105 * combine.c (can_change_dest_mode): Likewise.
15106 (move_deaths): Likewise.
15107 (distribute_notes): Likewise.
15108 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15109 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15110 (rs6000_split_multireg_move): Likewise.
15111 (rs6000_register_move_cost): Likewise.
15112 (rs6000_memory_move_cost): Likewise.
15113 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15114 (rs6000_split_multireg_move): Likewise.
15115 (rs6000_register_move_cost): Likewise.
15116 (rs6000_memory_move_cost): Likewise.
15117 * cselib.c (cselib_reset_table): Likewise.
15118 (cselib_lookup_1): Likewise.
15119 * emit-rtl.c (set_mode_and_regno): Likewise.
15120 * function.c (aggregate_value_p): Likewise.
15121 * ira-color.c (setup_profitable_hard_regs): Likewise.
15122 (check_hard_reg_p): Likewise.
15123 (calculate_saved_nregs): Likewise.
15124 (assign_hard_reg): Likewise.
15125 (improve_allocation): Likewise.
15126 (calculate_spill_cost): Likewise.
15127 * ira-emit.c (modify_move_list): Likewise.
15128 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15129 (ira_hard_reg_in_set_p): Likewise.
15130 * ira.c (setup_reg_mode_hard_regset): Likewise.
15131 (clarify_prohibited_class_mode_regs): Likewise.
15132 (check_allocation): Likewise.
15133 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15134 (lra_setup_reg_renumber): Likewise.
15135 (setup_try_hard_regno_pseudos): Likewise.
15136 (spill_for): Likewise.
15137 (assign_hard_regno): Likewise.
15138 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15139 * lra-constraints.c (in_class_p): Likewise.
15140 (lra_constraint_offset): Likewise.
15141 (simplify_operand_subreg): Likewise.
15142 (lra_constraints): Likewise.
15143 (split_reg): Likewise.
15144 (split_if_necessary): Likewise.
15145 (invariant_p): Likewise.
15146 (inherit_in_ebb): Likewise.
15147 * lra-lives.c (process_bb_lives): Likewise.
15148 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15149 (get_hard_regs): Likewise.
15150 (do_remat): Likewise.
15151 * lra-spills.c (assign_spill_hard_regs): Likewise.
15152 * mode-switching.c (create_pre_exit): Likewise.
15153 * postreload.c (reload_combine_recognize_pattern): Likewise.
15154 * recog.c (peep2_find_free_register): Likewise.
15155 * regcprop.c (kill_value_regno): Likewise.
15156 (set_value_regno): Likewise.
15157 (copy_value): Likewise.
15158 (maybe_mode_change): Likewise.
15159 (find_oldest_value_reg): Likewise.
15160 (copyprop_hardreg_forward_1): Likewise.
15161 * regrename.c (check_new_reg_p): Likewise.
15162 (regrename_do_replace): Likewise.
15163 * reload.c (push_reload): Likewise.
15164 (combine_reloads): Likewise.
15165 (find_dummy_reload): Likewise.
15166 (operands_match_p): Likewise.
15167 (find_reloads): Likewise.
15168 (find_equiv_reg): Likewise.
15169 (reload_adjust_reg_for_mode): Likewise.
15170 * reload1.c (count_pseudo): Likewise.
15171 (count_spilled_pseudo): Likewise.
15172 (find_reg): Likewise.
15173 (clear_reload_reg_in_use): Likewise.
15174 (free_for_value_p): Likewise.
15175 (allocate_reload_reg): Likewise.
15176 (choose_reload_regs): Likewise.
15177 (reload_adjust_reg_for_temp): Likewise.
15178 (emit_reload_insns): Likewise.
15179 (delete_output_reload): Likewise.
15180 * rtlanal.c (subreg_get_info): Likewise.
15181 * sched-deps.c (sched_analyze_reg): Likewise.
15182 * sel-sched.c (init_regs_for_mode): Likewise.
15183 (mark_unavailable_hard_regs): Likewise.
15184 (choose_best_reg_1): Likewise.
15185 (verify_target_availability): Likewise.
15186 * valtrack.c (dead_debug_insert_temp): Likewise.
15187 * var-tracking.c (track_loc_p): Likewise.
15188 (emit_note_insn_var_location): Likewise.
15189 * varasm.c (make_decl_rtl): Likewise.
15190 * reginfo.c (choose_hard_reg_mode): Likewise.
15191 (init_reg_modes_target): Refer directly to
15192 this_target_regs->x_hard_regno_nregs.
15194 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15196 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15197 instead of hard_regno_nregs.
15199 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15201 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15202 end_hard_regno instead of hard_regno_nregs.
15203 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15204 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15205 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15206 * ira-color.c (improve_allocation): Likewise.
15207 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15208 * lra-lives.c (mark_regno_live): Likewise.
15209 (mark_regno_dead): Likewise.
15210 * lra-remat.c (operand_to_remat): Likewise.
15211 * lra.c (collect_non_operand_hard_regs): Likewise.
15212 * postreload.c (reload_combine_note_store): Likewise.
15213 (move2add_valid_value_p): Likewise.
15214 * reload.c (regno_clobbered_p): Likewise.
15216 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15218 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15220 * config/v850/v850.c (v850_reorg): Likewise.
15221 * reload.c (refers_to_regno_for_reload_p): Likewise.
15222 (find_equiv_reg): Likewise.
15223 * reload1.c (reload_reg_reaches_end_p): Likewise.
15225 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15227 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15229 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15230 * config/arm/arm.c (output_move_neon): Likewise.
15231 (arm_attr_length_move_neon): Likewise.
15232 (neon_split_vcombine): Likewise.
15233 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15234 (c6x_mark_reg_written): Likewise.
15235 (c6x_dwarf_register_span): Likewise.
15236 * config/i386/i386.c (ix86_save_reg): Likewise.
15237 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15238 (rws_access_reg): Likewise.
15239 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15240 * mode-switching.c (create_pre_exit): Likewise.
15241 * ree.c (combine_reaching_defs): Likewise.
15242 (add_removable_extension): Likewise.
15243 * regcprop.c (find_oldest_value_reg): Likewise.
15244 (copyprop_hardreg_forward_1): Likewise.
15245 * reload.c (reload_inner_reg_of_subreg): Likewise.
15246 (push_reload): Likewise.
15247 (combine_reloads): Likewise.
15248 (find_dummy_reload): Likewise.
15249 (reload_adjust_reg_for_mode): Likewise.
15250 * reload1.c (find_reload_regs): Likewise.
15251 (forget_old_reloads_1): Likewise.
15252 (reload_reg_free_for_value_p): Likewise.
15253 (reload_adjust_reg_for_temp): Likewise.
15254 (emit_reload_insns): Likewise.
15255 (delete_output_reload): Likewise.
15256 * sel-sched.c (choose_best_reg_1): Likewise.
15257 (choose_best_pseudo_reg): Likewise.
15259 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15260 Alan Hayward <alan.hayward@arm.com>
15261 David Sherwood <david.sherwood@arm.com>
15263 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15264 * target.def (slow_unaligned_access): New hook.
15265 * targhooks.h (default_slow_unaligned_access): Declare.
15266 * targhooks.c (default_slow_unaligned_access): New function.
15267 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15268 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15269 * doc/tm.texi: Regenerate.
15270 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15271 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15272 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15274 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15275 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15277 (rs6000_slow_unaligned_access): New function.
15278 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15279 (expand_block_compare): Likewise.
15280 (expand_strn_compare): Likewise.
15281 (rs6000_rtx_costs): Likewise.
15282 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15283 (riscv_slow_unaligned_access): Likewise.
15284 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15285 (riscv_slow_unaligned_access_p): ...this and make static.
15286 (riscv_option_override): Update accordingly.
15287 (riscv_slow_unaligned_access): New function.
15288 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15289 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15290 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15291 (rs6000_slow_unaligned_access): New function.
15292 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15293 (rs6000_rtx_costs): Likewise.
15294 * config/rs6000/rs6000-string.c (expand_block_compare)
15295 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15296 of SLOW_UNALIGNED_ACCESS.
15297 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15298 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15299 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15300 of SLOW_UNALIGNED_ACCESS.
15301 * expmed.c (simple_mem_bitfield_p): Likewise.
15302 * expr.c (alignment_for_piecewise_move): Likewise.
15303 (emit_group_load_1): Likewise.
15304 (emit_group_store): Likewise.
15305 (copy_blkmode_from_reg): Likewise.
15306 (emit_push_insn): Likewise.
15307 (expand_assignment): Likewise.
15308 (store_field): Likewise.
15309 (expand_expr_real_1): Likewise.
15310 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15311 * lra-constraints.c (simplify_operand_subreg): Likewise.
15312 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15313 * gimple-ssa-store-merging.c: Likewise in block comment at start
15315 * tree-ssa-strlen.c: Include target.h.
15316 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15317 of SLOW_UNALIGNED_ACCESS.
15318 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15320 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15322 PR rtl-optimization/82185
15323 * expmed.c (emit_store_flag_int): Only test tem if it has been
15326 2017-09-12 Richard Biener <rguenther@suse.de>
15328 PR middle-end/82149
15329 * match.pd ((FTYPE) N CMP CST): Fix typo.
15331 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15333 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15335 (mips_near_type_p): Add 'short_call' attribute as a synonym
15337 * doc/extend.texi (short_call): Document new function attribute.
15339 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15342 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15343 assertion check that in the condition.
15344 (get_atomic_generic_size): Likewise. Before testing if parameter
15345 has pointer type, if it has array type, call for C++
15346 default_conversion to perform array-to-pointer conversion.
15348 2017-09-12 Richard Biener <rguenther@suse.de>
15350 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15351 for operations we cannot scalarize.
15353 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15355 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15356 vectors heap vectors. Clean up comments.
15357 Make visited_bbs a reference.
15358 (profitable_jump_thread_path): Make GC
15359 vectors heap vectors. Clean up comments.
15361 (convert_and_register_jump_thread_path): Make GC vectors heap
15363 (check_subpath_and_update_thread_path): Same. Clean up comments.
15364 Make visited_bbs a reference.
15365 (handle_phi): Abstract common code to to
15366 register_jump_thread_path_if_profitable.
15367 Rename VAR_BB to DEF_BB.
15369 Make GC vectors heap vectors.
15370 Make visited_bbs a reference.
15371 (handle_assignment): Same.
15372 (register_jump_thread_path_if_profitable): New.
15373 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15375 Make GC vectors heap vectors. Clean up comments.
15376 Make visited_bbs a reference.
15377 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15378 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15381 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15384 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15385 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15387 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15389 Revert r251800 and r251799.
15391 2017-09-11 Martin Jambor <mjambor@suse.cz>
15394 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15395 arguments in advance.
15396 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15397 use it to find predecessor edges.
15398 (naive_outof_ssa): Collect vector of predecessors.
15400 2017-09-08 Jason Merrill <jason@redhat.com>
15402 PR c++/70029 - ICE with ref-qualifier and -flto
15403 * langhooks.h (struct lang_hooks_for_types): Add
15404 copy_lang_qualifiers.
15405 * attribs.c (build_type_attribute_qual_variant): Use it.
15406 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15408 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15409 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15411 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15414 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15415 (*mulsi3_sp64): New instruction.
15416 (mulsi3): New expander.
15418 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15420 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15422 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15424 * sancov.c: Include memmodel.h.
15426 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15429 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15432 2017-09-07 Carl Love <cel@us.ibm.com>
15434 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15435 the sldi instruction.
15437 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15439 * sancov.c: Include tm_p.h.
15441 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15444 * output.h (switch_to_other_text_partition): New declaration.
15445 * varasm.c (switch_to_other_text_partition): New function.
15446 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15447 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15448 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15449 to the other text partition before emitting LCL label and switch back
15450 after emitting the word after it.
15452 2017-09-07 Richard Biener <rguenther@suse.de>
15454 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15455 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15456 critical edges here, after loop init.
15457 (pass_data_pre): Remove PROP_no_crit_edges flags.
15458 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15459 for valueization of call args to avoid leaking VN_TOP.
15460 (visit_use): Assert we do not visit default defs.
15461 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15462 Use error_mark_node to more easily detect leaking VN_TOP.
15463 All default-defs are varying, not VN_TOP. Mark them visited.
15464 (run_scc_vn): Make code match comment.
15466 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15468 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15469 OPTION_MASK_FLOAT128_KEYWORD.
15470 (POWERPC_MASKS): Likewise.
15471 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15472 support for the -mfloat128-type option, and make -mfloat128
15473 default on PowerPC Linux systems. Define or undefine
15474 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15475 Define __float128 to be __ieee128 if IEEE 128-bit support is
15476 enabled, or undefine it.
15477 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15478 Delete defining __FLOAT128_TYPE__.
15479 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15480 -mfloat128-type option and make -mfloat128 default on PowerPC
15482 (TARGET_FLOAT128_TYPE): Likewise.
15483 (-mfloat128-type): Likewise.
15484 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15485 Delete the -mfloat128-type option and make -mfloat128 default on
15486 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15487 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15488 128-bit floating point is enabled. Change tests from using
15489 -mfloat128-type to -mfloat128.
15490 (rs6000_mangle_type): Use the correct mangling for the __float128
15491 type even if normal long double is restricted to 64-bits.
15492 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15494 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15495 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15496 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15498 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15499 documentation for -mfloat128.
15501 2017-09-06 Olivier Hainque <hainque@adacore.com>
15503 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15505 2017-09-06 Wish Wu <wishwu007@gmail.com>
15506 Jakub Jelinek <jakub@redhat.com>
15508 * asan.c (initialize_sanitizer_builtins): Add
15509 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15510 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15511 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15512 BT_FN_VOID_UINT64_PTR variables.
15513 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15514 (BT_FN_VOID_UINT16_UINT16): Likewise.
15515 (BT_FN_VOID_UINT32_UINT32): Likewise.
15516 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15517 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15518 (BT_FN_VOID_UINT64_PTR): Likewise.
15519 * common.opt (flag_sanitize_coverage): New variable.
15520 (fsanitize-coverage=trace-pc): Remove.
15521 (fsanitize-coverage=): Add.
15522 * flag-types.h (enum sanitize_coverage_code): New enum.
15523 * fold-const.c (fold_range_test): Disable non-short-circuit
15524 optimization if flag_sanitize_coverage.
15525 (fold_truth_andor): Likewise.
15526 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15527 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15528 (coverage_sanitizer_opts): New array.
15529 (get_closest_sanitizer_option): Add OPTS argument, handle also
15530 OPT_fsanitize_coverage_.
15531 (parse_sanitizer_options): Adjusted to also handle
15532 OPT_fsanitize_coverage_.
15533 (common_handle_option): Add OPT_fsanitize_coverage_.
15534 * sancov.c (instrument_comparison, instrument_switch): New function.
15535 (sancov_pass): Add trace-cmp support.
15536 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15537 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15538 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15539 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15540 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15541 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15542 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15543 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15544 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15545 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15547 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15549 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15550 error. Only quit immediately if parsing is complete.
15551 (BEGIN): Initialize fatal_err and parse_done.
15552 (begin fpu, end fpu): Check number of arguments.
15553 (begin arch, end arch): Likewise.
15554 (begin cpu, end cpu): Likewise.
15555 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15556 (optalias): Likewise.
15558 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15560 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15561 * config/arm/arm-isa.h: Delete. Move definitions to ...
15562 * arm-cpus.in: ... here. Use new feature and fgroup values.
15563 * config/arm/arm.c (arm_option_override): Use lower case for feature
15565 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15566 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15567 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15569 (print_isa_bits_for): New function.
15570 (gen_isa): New function.
15571 (gen_comm_data): Use print_isa_bits_for.
15572 (define feature): New keyword.
15573 (define fgroup): New keyword.
15574 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15575 (arm-isa.h): Add rule to generate file.
15576 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15577 case for feature bit names.
15579 2017-09-06 Richard Biener <rguenther@suse.de>
15581 * tree-ssa-pre.c (NECESSARY): Remove.
15582 (create_expression_by_pieces): Do not touch pass-local flags.
15583 (insert_into_preds_of_block): Likewise.
15584 (do_pre_regular_insertion): Likewise.
15585 (eliminate_insert): Likewise.
15586 (eliminate_dom_walker::before_dom_children): Likewise.
15587 (fini_eliminate): Do not look at inserted_exprs.
15588 (mark_operand_necessary): Remove.
15589 (remove_dead_inserted_code): Replace with simple work-list
15590 algorithm based on inserted_exprs and SSA uses.
15591 (pass_pre::execute): Re-order fini_eliminate and
15592 remove_dead_inserted_code.
15594 2017-09-06 Olivier Hainque <hainque@adacore.com>
15596 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15597 for VxWorks 7. Adjust surrounding comments.
15599 2017-09-06 Richard Biener <rguenther@suse.de>
15601 * gimple-ssa-strength-reduction.c
15602 (find_candidates_dom_walker::before_dom_children): Also allow
15605 2017-09-06 Richard Biener <rguenther@suse.de>
15607 PR tree-optimization/82108
15608 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15609 for gap in the non-permutation SLP case.
15611 2017-09-06 Martin Jambor <mjambor@suse.cz>
15613 PR tree-optimization/82078
15614 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15615 add_access_to_work_queue...
15616 (build_accesses_from_assign): ...here.
15617 (propagate_all_subaccesses): Make sure racc is the group
15618 representative, if there is one.
15620 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15622 PR middle-end/82095
15623 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15626 2017-09-06 Richard Biener <rguenther@suse.de>
15628 * gimple-ssa-strength-reduction.c
15629 (find_candidates_dom_walker::before_doom_children): Use a
15630 type and not a mode check.
15632 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15635 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15637 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15639 2017-09-05 Jeff Law <law@redhat.com>
15641 PR tree-optimization/64910
15642 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15643 cases where we have 3 or more operands.
15645 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15647 PR middle-end/81768
15648 * omp-low.c (lower_omp_for): Recompute tree invariant if
15649 gimple_omp_for_initial/final is ADDR_EXPR.
15651 PR middle-end/81768
15652 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15653 into gimple val before gimplification fo the COND_EXPR.
15655 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15657 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15658 REGION_COPY argument.
15659 (thread_through_all_blocks): Remove unused argument to
15660 duplicate_thread_path.
15662 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15663 Alan Hayward <alan.hayward@arm.com>
15664 David Sherwood <david.sherwood@arm.com>
15666 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15667 Take a scalar_mode rather than a machine_mode.
15668 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15669 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15670 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15671 (aarch64_gen_adjusted_ldpstp): Likewise.
15672 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15674 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15675 Alan Hayward <alan.hayward@arm.com>
15676 David Sherwood <david.sherwood@arm.com>
15678 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15679 Take a scalar_int_mode instead of a machine_mode.
15680 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15681 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15682 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15683 (aarch64_simd_attr_length_rglist): Delete.
15684 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15685 a scalar_int_mode instead of a machine_mode.
15686 (aarch64_add_offset): Likewise.
15687 (aarch64_internal_mov_immediate): Likewise
15688 (aarch64_add_constant_internal): Likewise.
15689 (aarch64_add_constant): Likewise.
15690 (aarch64_movw_imm): Likewise.
15691 (aarch64_rtx_arith_op_extract_p): Likewise.
15692 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15693 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15694 Remove assert that the mode isn't a vector.
15695 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15696 (aarch64_expand_mov_immediate): Update calls after above changes.
15697 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15698 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15699 (aarch64_move_imm): Likewise.
15700 (aarch64_can_const_movi_rtx_p): Likewise.
15701 (aarch64_strip_extend): Likewise.
15702 (aarch64_extr_rtx_p): Likewise.
15703 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15704 a CONST_INT when the mode parameter is VOIDmode.
15705 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15707 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15709 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15710 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15711 (bitwise_type_for_mode): Update accordingly.
15713 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15715 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15716 * stor-layout.c (mode_for_size_tree): Likewise.
15717 (mode_for_array): Update accordingly.
15718 (layout_decl): Likewise.
15719 (compute_record_mode): Likewise. Only set the mode once.
15721 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15723 * target.def (get_mask_mode): Change return type to opt_mode.
15725 * doc/tm.texi: Regenerate.
15726 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15727 * targhooks.c (default_get_mask_mode): Likewise.
15728 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15729 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15730 targetm.get_mask_mode.
15731 * tree.c (build_truth_vector_type): Likewise.
15733 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15735 * machmode.h (mode_for_vector): Return an opt_mode.
15736 * stor-layout.c (mode_for_vector): Likewise.
15737 (mode_for_int_vector): Update accordingly.
15738 (layout_type): Likewise.
15739 * config/i386/i386.c (emit_memmov): Likewise.
15740 (ix86_expand_set_or_movmem): Likewise.
15741 (ix86_expand_vector_init): Likewise.
15742 (ix86_get_mask_mode): Likewise.
15743 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15745 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15746 * expmed.c (extract_bit_field_1): Likewise.
15747 * expr.c (expand_expr_real_2): Likewise.
15748 * optabs-query.c (can_vec_perm_p): Likewise.
15749 (can_vec_mask_load_store_p): Likewise.
15750 * optabs.c (expand_vec_perm): Likewise.
15751 * targhooks.c (default_get_mask_mode): Likewise.
15752 * tree-vect-stmts.c (vectorizable_store): Likewise.
15753 (vectorizable_load): Likewise.
15754 (get_vectype_for_scalar_type_and_size): Likewise.
15756 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15758 * machmode.h (mode_for_int_vector): New function.
15759 * stor-layout.c (mode_for_int_vector): Likewise.
15760 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15761 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15762 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15763 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15764 (s390_expand_vcond): Likewise.
15766 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15768 * machmode.h (opt_machine_mode): New type.
15769 (opt_mode<T>): Allow construction from anything that can be
15771 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15772 (mode_for_size): Return an opt_machine_mode.
15773 * stor-layout.c (mode_for_size): Likewise.
15774 (mode_for_size_tree): Update call accordingly.
15775 (bitwise_mode_for_mode): Likewise.
15776 (make_fract_type): Likewise.
15777 (make_accum_type): Likewise.
15778 * caller-save.c (replace_reg_with_saved_mem): Update call
15780 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15781 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15782 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15783 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15784 * expmed.c (extract_bit_field_1): Likewise.
15785 * reload.c (get_secondary_mem): Likewise.
15786 * varasm.c (assemble_integer): Likewise.
15787 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15790 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15792 * machmode.h (decimal_float_mode_for_size): New function.
15793 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
15794 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
15795 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
15796 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
15797 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
15798 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
15800 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15802 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
15803 (get_builtin_sync_mode): Likewise.
15804 (expand_ifn_atomic_compare_exchange): Likewise.
15805 (expand_builtin_atomic_clear): Likewise.
15806 (expand_builtin_atomic_test_and_set): Likewise.
15807 (fold_builtin_atomic_always_lock_free): Likewise.
15808 * calls.c (compute_argument_addresses): Likewise.
15809 (emit_library_call_value_1): Likewise.
15810 (store_one_arg): Likewise.
15811 * combine.c (combine_instructions): Likewise.
15812 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
15813 * config/arm/arm.c (arm_function_value): Likewise.
15814 (aapcs_allocate_return_reg): Likewise.
15815 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
15816 * config/i386/i386.c (construct_container): Likewise.
15817 (ix86_gimplify_va_arg): Likewise.
15818 (ix86_expand_sse_cmp): Likewise.
15819 (emit_memmov): Likewise.
15820 (emit_memset): Likewise.
15821 (expand_small_movmem_or_setmem): Likewise.
15822 (ix86_expand_pextr): Likewise.
15823 (ix86_expand_pinsr): Likewise.
15824 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
15825 * config/microblaze/microblaze.c (microblaze_block_move_straight):
15827 * config/mips/mips.c (mips_function_value_1) Likewise.
15828 (mips_block_move_straight): Likewise.
15829 (mips_expand_ins_as_unaligned_store): Likewise.
15830 * config/powerpcspe/powerpcspe.c
15831 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15832 (rs6000_darwin64_record_arg_flush): Likewise.
15833 * config/rs6000/rs6000.c
15834 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15835 (rs6000_darwin64_record_arg_flush): Likewise.
15836 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
15837 (sparc_function_value_1): Likewise.
15838 * config/spu/spu.c (adjust_operand): Likewise.
15839 (spu_emit_branch_or_set): Likewise.
15840 (arith_immediate_p): Likewise.
15841 * emit-rtl.c (gen_lowpart_common): Likewise.
15842 * expr.c (expand_expr_real_1): Likewise.
15843 * function.c (assign_parm_setup_block): Likewise.
15844 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
15845 * reload1.c (alter_reg): Likewise.
15846 * stor-layout.c (mode_for_vector): Likewise.
15847 (layout_type): Likewise.
15849 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15851 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
15852 (spu_convert_move): Likewise.
15853 * lower-subreg.c (resolve_simple_move): Likewise.
15855 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15858 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
15859 define_insn to a define_expand.
15860 (altivec_vsum2sws_direct): New define_insn.
15861 (altivec_vsumsws): Convert from a define_insn to a define_expand.
15863 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15865 * config/arm/arm.c (arm_option_params_internal): Improve setting of
15868 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
15873 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
15874 realign stack if stack alignment needed is less than incoming
15877 2017-09-05 Marek Polacek <polacek@redhat.com>
15880 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
15883 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15885 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
15887 2017-09-05 Richard Biener <rguenther@suse.de>
15889 PR tree-optimization/82084
15890 * fold-const.c (can_native_encode_string_p): Handle wide characters.
15892 2017-09-05 Richard Biener <rguenther@suse.de>
15894 PR tree-optimization/82102
15895 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
15897 2017-09-05 Martin Liska <mliska@suse.cz>
15899 PR tree-optimization/82032
15900 * tree-cfg.c (generate_range_test): New function.
15901 * tree-cfg.h (generate_range_test): Declared here.
15902 * tree-cfgcleanup.c (convert_single_case_switch): New function.
15903 (cleanup_control_expr_graph): Use it.
15904 * tree-switch-conversion.c (try_switch_expansion): Remove
15906 (emit_case_nodes): Use generate_range_test.
15908 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15911 * config/i386/i386.md (*<btsc><mode>_mask): Add
15912 TARGET_USE_BT to insn constraint.
15913 (*btr<mode>_mask): Ditto.
15915 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
15917 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
15918 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
15920 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
15923 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
15924 TARGET_NEON and TARGET_IWMMXT.
15925 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
15926 TARGET_NEON and TARGET_IWMMXT.
15927 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
15929 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15931 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
15932 (ix86_rewrite_tls_address): Ditto.
15933 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
15934 (ix86_rewrite_tls_address_1): Ditto.
15935 (ix86_rewrite_tls_address): Ditto.
15936 * config/i386/predicates.md (tls_address_pattern): New predicate.
15937 * config/i386/i386.md (TLS address splitter): New splitter.
15939 2017-09-04 Richard Biener <rguenther@suse.de>
15941 PR tree-optimization/82084
15942 * fold-const.h (can_native_encode_string_p): Declare.
15943 * fold-const.c (can_native_encode_string_p): Factor out from ...
15944 (native_encode_string): ... here.
15945 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
15946 vectorizing stores from constants we later cannot handle.
15948 2017-09-04 Marek Polacek <polacek@redhat.com>
15951 * doc/invoke.texi: Update -Wtautological-compare documentation.
15953 2017-09-04 Jeff Law <law@redhat.com>
15955 PR tree-optimization/64910
15956 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
15957 swap the first and last operand if the last is a constant.
15959 2017-09-04 Marek Polacek <polacek@redhat.com>
15962 * convert.c (do_narrow): When sanitizing signed integer overflows,
15963 bail out for signed types.
15964 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
15966 2017-09-04 Richard Biener <rguenther@suse.de>
15968 PR tree-optimization/82060
15969 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15970 Move devirtualization after stmt folding and before EH/AB/noreturn
15971 cleanup to get the stmt refs canonicalized. Use a bool instead
15972 of gimple_modified_p since that doesn't work for NOPs. Schedule
15973 NOPs generated by folding for removal.
15975 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15976 Alan Hayward <alan.hayward@arm.com>
15977 David Sherwood <david.sherwood@arm.com>
15979 * coretypes.h (pad_direction): New enum.
15980 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
15981 (FUNCTION_ARG_PADDING): Likewise.
15982 * target.def (function_arg_padding): New hook.
15983 * targhooks.h (default_function_arg_padding): Declare.
15984 * targhooks.c (default_function_arg_padding): New function.
15985 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
15986 (TARGET_FUNCTION_ARG_PADDING): ...this.
15987 * doc/tm.texi: Regenerate.
15988 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
15989 instead of direction.
15990 (compute_argument_addresses): Likewise.
15991 (load_register_parameters): Likewise.
15992 (emit_library_call_value_1): Likewise.
15993 (store_one_arg): Use targetm.calls.function_arg_padding instead
15994 of FUNCTION_ARG_PADDING.
15995 (must_pass_in_stack_var_size_or_pad): Likewise.
15996 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
15997 (emit_group_store): Likewise.
15998 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
15999 instead of FUNCTION_ARG_PADDING.
16000 (emit_push_insn): Likewise, and propagate enum change throughout
16002 * function.h (direction): Delete.
16003 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16005 * function.c (assign_parm_find_stack_rtl): Likewise.
16006 (assign_parm_setup_block_p): Likewise.
16007 (assign_parm_setup_block): Likewise.
16008 (gimplify_parameters): Likewise.
16009 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16010 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16012 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16013 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16014 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16015 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16016 (aarch64_function_arg_padding): ...this new function.
16017 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16018 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16019 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16020 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16021 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16022 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16023 (arm_pad_arg_upward): Replace with...
16024 (arm_function_arg_padding): ...this new function.
16025 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16027 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16028 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16029 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16030 (ia64_hpux_function_arg_padding): Replace with...
16031 (ia64_function_arg_padding): ...this new function. Use pad_direction
16032 instead of direction. Check for TARGET_HPUX.
16033 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16034 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16035 (iq2000_function_arg_padding): New function.
16036 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16037 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16038 (mips_function_arg_padding): ...this new function.
16039 (mips_pad_reg_upward): Update accordingly.
16040 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16041 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16042 targetm.calls.function_arg_padding.
16043 (FUNCTION_ARG_PADDING): Delete.
16044 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16045 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16046 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16047 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16048 (nios2_block_reg_padding): Return pad_direction instead of direction.
16049 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16050 instead of direction.
16051 (nios2_function_arg_padding): Likewise. Make static.
16052 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16053 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16054 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16055 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16056 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16057 (pa_function_arg_padding): Make static. Return pad_direction instead
16059 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16060 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16061 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16062 instead of direction. Use targetm.calls.function_arg_padding.
16063 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16064 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16065 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16066 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16067 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16069 (function_arg_padding): Rename to...
16070 (rs6000_function_arg_padding): ...this. Make static. Return
16071 pad_direction instead of direction.
16072 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16073 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16074 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16075 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16076 instead of direction. Use targetm.calls.function_arg_padding.
16077 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16078 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16079 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16080 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16081 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16082 (function_arg_padding): Rename to...
16083 (rs6000_function_arg_padding): ...this. Make static. Return
16084 pad_direction instead of direction.
16085 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16086 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16087 * config/s390/s390.c (s390_function_arg_padding): New function.
16088 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16089 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16090 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16091 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16092 (function_arg_padding): Rename to...
16093 (sparc_function_arg_padding): ...this. Make static. Return
16094 pad_direction instead of direction.
16095 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16096 * config/spu/spu.c (spu_function_arg_padding): New function.
16097 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16098 * system.h (FUNCTION_ARG_PADDING): Poison.
16100 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16101 Alan Hayward <alan.hayward@arm.com>
16102 David Sherwood <david.sherwood@arm.com>
16104 * target.def (modes_tieable_p): New hook.
16105 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16106 (TARGET_MODES_TIEABLE_P): ...this.
16107 * doc/tm.texi.in: Regenerate.
16108 * hooks.h (hook_bool_mode_mode_true): Declare.
16109 * hooks.c (hook_bool_mode_mode_true): New function.
16110 * combine.c (subst): Use targetm.modes_tieable_p instead of
16112 * dse.c (find_shift_sequence): Likewise.
16113 * expmed.c (extract_low_bits): Likewise.
16114 * lower-subreg.c: Include target.h.
16115 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16117 * rtlanal.c (rtx_cost): Likewise.
16118 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16119 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16120 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16121 (TARGET_MODES_TIEABLE_P): Redefine.
16122 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16123 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16124 (TARGET_MODES_TIEABLE_P): Redefine.
16125 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16126 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16127 (arc_modes_tieable_p): New function.
16128 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16129 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16130 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16131 (arm_modes_tieable_p): Make static.
16132 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16133 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16134 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16135 (TARGET_MODES_TIEABLE_P): Redefine.
16136 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16137 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16138 (TARGET_MODES_TIEABLE_P): Redefine.
16139 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16140 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16141 (cr16_modes_tieable_p): New function.
16142 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16143 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16144 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16145 (TRULY_NOOP_TRUNCATION): Update comment.
16146 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16147 (TRULY_NOOP_TRUNCATION): Update comment.
16148 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16149 (frv_modes_tieable_p): New function.
16150 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16151 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16152 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16153 (TARGET_MODES_TIEABLE_P): Redefine.
16154 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16155 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16156 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16157 (TARGET_MODES_TIEABLE_P): Redefine.
16158 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16159 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16160 (ia64_modes_tieable_p): New function.
16161 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16162 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16163 (iq2000_modes_tieable_p): New function.
16164 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16165 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16166 (lm32_modes_tieable_p): New function.
16167 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16168 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16169 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16170 (TARGET_MODES_TIEABLE_P): Redefine.
16171 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16172 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16173 (m32r_modes_tieable_p): New function.
16174 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16175 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16176 (m68k_modes_tieable_p): New function.
16177 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16178 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16179 (mcore_modes_tieable_p): New function.
16180 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16181 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16183 (TARGET_MODES_TIEABLE_P): Redefine.
16184 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16185 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16186 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16187 (TARGET_MODES_TIEABLE_P): Redefine.
16188 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16189 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16190 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16191 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16192 (mn10300_modes_tieable_p): ...this and make static.
16193 (TARGET_MODES_TIEABLE_P): Redefine.
16194 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16195 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16196 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16197 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16198 (msp430_modes_tieable_p): Make static.
16199 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16200 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16201 (TARGET_MODES_TIEABLE_P): Redefine.
16202 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16203 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16204 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16205 (TARGET_MODES_TIEABLE_P): Redefine.
16206 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16207 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16208 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16209 (TARGET_MODES_TIEABLE_P): Redefine.
16210 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16211 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16212 (pdp11_modes_tieable_p): New function.
16213 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16214 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16215 (rs6000_modes_tieable_p): New function.
16216 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16217 * config/powerpcspe/powerpcspe.md: Update comment.
16218 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16219 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16220 (TARGET_MODES_TIEABLE_P): Redefine.
16221 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16222 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16223 (rl78_modes_tieable_p): New function.
16224 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16225 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16226 (rs6000_modes_tieable_p): New function.
16227 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16228 * config/rs6000/rs6000.md: Update comment.
16229 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16230 * config/rx/rx.c (rx_modes_tieable_p): New function.
16231 (TARGET_MODES_TIEABLE_P): Redefine.
16232 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16233 * config/s390/s390.c (s390_modes_tieable_p): New function.
16234 (TARGET_MODES_TIEABLE_P): Redefine.
16235 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16236 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16237 (sh_modes_tieable_p): New function.
16238 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16239 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16240 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16241 (sparc_modes_tieable_p): Make static.
16242 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16243 * config/spu/spu.c (spu_modes_tieable_p): New function.
16244 (TARGET_MODES_TIEABLE_P): Redefine.
16245 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16246 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16247 (TARGET_MODES_TIEABLE_P): Redefine.
16248 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16249 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16250 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16251 * config/v850/v850.c (v850_modes_tieable_p): New function.
16252 (TARGET_MODES_TIEABLE_P): Redefine.
16253 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16254 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16255 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16256 (visium_modes_tieable_p): New function.
16257 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16258 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16259 (xtensa_modes_tieable_p): New function.
16260 * system.h (MODES_TIEABLE_P): Poison.
16262 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16263 Alan Hayward <alan.hayward@arm.com>
16264 David Sherwood <david.sherwood@arm.com>
16266 * target.def (hard_regno_mode_ok): New hook.
16267 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16268 (TARGET_HARD_REGNO_MODE_OK): ...this.
16269 * doc/tm.texi.in: Regenerate.
16270 * hooks.h (hook_bool_uint_mode_true): Declare.
16271 * hooks.c (hook_bool_uint_mode_true): New function.
16272 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16273 HARD_REGNO_MODE_OK.
16274 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16275 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16276 instead of HARD_REGNO_MODE_OK.
16277 * caller-save.c: Include target.h.
16278 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16279 HARD_REGNO_MODE_OK.
16280 * combine.c (can_combine_p): Likewise.
16281 (combinable_i3pat): Likewise.
16282 (can_change_dest_mode): Likewise.
16283 * expr.c (init_expr_target): Likewise.
16284 (convert_move): Likewise.
16285 (convert_modes): Likewise.
16286 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16287 (setup_prohibited_mode_move_regs): Likewise.
16288 * ira.h (target_ira): Likewise.
16289 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16290 * lra-constraints.c (process_alt_operands): Likewise.
16291 (split_reg): Likewise.
16292 * recog.c (peep2_find_free_register): Likewise.
16293 * ree.c (combine_reaching_defs): Likewise.
16294 * regcprop.c (maybe_mode_change): Likewise.
16295 * reginfo.c (init_reg_sets_1): Likewise.
16296 (choose_hard_reg_mode): Likewise.
16297 (simplifiable_subregs): Likewise.
16298 * regrename.c (check_new_reg_p): Likewise.
16299 * reload.c (find_valid_class): Likewise.
16300 (find_valid_class_1): Likewise.
16301 (reload_inner_reg_of_subreg): Likewise.
16302 (push_reload): Likewise.
16303 (combine_reloads): Likewise.
16304 (find_dummy_reload): Likewise.
16305 (find_reloads): Likewise.
16306 * reload1.c (find_reg): Likewise.
16307 (set_reload_reg): Likewise.
16308 (allocate_reload_reg): Likewise.
16309 (choose_reload_regs): Likewise.
16310 (reload_adjust_reg_for_temp): Likewise.
16311 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16312 (simplify_subreg_regno): Likewise.
16313 * sel-sched.c (init_regs_for_mode): Likewise.
16314 * varasm.c (make_decl_rtl): Likewise.
16315 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16316 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16317 HARD_REGNO_MODE_OK.
16318 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16319 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16320 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16321 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16322 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16323 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16324 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16325 (arc_mode_class): Delete.
16326 (HARD_REGNO_MODE_OK): Delete.
16327 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16328 (arc_hard_regno_mode_ok): Rename old array to...
16329 (arc_hard_regno_mode_ok_modes): ...this.
16330 (arc_conditional_register_usage): Update accordingly.
16331 (arc_mode_class): Make static.
16332 (arc_hard_regno_mode_ok): New function.
16333 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16334 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16335 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16336 (arm_hard_regno_mode_ok): Make static.
16337 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16338 HARD_REGNO_MODE_OK.
16339 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16340 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16341 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16343 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16344 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16345 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16346 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16347 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16348 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16349 * config/bfin/predicates.md (valid_reg_operand): Use
16350 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16351 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16352 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16353 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16354 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16355 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16356 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16357 (cr16_hard_regno_mode_ok): Make static and return a bool.
16358 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16359 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16360 (cris_hard_regno_mode_ok): New function.
16361 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16362 (epiphany_mode_class): Delete.
16363 (HARD_REGNO_MODE_OK): Delete.
16364 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16365 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16366 (hard_regno_mode_ok): Rename to...
16367 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16368 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16369 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16370 HARD_REGNO_MODE_OK.
16371 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16372 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16373 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16374 (frv_hard_regno_mode_ok): Make static and return a bool.
16375 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16376 HARD_REGNO_MODE_OK.
16377 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16378 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16379 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16380 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16382 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16383 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16384 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16385 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16387 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16388 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16389 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16390 (ia64_hard_regno_mode_ok): New function.
16391 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16392 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16393 (iq2000_hard_regno_mode_ok): New function.
16394 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16395 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16396 (lm32_hard_regno_mode_ok): New function.
16397 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16398 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16399 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16400 instead of HARD_REGNO_MODE_OK.
16401 (m32c_hard_regno_ok): Rename to...
16402 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16403 (m32c_cannot_change_mode_class): Update accordingly.
16404 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16405 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16406 (m32r_mode_class): Delete.
16407 (HARD_REGNO_MODE_OK): Delete.
16408 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16409 (m32r_hard_regno_mode_ok): Rename to...
16410 (m32r_hard_regno_modes): ...this.
16411 (m32r_mode_class): Make static.
16412 (m32r_hard_regno_mode_ok): New function.
16413 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16414 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16415 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16416 (m68k_hard_regno_mode_ok): Make static.
16417 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16418 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16419 (mcore_hard_regno_mode_ok): New function.
16420 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16421 (HARD_REGNO_MODE_OK): Delete.
16422 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16424 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16425 (microblaze_hard_regno_mode_ok): New function.
16426 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16427 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16428 (mips_hard_regno_mode_ok): Delete.
16429 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16430 (mips_hard_regno_mode_ok_p): ...this and make static.
16431 (mips_hard_regno_mode_ok_p): Rename to...
16432 (mips_hard_regno_mode_ok_uncached): ...this.
16433 (mips_hard_regno_mode_ok): New function.
16434 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16435 of HARD_REGNO_MODE_OK.
16436 (mips_option_override): Update after above name changes.
16437 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16438 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16439 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16440 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16441 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16442 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16443 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16444 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16445 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16446 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16447 (msp430_hard_regno_mode_ok): Make static and return a bool.
16448 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16449 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16450 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16452 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16453 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16454 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16455 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16456 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16457 (PA_HARD_REGNO_MODE_OK): ...this
16458 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16459 (PA_HARD_REGNO_MODE_OK): ...this.
16460 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16461 (pa_hard_regno_mode_ok): New function.
16462 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16463 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16464 (pdp11_hard_regno_mode_ok): New function.
16465 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16466 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16468 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16470 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16471 (rs6000_hard_regno_mode_ok): Rename to...
16472 (rs6000_hard_regno_mode_ok_uncached): ...this.
16473 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16474 (rs6000_hard_regno_mode_ok): New function.
16475 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16476 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16477 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16478 (riscv_hard_regno_mode_ok): ...this and make static.
16479 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16480 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16481 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16482 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16483 (rl78_hard_regno_mode_ok): Make static and return bool.
16484 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16485 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16487 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16488 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16489 (rs6000_hard_regno_mode_ok): Rename to...
16490 (rs6000_hard_regno_mode_ok_uncached): ...this.
16491 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16492 (rs6000_hard_regno_mode_ok): New function.
16493 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16494 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16495 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16496 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16497 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16498 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16499 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16500 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16501 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16502 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16503 (sh_hard_regno_mode_ok): Make static.
16504 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16505 instead of HARD_REGNO_MODE_OK.
16506 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16507 (sparc_mode_class): Delete.
16508 (HARD_REGNO_MODE_OK): Delete.
16509 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16510 (hard_regno_mode_classes): Make static.
16511 (sparc_mode_class): Likewise.
16512 (sparc_hard_regno_mode_ok): New function.
16513 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16514 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16515 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16517 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16518 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16519 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16520 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16521 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16522 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16523 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16524 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16525 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16526 (visium_hard_regno_mode_ok): New function.
16527 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16528 instead of HARD_REGNO_MODE_OK.
16529 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16530 (HARD_REGNO_MODE_OK): Delete.
16531 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16532 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16533 (xtensa_option_override): Update accordingly.
16534 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16535 (xtensa_hard_regno_mode_ok): New function.
16536 * system.h (HARD_REGNO_MODE_OK): Poison.
16538 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16539 Alan Hayward <alan.hayward@arm.com>
16540 David Sherwood <david.sherwood@arm.com>
16542 * target.def (hard_regno_call_part_clobbered): New hook.
16543 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16544 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16545 * doc/tm.texi: Regenerate.
16546 * hooks.h (hook_bool_uint_mode_false): Declare.
16547 * hooks.c (hook_bool_uint_mode_false): New function.
16548 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16549 * cselib.c (cselib_process_insn): Use
16550 targetm.hard_regno_call_part_clobbered instead of
16551 HARD_REGNO_CALL_PART_CLOBBERED.
16552 * ira-conflicts.c (ira_build_conflicts): Likewise.
16553 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16554 * lra-constraints.c (need_for_call_save_p): Likewise.
16555 * lra-lives.c: Include target.h.
16556 (check_pseudos_live_through_calls): Use
16557 targetm.hard_regno_call_part_clobbered instead of
16558 HARD_REGNO_CALL_PART_CLOBBERED.
16559 * regcprop.c: Include target.h.
16560 (copyprop_hardreg_forward_1): Use
16561 targetm.hard_regno_call_part_clobbered instead of
16562 HARD_REGNO_CALL_PART_CLOBBERED.
16563 * reginfo.c (choose_hard_reg_mode): Likewise.
16564 * regrename.c (check_new_reg_p): Likewise.
16565 * reload.c (find_equiv_reg): Likewise.
16566 * reload1.c (emit_reload_insns): Likewise.
16567 * sched-deps.c (deps_analyze_insn): Likewise.
16568 * sel-sched.c (init_regs_for_mode): Likewise.
16569 (mark_unavailable_hard_regs): Likewise.
16570 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16571 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16572 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16574 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16575 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16576 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16578 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16580 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16581 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16582 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16584 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16585 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16586 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16588 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16589 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16591 * config/powerpcspe/powerpcspe.c
16592 (rs6000_hard_regno_call_part_clobbered): New function.
16593 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16594 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16595 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16597 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16598 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16599 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16601 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16602 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16603 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16605 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16606 Alan Hayward <alan.hayward@arm.com>
16607 David Sherwood <david.sherwood@arm.com>
16609 * rtl.h (subreg_memory_offset): Declare.
16610 * emit-rtl.c (subreg_memory_offset): New function.
16611 * expmed.c (store_bit_field_1): Use it.
16612 * expr.c (undefined_operand_subword_p): Likewise.
16613 * simplify-rtx.c (simplify_subreg): Likewise.
16615 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16617 PR rtl-optimization/57448
16620 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16621 using atomic_load pattern.
16622 (expand_atomic_store): Likewise.
16624 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16627 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16628 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16629 handling. Use replace_call_with_value with NULL instead of
16630 gsi_replace, unlink_stmt_vdef and release_defs.
16632 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16635 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16637 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16640 * rtl.h (emit_library_call_value_1): Declare.
16641 (emit_library_call): Replace declaration with a series of overloads.
16642 Remove the parameter count argument.
16643 (emit_library_call_value): Likewise.
16644 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16645 with an "rtx_mode_t *".
16646 (emit_library_call_value): Delete.
16647 (emit_library_call): Likewise.
16648 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16649 (asan_emit_allocas_unpoison): Likewise.
16650 * builtins.c (expand_builtin_powi): Likewise.
16651 (expand_asan_emit_allocas_unpoison): Likewise.
16652 * cfgexpand.c (expand_main_function): Likewise.
16653 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16654 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16655 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16656 * config/arm/arm.c (arm_trampoline_init): Likewise.
16657 (arm_call_tls_get_addr): Likewise.
16658 (arm_expand_divmod_libfunc): Likewise.
16659 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16660 (smulsi3_highpart): Likewise.
16661 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16662 (c6x_expand_compare): Likewise.
16663 (c6x_expand_movmem): Likewise.
16664 * config/frv/frv.c (frv_trampoline_init): Likewise.
16665 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16666 (ix86_expand_divmod_libfunc): Likewise.
16667 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16668 (ia64_expand_compare): Likewise.
16669 (ia64_profile_hook): Likewise.
16670 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16671 (nonlocal_goto): Likewise.
16672 (restore_stack_nonlocal): Likewise.
16673 * config/m32r/m32r.c (block_move_call): Likewise.
16674 (m32r_trampoline_init): Likewise.
16675 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16676 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16677 (m68k_call_m68k_read_tp): Likewise.
16678 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16679 (microblaze_expand_divide): Likewise.
16680 * config/mips/mips.h (mips_args): Likewise.
16681 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16682 (MIPS_ICACHE_SYNC): Likewise.
16683 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16684 (nios2_trampoline_init): Likewise.
16685 * config/pa/pa.c (hppa_tls_call): Likewise.
16686 (pa_trampoline_init): Likewise.
16687 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16688 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16689 (expand_strn_compare): Likewise.
16690 (rs6000_generate_compare): Likewise.
16691 (rs6000_expand_float128_convert): Likewise.
16692 (output_profile_hook): Likewise.
16693 (rs6000_trampoline_init): Likewise.
16694 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16695 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16696 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16697 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16698 (rs6000_generate_compare): Likewise.
16699 (rs6000_expand_float128_convert): Likewise.
16700 (output_profile_hook): Likewise.
16701 (rs6000_trampoline_init): Likewise.
16702 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16703 * config/sh/sh.c (sh_trampoline_init): Likewise.
16704 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16705 (sparc_emit_float_lib_cmp): Likewise.
16706 (sparc32_initialize_trampoline): Likewise.
16707 (sparc64_initialize_trampoline): Likewise.
16708 (sparc_profile_hook): Likewise.
16709 * config/spu/spu.c (ea_load_store): Likewise.
16710 * config/spu/spu.md (floatunssidf2): Likewise.
16711 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16712 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16713 * config/visium/visium.c (expand_block_move_4): Likewise.
16714 (expand_block_move_2): Likewise.
16715 (expand_block_move_1): Likewise.
16716 (expand_block_set_4): Likewise.
16717 (expand_block_set_2): Likewise.
16718 (expand_block_set_1): Likewise.
16719 (visium_trampoline_init): Likewise.
16720 (visium_profile_hook): Likewise.
16721 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16722 (xtensa_setup_frame_addresses): Likewise.
16723 (xtensa_trampoline_init): Likewise.
16724 * except.c (sjlj_emit_function_enter): Likewise.
16725 (sjlj_emit_function_exit): Likewise.
16726 * explow.c (allocate_dynamic_stack_space): Likewise.
16727 (probe_stack_range): Likewise.
16728 * expr.c (convert_mode_scalar): Likewise.
16729 * optabs.c (expand_binop): Likewise.
16730 (expand_twoval_binop_libfunc): Likewise.
16731 (expand_unop): Likewise.
16732 (prepare_cmp_insn): Likewise.
16733 (prepare_float_lib_cmp): Likewise.
16734 (expand_float): Likewise.
16735 (expand_fix): Likewise.
16736 (expand_fixed_convert): Likewise.
16737 (maybe_emit_sync_lock_test_and_set): Likewise.
16738 (expand_atomic_compare_and_swap): Likewise.
16739 (expand_mem_thread_fence): Likewise.
16740 (expand_atomic_fetch_op): Likewise.
16742 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16744 * doc/generic.texi (OpenACC): Adjust URL.
16745 * doc/invoke.texi (C Dialect Options): Ditto.
16747 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16749 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16750 predicate for operand 1. Add (m,<S>) constraint.
16751 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16752 Prevent memory operand 1 with register operand 2.
16754 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16756 PR rtl-optimization/82024
16757 * combine.c (try_combine): If the combination result is a PARALLEL,
16758 and we only need to retain the SET in there that would be placed
16759 at I2, check that we can place that at I3 instead, before doing so.
16761 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16764 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16766 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16767 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16770 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16771 Jeff Law <law@redhat.com>
16773 * varasm.c (bss_initializer_p): Do not put constants into .bss
16774 (categorize_decl_for_section): Handle bss_initializer_p returning
16775 false when DECL_INITIAL is NULL.
16777 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16780 * config/s390/s390.c (s390_can_inline_p): New function.
16782 2017-09-01 Jeff Law <law@redhat.com>
16784 PR tree-optimization/82052
16785 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16786 Always initialize the returned slot after a hash table miss
16787 when INSERT is true.
16789 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16791 * config/s390/s390.md (mem_signal_fence): Remove.
16792 * doc/md.texi (mem_signal_fence): Remove.
16793 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
16795 * target-insns.def (mem_signal_fence): Remove.
16797 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16800 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
16803 * asan.c (create_odr_indicator): Strip name encoding from assembler
16804 name before appending it after __odr_asan_.
16806 2017-09-01 Martin Liska <mliska@suse.cz>
16808 PR tree-optimization/82059
16809 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
16810 frequency only when an edge is redirected.
16812 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16814 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
16815 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
16816 (arc_conditional_register_usage): Remove ARC600 lp_count
16818 (arc_file_start): Emit Tag_ARC_CPU_variation.
16819 (arc_can_use_doloop_p): New conditions to use ZOLs.
16820 (hwloop_fail): New function.
16821 (hwloop_optimize): Likewise.
16822 (hwloop_pattern_reg): Likewise.
16823 (arc_doloop_hooks): New struct, to be used with reorg_loops.
16824 (arc_reorg_loops): New function, calls reorg_loops.
16825 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
16826 (arc600_corereg_hazard): Remove ZOL checking, case handled by
16828 (arc_loop_hazard): Remove function, functionality moved into
16830 (arc_hazard): Remove arc_loop_hazard call.
16831 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
16832 into hwloop_optimize.
16833 (arc_label_align): Remove ZOL handling.
16834 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
16835 * config/arc/arc.md (doloop_begin): Remove pattern.
16836 (doloop_begin_i): Likewise.
16837 (doloop_end_i): Likewise.
16838 (doloop_fallback): Likewise.
16839 (doloop_fallback_m): Likewise.
16840 (doloop_end): Reimplement expand.
16841 (arc_lp): New pattern for LP instruction.
16842 (loop_end): New pattern.
16843 (loop_fail): Likewise.
16844 (decrement_and_branch_until_zero): Likewise.
16845 * config/arc/arc.opt (mlpc-width): New option.
16846 * doc/invoke.texi (mlpc-width): Document option.
16848 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16850 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
16851 (arc_ccfsm_advance): Fix checking for delay slots.
16852 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
16854 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16856 * config/arc/arc.md (movqi_insn): Add stores to save constant long
16858 (movhi_insn): Update store instruction constraint which are saving
16859 6-bit short immediates.
16860 (movsi_insn): Consider also short scaled load operations.
16861 (zero_extendhisi2_i): Use Usd constraint instead of T.
16862 (extendhisi2_i): Add q constraint.
16863 (arc_clzsi2): Add type and length attributes.
16864 (arc_ctzsi2): Likewise.
16865 * config/arc/constraints.md (Usc): Update constraint, the
16866 assembler can parse two relocations for a single instruction.
16868 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16870 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
16871 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
16873 2017-08-31 Olivier Hainque <hainque@adacore.com>
16875 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
16876 match as powerpc-wrs-vxworks*.
16878 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
16880 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
16881 register constraint for by-element operand.
16882 (aarch64_mls_elt_merge<mode>): Likewise.
16884 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16886 * config/arc/arc.c (arc_can_follow_jump): Check for short
16889 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16891 * config.gcc: Use g.opt for arc.
16892 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
16893 functionality moved to ...
16894 (legitimate_scaled_address_p): New function, ...here.
16895 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
16896 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
16897 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
16899 (arc_override_options): Handle G option.
16900 (arc_output_pic_addr_const): Correct function definition.
16901 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
16902 (arc_decl_anon_ns_mem_p): Delete.
16903 (arc_in_small_data_p): Overhaul this function to take into
16904 consideration the value given via G option.
16905 (arc_rewrite_small_data_1): Renamed and corrected old
16906 arc_rewrite_small_data function.
16907 (arc_rewrite_small_data): New function.
16908 (small_data_pattern): Don't use pic_offset_table_rtx.
16909 * config/arc/arc.h (CC1_SPEC): Recognize G option.
16910 * config/arc/simdext.md (movmisalignv2hi): Use
16911 prepare_move_operands function.
16913 (movmisalign*): Likewise.
16914 * doc/invoke.texi (ARC options): Document -G option.
16916 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16918 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
16920 * config/arc/arc.c (arc_print_operand): Output scalled address for
16921 sdata whenever is possible.
16922 (arc_in_small_data_p): Allow sdata for 64bit datum when double
16923 load/stores are available.
16924 (compact_sda_memory_operand): Check for the alignment required by
16925 code density instructions.
16926 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
16928 * config/arc/constraints.md (Usd): Update constraint.
16929 (Us0): New constraint.
16930 (Usc): Update constraint.
16932 2017-08-31 Richard Biener <rguenther@suse.de>
16934 PR middle-end/82054
16935 * dwarf2out.c (dwarf2out_early_global_decl): Process each
16936 function only once.
16938 2017-08-31 Tamar Christina <tamar.christina@arm.com>
16940 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
16941 Resize type_signature.
16943 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16944 Alan Hayward <alan.hayward@arm.com>
16945 David Sherwood <david.sherwood@arm.com>
16947 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
16948 subregs whose inner modes can be stored in GPRs.
16949 (aarch64_classify_index): Likewise.
16951 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16952 Alan Hayward <alan.hayward@arm.com>
16953 David Sherwood <david.sherwood@arm.com>
16955 * config/aarch64/iterators.md (V_cmp_result): Rename to...
16956 (V_INT_EQUIV): ...this.
16957 (v_cmp_result): Rename to...
16958 (v_int_equiv): ...this.
16959 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
16960 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
16961 (copysign<mode>3): Likewise.
16962 (aarch64_simd_bsl<mode>_internal): Likewise.
16963 (aarch64_simd_bsl<mode>): Likewise.
16964 (vec_cmp<mode><mode>): Likewise.
16965 (vcond<mode><mode>): Likewise.
16966 (vcond<v_cmp_mixed><mode>): Likewise.
16967 (vcondu<mode><v_cmp_mixed>): Likewise.
16968 (aarch64_cm<optab><mode>): Likewise.
16969 (aarch64_cmtst<mode>): Likewise.
16970 (aarch64_fac<optab><mode>): Likewise.
16971 (vec_perm_const<mode>): Likewise.
16972 (vcond_mask_<mode><v_cmp_result>): Rename to...
16973 (vcond_mask_<mode><v_int_equiv>): ...this.
16974 (vec_cmp<mode><v_cmp_result>): Rename to...
16975 (vec_cmp<mode><v_int_equiv>): ...this.
16977 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16978 Alan Hayward <alan.hayward@arm.com>
16979 David Sherwood <david.sherwood@arm.com>
16981 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
16983 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
16984 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
16985 (UNSPEC_LD4_DREG): New unspecs.
16986 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
16987 (aarch64_ld2<mode>_dreg_be): Replace with...
16988 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
16990 (aarch64_ld3<mode>_dreg_le)
16991 (aarch64_ld3<mode>_dreg_be): Replace with...
16992 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
16994 (aarch64_ld4<mode>_dreg_le)
16995 (aarch64_ld4<mode>_dreg_be): Replace with...
16996 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
16999 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17001 PR tree-optimization/81987
17002 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17003 insert an initializer in a location not dominated by the stride
17006 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17008 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17009 on the entire header of the finally block in the fallthru case.
17011 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17013 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17015 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17017 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17018 rs6000_emit_move_from_cr and call renamed function.
17019 (rs6000_emit_prologue): Call renamed functions.
17020 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17021 movesi_from_cr, remove volatile CRs.
17023 2017-08-30 Jon Beniston <jon@beniston.com>
17024 Richard Biener <rguenther@suse.de>
17026 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17027 of VECTOR_MODE_P check.
17028 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17029 element vector types.
17031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17033 * df.h (df_read_modify_subreg_p): Remove in favor of...
17034 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17035 const_rtx instead of an rtx.
17036 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17037 * df-problems.c (df_word_lr_mark_ref): Likewise.
17038 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17039 (mark_pseudo_reg_dead): Likewise.
17040 (mark_ref_dead): Likewise.
17041 * reginfo.c (init_subregs_of_mode): Likewise.
17042 * sched-deps.c (sched_analyze_1): Likewise.
17043 * df-scan.c (df_def_record_1): Likewise.
17044 (df_uses_record): Likewise.
17045 (df_read_modify_subreg_p): Remove in favor of...
17046 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17047 const_rtx instead of an rtx.
17049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17050 Alan Hayward <alan.hayward@arm.com>
17051 David Sherwood <david.sherwood@arm.com>
17053 * rtl.h (partial_subreg_p): New function.
17054 * caller-save.c (save_call_clobbered_regs): Use it.
17055 * calls.c (expand_call): Likewise.
17056 * combine.c (combinable_i3pat): Likewise.
17057 (simplify_set): Likewise.
17058 (make_extraction): Likewise.
17059 (make_compound_operation_int): Likewise.
17060 (gen_lowpart_or_truncate): Likewise.
17061 (force_to_mode): Likewise.
17062 (make_field_assignment): Likewise.
17063 (reg_truncated_to_mode): Likewise.
17064 (record_truncated_value): Likewise.
17065 (move_deaths): Likewise.
17066 * cse.c (record_jump_cond): Likewise.
17067 (cse_insn): Likewise.
17068 * cselib.c (cselib_lookup_1): Likewise.
17069 * expmed.c (extract_bit_field_using_extv): Likewise.
17070 * function.c (assign_parm_setup_reg): Likewise.
17071 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17072 * ira-build.c (create_insn_allocnos): Likewise.
17073 * lra-coalesce.c (merge_pseudos): Likewise.
17074 * lra-constraints.c (match_reload): Likewise.
17075 (simplify_operand_subreg): Likewise.
17076 (curr_insn_transform): Likewise.
17077 * lra-lives.c (process_bb_lives): Likewise.
17078 * lra.c (new_insn_reg): Likewise.
17079 (lra_substitute_pseudo): Likewise.
17080 * regcprop.c (mode_change_ok): Likewise.
17081 (maybe_mode_change): Likewise.
17082 (copyprop_hardreg_forward_1): Likewise.
17083 * reload.c (push_reload): Likewise.
17084 (find_reloads): Likewise.
17085 (find_reloads_subreg_address): Likewise.
17086 * reload1.c (alter_reg): Likewise.
17087 (eliminate_regs_1): Likewise.
17088 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17090 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17092 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17093 back to if statements, including unpack.
17095 2017-08-30 Martin Liska <mliska@suse.cz>
17097 PR inline-asm/82001
17098 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17100 (func_checker::compare_asm_inputs_outputs): ... this function.
17101 (func_checker::compare_gimple_asm): Use the function to compare
17102 also ASM constrains.
17103 * ipa-icf-gimple.h: Rename the function.
17105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17106 Alan Hayward <alan.hayward@arm.com>
17107 David Sherwood <david.sherwood@arm.com>
17109 * coretypes.h (complex_mode): New type.
17110 * gdbhooks.py (build_pretty_printer): Handle it.
17111 * machmode.h (complex_mode): New class.
17112 (complex_mode::includes_p): New function.
17113 (is_complex_int_mode): Likewise.
17114 (is_complex_float_mode): Likewise.
17115 * genmodes.c (get_mode_class): Handle complex mode classes.
17116 * function.c (expand_function_end): Use is_complex_int_mode.
17118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17119 Alan Hayward <alan.hayward@arm.com>
17120 David Sherwood <david.sherwood@arm.com>
17122 * coretypes.h (scalar_mode_pod): New typedef.
17123 * gdbhooks.py (build_pretty_printer): Handle it.
17124 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17125 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17126 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17127 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17128 as_a <scalar_mode>.
17130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17131 Alan Hayward <alan.hayward@arm.com>
17132 David Sherwood <david.sherwood@arm.com>
17134 * machmode.h (mode_for_vector): Take a scalar_mode instead
17136 * stor-layout.c (mode_for_vector): Likewise.
17137 * explow.c (promote_mode): Use as_a <scalar_mode>.
17138 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17141 Alan Hayward <alan.hayward@arm.com>
17142 David Sherwood <david.sherwood@arm.com>
17144 * target.def (preferred_simd_mode): Take a scalar_mode
17145 instead of a machine_mode.
17146 * targhooks.h (default_preferred_simd_mode): Likewise.
17147 * targhooks.c (default_preferred_simd_mode): Likewise.
17148 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17149 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17150 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17151 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17152 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17153 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17154 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17155 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17157 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17158 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17159 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17160 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17161 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17162 * doc/tm.texi: Regenerate.
17163 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17166 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17167 Alan Hayward <alan.hayward@arm.com>
17168 David Sherwood <david.sherwood@arm.com>
17170 * target.def (scalar_mode_supported_p): Take a scalar_mode
17171 instead of a machine_mode.
17172 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17173 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17174 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17175 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17176 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17177 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17178 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17179 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17180 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17181 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17182 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17183 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17184 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17185 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17187 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17188 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17189 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17190 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17191 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17193 * doc/tm.texi: Regenerate.
17195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17196 Alan Hayward <alan.hayward@arm.com>
17197 David Sherwood <david.sherwood@arm.com>
17199 * coretypes.h (opt_scalar_mode): New typedef.
17200 * gdbhooks.py (build_pretty_printers): Handle it.
17201 * machmode.h (mode_iterator::get_2xwider): Add overload for
17203 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17205 * expr.c (convert_mode_scalar): Likewise.
17206 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17207 * optabs.c (expand_float): Likewise.
17208 (expand_fix): Likewise.
17209 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17211 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17212 Alan Hayward <alan.hayward@arm.com>
17213 David Sherwood <david.sherwood@arm.com>
17215 * optabs.c (expand_float): Explicitly check for scalars before
17216 using a branching expansion.
17217 (expand_fix): Likewise.
17219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17220 Alan Hayward <alan.hayward@arm.com>
17221 David Sherwood <david.sherwood@arm.com>
17223 * expr.c (convert_mode): Split scalar handling out into...
17224 (convert_mode_scalar): ...this new function. Treat the modes
17227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17228 Alan Hayward <alan.hayward@arm.com>
17229 David Sherwood <david.sherwood@arm.com>
17231 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17233 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17236 Alan Hayward <alan.hayward@arm.com>
17237 David Sherwood <david.sherwood@arm.com>
17239 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17240 rather than a machine_mode.
17241 (fixed_from_string): Likewise.
17242 (fixed_convert): Likewise.
17243 (fixed_convert_from_int): Likewise.
17244 (fixed_convert_from_real): Likewise.
17245 (real_convert_from_fixed): Likewise.
17246 * fixed-value.c (fixed_from_double_int): Likewise.
17247 (fixed_from_string): Likewise.
17248 (fixed_convert): Likewise.
17249 (fixed_convert_from_int): Likewise.
17250 (fixed_convert_from_real): Likewise.
17251 (real_convert_from_fixed): Likewise.
17252 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17255 Alan Hayward <alan.hayward@arm.com>
17256 David Sherwood <david.sherwood@arm.com>
17258 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17259 of separate mode class checks. Do not allow vector modes here.
17260 (immed_wide_int_const): Use as_a <scalar_mode>.
17261 * explow.c (trunc_int_for_mode): Likewise.
17262 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17263 (wi::shwi): Likewise.
17264 (wi::min_value): Likewise.
17265 (wi::max_value): Likewise.
17266 * dwarf2out.c (loc_descriptor): Likewise.
17267 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17268 for CONST_WIDE_INT.
17270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17271 Alan Hayward <alan.hayward@arm.com>
17272 David Sherwood <david.sherwood@arm.com>
17274 * tree.h (SCALAR_TYPE_MODE): New macro.
17275 * expr.c (expand_expr_addr_expr_1): Use it.
17276 (expand_expr_real_2): Likewise.
17277 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17278 (fold_convert_const_fixed_from_int): Likewise.
17279 (fold_convert_const_fixed_from_real): Likewise.
17280 (native_encode_fixed): Likewise
17281 (native_encode_complex): Likewise
17282 (native_encode_vector): Likewise.
17283 (native_interpret_fixed): Likewise.
17284 (native_interpret_real): Likewise.
17285 (native_interpret_complex): Likewise.
17286 (native_interpret_vector): Likewise.
17287 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17288 (simd_clone_adjust_argument_types): Likewise.
17289 (simd_clone_init_simd_arrays): Likewise.
17290 (simd_clone_adjust): Likewise.
17291 * stor-layout.c (layout_type): Likewise.
17292 * tree.c (build_minus_one_cst): Likewise.
17293 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17294 * tree-inline.c (estimate_move_cost): Likewise.
17295 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17296 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17297 (vectorizable_reduction): Likewise.
17298 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17299 (vect_recog_mixed_size_cond_pattern): Likewise.
17300 (check_bool_pattern): Likewise.
17301 (adjust_bool_pattern): Likewise.
17302 (search_type_for_mask_1): Likewise.
17303 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17304 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17305 (vectorizable_load): Likewise.
17306 (vectorizable_store): Likewise.
17307 * ubsan.c (ubsan_encode_value): Likewise.
17308 * varasm.c (output_constant): Likewise.
17310 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17311 Alan Hayward <alan.hayward@arm.com>
17312 David Sherwood <david.sherwood@arm.com>
17314 * coretypes.h (scalar_mode): New class.
17315 * machmode.h (scalar_mode): Likewise.
17316 (scalar_mode::includes_p): New function.
17317 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17318 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17319 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17320 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17321 * expmed.c (store_bit_field_1): Likewise.
17322 (extract_bit_field_1): Likewise.
17323 * expr.c (write_complex_part): Likewise.
17324 (read_complex_part): Likewise.
17325 (emit_move_complex_push): Likewise.
17326 (expand_expr_real_2): Likewise.
17327 * function.c (assign_parm_setup_reg): Likewise.
17328 (assign_parms_unsplit_complex): Likewise.
17329 * optabs.c (expand_binop): Likewise.
17330 * rtlanal.c (subreg_get_info): Likewise.
17331 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17332 * varasm.c (output_constant_pool_2): Likewise.
17334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17335 Alan Hayward <alan.hayward@arm.com>
17336 David Sherwood <david.sherwood@arm.com>
17338 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17340 (expmed_mult_highpart_optab): Likewise.
17341 (expmed_mult_highpart): Likewise.
17343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17344 Alan Hayward <alan.hayward@arm.com>
17345 David Sherwood <david.sherwood@arm.com>
17347 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17348 instead of a machine_mode.
17349 (builtin_memset_read_str): Likewise.
17350 * builtins.c (c_readstr): Likewise.
17351 (builtin_memcpy_read_str): Likewise.
17352 (builtin_strncpy_read_str): Likewise.
17353 (builtin_memset_read_str): Likewise.
17354 (builtin_memset_gen_str): Likewise.
17355 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17356 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17357 instead of a machine_mode.
17358 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17360 (make_extraction): Likewise.
17361 (try_widen_shift_mode): Take and return scalar_int_modes instead
17363 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17364 a scalar_int_mode instead of a machine_mode.
17365 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17366 (avr_addr_space_pointer_mode): Likewise.
17367 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17368 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17369 (msp430_unwind_word_mode): Likewise.
17370 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17371 (spu_addr_space_pointer_mode): Likewise.
17372 (spu_addr_space_address_mode): Likewise.
17373 (spu_libgcc_cmp_return_mode): Likewise.
17374 (spu_libgcc_shift_count_mode): Likewise.
17375 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17376 (rl78_addr_space_pointer_mode): Likewise.
17377 (fl78_unwind_word_mode): Likewise.
17378 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17380 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17381 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17382 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17383 (mips_valid_pointer_mode): Likewise.
17384 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17385 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17386 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17388 (ft32_addr_space_address_mode): Likewise.
17389 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17390 scalar_int_mode instead of a machine_mode.
17391 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17393 (m32c_addr_space_address_mode): Likewise.
17394 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17395 (rs6000_eh_return_filter_mode): Likewise.
17396 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17397 (rs6000_eh_return_filter_mode): Likewise.
17398 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17399 (s390_libgcc_shift_count_mode): Likewise.
17400 (s390_unwind_word_mode): Likewise.
17401 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17403 * target.def (mode_rep_extended): Likewise.
17404 (valid_pointer_mode): Likewise.
17405 (addr_space.valid_pointer_mode): Likewise.
17406 (eh_return_filter_mode): Return a scalar_int_mode rather than
17408 (libgcc_cmp_return_mode): Likewise.
17409 (libgcc_shift_count_mode): Likewise.
17410 (unwind_word_mode): Likewise.
17411 (addr_space.pointer_mode): Likewise.
17412 (addr_space.address_mode): Likewise.
17413 * doc/tm.texi: Regenerate.
17414 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17416 (do_jump): Use scalar_int_mode for local variables.
17417 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17418 rather than a machine_mode.
17419 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17420 (scompare_loc_descriptor_wide): Likewise.
17421 (scompare_loc_descriptor_narrow): Likewise.
17422 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17424 * except.c (sjlj_emit_dispatch_table): Likewise.
17425 (expand_builtin_eh_copy_values): Likewise.
17426 * explow.c (convert_memory_address_addr_space_1): Likewise.
17427 Take a scalar_int_mode rather than a machine_mode.
17428 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17429 than a machine_mode.
17430 (memory_address_addr_space): Use scalar_int_mode for local variables.
17431 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17432 rather than a machine_mode.
17433 * expmed.c (mask_rtx): Likewise.
17434 (init_expmed_one_conv): Likewise.
17435 (expand_mult_highpart_adjust): Likewise.
17436 (extract_high_half): Likewise.
17437 (expmed_mult_highpart_optab): Likewise.
17438 (expmed_mult_highpart): Likewise.
17439 (expand_smod_pow2): Likewise.
17440 (expand_sdiv_pow2): Likewise.
17441 (emit_store_flag_int): Likewise.
17442 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17444 (extract_low_bits): Likewise.
17445 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17447 * expr.c (pieces_addr::adjust): Likewise.
17448 (can_store_by_pieces): Likewise.
17449 (store_by_pieces): Likewise.
17450 (clear_by_pieces_1): Likewise.
17451 (expand_expr_addr_expr_1): Likewise.
17452 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17453 (expand_expr_real_1): Likewise.
17454 (try_casesi): Likewise.
17455 * final.c (shorten_branches): Likewise.
17456 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17457 type of "mode" to machine_mode.
17458 * internal-fn.c (expand_arith_overflow_result_store): Take a
17459 scalar_int_mode rather than a machine_mode.
17460 (expand_mul_overflow): Use scalar_int_mode for local variables.
17461 * loop-doloop.c (doloop_modify): Likewise.
17462 (doloop_optimize): Likewise.
17463 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17464 than a machine_mode.
17465 (expand_doubleword_shift_condmove): Likewise.
17466 (expand_doubleword_shift): Likewise.
17467 (expand_doubleword_clz): Likewise.
17468 (expand_doubleword_popcount): Likewise.
17469 (expand_doubleword_parity): Likewise.
17470 (expand_absneg_bit): Use scalar_int_mode for local variables.
17471 (prepare_float_lib_cmp): Likewise.
17472 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17473 rather than a machine_mode.
17474 (convert_memory_address_addr_space): Likewise.
17475 (get_mode_bounds): Likewise.
17476 (get_address_mode): Return a scalar_int_mode rather than a
17478 * rtlanal.c (get_address_mode): Likewise.
17479 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17480 than a machine_mode.
17481 * targhooks.c (default_mode_rep_extended): Likewise.
17482 (default_valid_pointer_mode): Likewise.
17483 (default_addr_space_valid_pointer_mode): Likewise.
17484 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17485 than a machine_mode.
17486 (default_libgcc_cmp_return_mode): Likewise.
17487 (default_libgcc_shift_count_mode): Likewise.
17488 (default_unwind_word_mode): Likewise.
17489 (default_addr_space_pointer_mode): Likewise.
17490 (default_addr_space_address_mode): Likewise.
17491 * targhooks.h (default_eh_return_filter_mode): Likewise.
17492 (default_libgcc_cmp_return_mode): Likewise.
17493 (default_libgcc_shift_count_mode): Likewise.
17494 (default_unwind_word_mode): Likewise.
17495 (default_addr_space_pointer_mode): Likewise.
17496 (default_addr_space_address_mode): Likewise.
17497 (default_mode_rep_extended): Take a scalar_int_mode rather than
17499 (default_valid_pointer_mode): Likewise.
17500 (default_addr_space_valid_pointer_mode): Likewise.
17501 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17503 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17504 rather than a machine_mode.
17505 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17506 for local variables.
17507 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17508 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17509 than a machine_mode.
17511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17512 Alan Hayward <alan.hayward@arm.com>
17513 David Sherwood <david.sherwood@arm.com>
17515 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17516 the mode argument to scalar_int_mode.
17517 (do_jump_by_parts_zero_rtx): Likewise.
17518 (do_jump_by_parts_equality_rtx): Likewise.
17519 (do_jump_by_parts_greater): Take a mode argument.
17520 (do_jump_by_parts_equality): Likewise.
17521 (do_jump_1): Update calls accordingly.
17523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17524 Alan Hayward <alan.hayward@arm.com>
17525 David Sherwood <david.sherwood@arm.com>
17527 * is-a.h (safe_dyn_cast): New function.
17528 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17529 (jump_table_for_label): Likewise.
17530 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17531 instead of an rtx_insn *.
17532 (shorten_branches): Use dyn_cast instead of LABEL_P and
17533 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17534 rtx_jump_table_data::get_data_mode.
17535 (final_scan_insn): Likewise.
17537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17538 Alan Hayward <alan.hayward@arm.com>
17539 David Sherwood <david.sherwood@arm.com>
17541 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17542 trying to combine a full-register integer set with a subreg
17545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17546 Alan Hayward <alan.hayward@arm.com>
17547 David Sherwood <david.sherwood@arm.com>
17549 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17550 that is always either address_mode or pointer_mode.
17552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17553 Alan Hayward <alan.hayward@arm.com>
17554 David Sherwood <david.sherwood@arm.com>
17556 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17557 when the two are known to be equal.
17559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17560 Alan Hayward <alan.hayward@arm.com>
17561 David Sherwood <david.sherwood@arm.com>
17563 * simplify-rtx.c (simplify_const_unary_operation): Use
17564 is_a <scalar_int_mode> instead of checking for a nonzero
17565 precision. Forcibly convert op_mode to a scalar_int_mode
17566 in that case. More clearly differentiate the operand and
17567 result modes and use the former when deciding what the value
17568 of a count-bits operation should be. Use is_int_mode instead
17569 of checking for a MODE_INT. Remove redundant check for whether
17570 this mode has a zero precision.
17572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17573 Alan Hayward <alan.hayward@arm.com>
17574 David Sherwood <david.sherwood@arm.com>
17576 * optabs.c (widen_leading): Change the type of the mode argument
17577 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17578 (widen_bswap): Likewise.
17579 (expand_parity): Likewise.
17580 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17581 (expand_ffs): Likewise.
17582 (epand_unop): Check for scalar integer modes before calling the
17585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17586 Alan Hayward <alan.hayward@arm.com>
17587 David Sherwood <david.sherwood@arm.com>
17589 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17591 (expand_expr_real_1): Update call accordingly.
17593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17594 Alan Hayward <alan.hayward@arm.com>
17595 David Sherwood <david.sherwood@arm.com>
17597 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17598 value_mode arguments. Use scalar_int_mode internally.
17599 (store_bit_field_1): Rename the new integer mode from imode
17600 to op0_mode and use it instead of GET_MODE (op0). Update calls
17601 to store_split_bit_field, store_bit_field_using_insv and
17602 store_fixed_bit_field.
17603 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17604 Use scalar_int_mode internally. Use a bit count rather than a mode
17605 when calculating the largest bit size for get_best_mode.
17606 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17607 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17608 Remove assertion that OP0 has a scalar integer mode.
17609 (store_split_bit_field): Add op0_mode and value_mode arguments.
17610 Update calls to extract_fixed_bit_field.
17611 (extract_bit_field_using_extv): Add an op0_mode argument.
17612 Use scalar_int_mode internally.
17613 (extract_bit_field_1): Rename the new integer mode from imode to
17614 op0_mode and use it instead of GET_MODE (op0). Update calls to
17615 extract_split_bit_field, extract_bit_field_using_extv and
17616 extract_fixed_bit_field.
17617 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17618 to extract_split_bit_field and extract_fixed_bit_field_1.
17619 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17620 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17621 on the target mode.
17622 (extract_split_bit_field): Add an op0_mode argument. Update call
17623 to extract_fixed_bit_field.
17625 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17626 Alan Hayward <alan.hayward@arm.com>
17627 David Sherwood <david.sherwood@arm.com>
17629 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17630 * explow.c (hard_function_value): Likewise.
17631 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17632 convert_to_mode call outside the loop.
17633 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17634 for the mode iterator. Require the mode specified by max_pieces
17636 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17638 (copy_blkmode_to_reg): Likewise.
17639 (set_storage_via_setmem): Likewise.
17640 * optabs.c (prepare_cmp_insn): Likewise.
17641 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17642 * stor-layout.c (finish_bitfield_representative): Likewise.
17644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17645 Alan Hayward <alan.hayward@arm.com>
17646 David Sherwood <david.sherwood@arm.com>
17648 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17649 * expr.c (convert_move): Use them.
17650 (convert_modes): Likewise.
17651 (store_expr_with_bounds): Likewise.
17653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17654 Alan Hayward <alan.hayward@arm.com>
17655 David Sherwood <david.sherwood@arm.com>
17657 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17658 parameter for the mode of "x". Remove the "known_x", "known_mode"
17659 and "known_ret" arguments. Change the type of the mode argument
17660 to scalar_int_mode.
17661 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17662 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17663 (reg_num_sign_bit_copies_for_combine): Likewise.
17664 * rtlanal.c (nonzero_bits1): Likewise.
17665 (num_sign_bit_copies1): Likewise.
17666 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17667 (reg_num_sign_bit_copies_general): Likewise.
17668 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17669 (reg_nonzero_bits_general): Likewise.
17671 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17672 Alan Hayward <alan.hayward@arm.com>
17673 David Sherwood <david.sherwood@arm.com>
17675 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17676 than in subroutines. Return 1 for non-integer modes.
17677 (cached_num_sign_bit_copies): Change the type of the mode parameter
17678 to scalar_int_mode.
17679 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17680 classes. Handle CONST_INT_P first and then check whether X also
17681 has a scalar integer mode. Check the same thing for inner registers
17682 of a SUBREG and for values that are being extended or truncated.
17684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17685 Alan Hayward <alan.hayward@arm.com>
17686 David Sherwood <david.sherwood@arm.com>
17688 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17689 in subroutines. Return the mode mask for non-integer modes.
17690 (cached_nonzero_bits): Change the type of the mode parameter
17691 to scalar_int_mode.
17692 (nonzero_bits1): Likewise. Remove early exit for other mode
17693 classes. Handle CONST_INT_P first and then check whether X
17694 also has a scalar integer mode.
17696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17697 Alan Hayward <alan.hayward@arm.com>
17698 David Sherwood <david.sherwood@arm.com>
17700 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17701 Return a scalar_int_mode and assert that the size is greater than
17703 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17704 (op_by_pieces_d::op_by_pieces_d): Likewise.
17705 (op_by_pieces_d::run): Likewise.
17706 (can_store_by_pieces): Likewise.
17708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17709 Alan Hayward <alan.hayward@arm.com>
17710 David Sherwood <david.sherwood@arm.com>
17712 * combine.c (extract_left_shift): Add a mode argument and update
17714 (make_compound_operation_int): Change the type of the mode parameter
17715 to scalar_int_mode and update the call to extract_left_shift.
17717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17718 Alan Hayward <alan.hayward@arm.com>
17719 David Sherwood <david.sherwood@arm.com>
17721 * combine.c (simplify_and_const_int): Change the type of the mode
17722 parameter to scalar_int_mode.
17723 (simplify_and_const_int_1): Likewise. Update recursive call.
17725 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17726 Alan Hayward <alan.hayward@arm.com>
17727 David Sherwood <david.sherwood@arm.com>
17729 * combine.c (simplify_compare_const): Check that the mode is a
17730 scalar_int_mode (rather than VOIDmode) before testing its
17732 (simplify_comparison): Move COMPARISON_P handling out of the
17733 loop and restrict the latter part of the loop to scalar_int_modes.
17734 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17735 and when considering SUBREG_REGs. Use is_int_mode instead of
17736 checking GET_MODE_CLASS against MODE_INT.
17738 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17739 Alan Hayward <alan.hayward@arm.com>
17740 David Sherwood <david.sherwood@arm.com>
17742 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17743 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17744 shift_unit_mode and for modes involved in scalar shifts.
17746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17747 Alan Hayward <alan.hayward@arm.com>
17748 David Sherwood <david.sherwood@arm.com>
17750 * combine.c (force_int_to_mode): New function, split out from...
17751 (force_to_mode): ...here. Keep xmode up-to-date and use it
17752 instead of GET_MODE (x).
17754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17755 Alan Hayward <alan.hayward@arm.com>
17756 David Sherwood <david.sherwood@arm.com>
17758 * optabs-query.h (extraction_insn::struct_mode): Change type to
17759 opt_scalar_int_mode and update comment.
17760 (extraction_insn::field_mode): Change type to scalar_int_mode.
17761 (extraction_insn::pos_mode): Likewise.
17762 * combine.c (make_extraction): Update accordingly.
17763 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17764 (get_optab_extraction_insn): Likewise.
17765 * recog.c (simplify_while_replacing): Likewise.
17766 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17767 parameter to opt_scalar_int_mode.
17769 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17770 Alan Hayward <alan.hayward@arm.com>
17771 David Sherwood <david.sherwood@arm.com>
17773 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17774 to a scalar_int_mode instead of a machine_mode.
17775 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17776 (get_best_mode): Return a boolean and use a pointer argument to store
17777 the selected mode. Replace the limit mode parameter with a bit limit.
17778 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17779 for the values returned by bit_field_mode_iterator::next_mode.
17780 (store_bit_field): Update call to get_best_mode.
17781 (store_fixed_bit_field): Likewise.
17782 (extract_fixed_bit_field): Likewise.
17783 * expr.c (optimize_bitfield_assignment_op): Likewise.
17784 * fold-const.c (optimize_bit_field_compare): Likewise.
17785 (fold_truth_andor_1): Likewise.
17786 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17787 Update for new type of m_mode.
17788 (get_best_mode): As above.
17790 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17791 Alan Hayward <alan.hayward@arm.com>
17792 David Sherwood <david.sherwood@arm.com>
17794 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
17795 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
17796 (store_bit_field): Check is_a <scalar_int_mode> before calling
17797 strict_volatile_bitfield_p.
17798 (extract_bit_field): Likewise.
17800 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17801 Alan Hayward <alan.hayward@arm.com>
17802 David Sherwood <david.sherwood@arm.com>
17804 * target.def (cstore_mode): Return a scalar_int_mode.
17805 * doc/tm.texi: Regenerate.
17806 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
17807 * targhooks.h (default_cstore_mode): Likewise.
17808 * targhooks.c (default_cstore_mode): Likewise, using a forced
17810 * expmed.c (emit_cstore): Expect the target of the cstore to be
17813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17814 Alan Hayward <alan.hayward@arm.com>
17815 David Sherwood <david.sherwood@arm.com>
17817 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
17819 (niter_desc): Likewise mode.
17820 (iv_analyze): Add a mode parameter.
17822 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
17823 and change its type to scalar_int_mode.
17824 * loop-iv.c: Update commentary at head of file.
17825 (iv_constant): Pass the mode paraeter before the rtx it describes
17826 and change its type to scalar_int_mode. Remove VOIDmode handling.
17827 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
17828 (iv_extend): Likewise.
17829 (shorten_into_mode): Likewise.
17830 (iv_add): Use scalar_int_mode.
17831 (iv_mult): Likewise.
17832 (iv_shift): Likewise.
17833 (canonicalize_iv_subregs): Likewise.
17834 (get_biv_step_1): Pass the outer_mode parameter before the rtx
17835 it describes and change its mode to scalar_int_mode. Also change
17836 the type of the returned inner_mode to scalar_int_mode.
17837 (get_biv_step): Likewise, turning outer_mode from a pointer
17838 into a direct parameter. Update call to get_biv_step_1.
17839 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
17840 iv_constant and get_biv_step.
17841 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
17842 and change its type to scalar_int_mode. Don't initialise iv->mode
17843 to VOIDmode and remove later checks for its still being VOIDmode.
17844 Update calls to iv_analyze_op and iv_analyze_expr. Check
17845 is_a <scalar_int_mode> when changing the mode under consideration.
17846 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
17847 Update call to iv_analyze_expr.
17848 (iv_analyze_op): Add a mode parameter. Reject subregs whose
17849 inner register is not also a scalar_int_mode. Update call to
17851 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
17852 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
17853 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
17854 separate mode class checks. Update calls to iv_analyze. Remove
17855 fix-up of VOIDmodes after iv_analyze_biv.
17856 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
17857 don't have a scalar_int_mode. Update call to biv_p.
17859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17860 Alan Hayward <alan.hayward@arm.com>
17861 David Sherwood <david.sherwood@arm.com>
17863 * cfgexpand.c (convert_debug_memory_address): Use
17864 as_a <scalar_int_mode>.
17865 * combine.c (expand_compound_operation): Likewise.
17866 (make_extraction): Likewise.
17867 (change_zero_ext): Likewise.
17868 (simplify_comparison): Likewise.
17869 * cse.c (cse_insn): Likewise.
17870 * dwarf2out.c (minmax_loc_descriptor): Likewise.
17871 (mem_loc_descriptor): Likewise.
17872 (loc_descriptor): Likewise.
17873 * expmed.c (init_expmed_one_mode): Likewise.
17874 (synth_mult): Likewise.
17875 (emit_store_flag_1): Likewise.
17876 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
17877 of a comparison with size.
17878 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
17879 (reduce_to_bit_field_precision): Likewise.
17880 * function.c (expand_function_end): Likewise.
17881 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
17882 * loop-doloop.c (doloop_modify): Likewise.
17883 * optabs.c (expand_binop): Likewise.
17884 (expand_unop): Likewise.
17885 (expand_copysign_absneg): Likewise.
17886 (prepare_cmp_insn): Likewise.
17887 (maybe_legitimize_operand): Likewise.
17888 * recog.c (const_scalar_int_operand): Likewise.
17889 * rtlanal.c (get_address_mode): Likewise.
17890 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17891 (simplify_cond_clz_ctz): Likewise.
17892 * tree-nested.c (get_nl_goto_field): Likewise.
17893 * tree.c (build_vector_type_for_mode): Likewise.
17894 * var-tracking.c (use_narrower_mode): Likewise.
17896 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17897 Alan Hayward <alan.hayward@arm.com>
17898 David Sherwood <david.sherwood@arm.com>
17900 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
17901 * builtins.c (expand_builtin_signbit): Use it.
17902 * cfgexpand.c (expand_debug_expr): Likewise.
17903 * dojump.c (do_jump): Likewise.
17904 (do_compare_and_jump): Likewise.
17905 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
17906 * expmed.c (make_tree): Likewise.
17907 * expr.c (expand_expr_real_2): Likewise.
17908 (expand_expr_real_1): Likewise.
17909 (try_casesi): Likewise.
17910 * fold-const-call.c (fold_const_call_ss): Likewise.
17911 * fold-const.c (unextend): Likewise.
17912 (extract_muldiv_1): Likewise.
17913 (fold_single_bit_test): Likewise.
17914 (native_encode_int): Likewise.
17915 (native_encode_string): Likewise.
17916 (native_interpret_int): Likewise.
17917 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
17918 * internal-fn.c (expand_addsub_overflow): Likewise.
17919 (expand_neg_overflow): Likewise.
17920 (expand_mul_overflow): Likewise.
17921 (expand_arith_overflow): Likewise.
17922 * match.pd: Likewise.
17923 * stor-layout.c (layout_type): Likewise.
17924 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17925 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
17926 * tree-ssanames.c (get_range_info): Likewise.
17927 * tree-switch-conversion.c (array_value_type) Likewise.
17928 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
17929 (vect_recog_divmod_pattern): Likewise.
17930 (vect_recog_mixed_size_cond_pattern): Likewise.
17931 * tree-vrp.c (extract_range_basic): Likewise.
17932 (simplify_float_conversion_using_ranges): Likewise.
17933 * tree.c (int_fits_type_p): Likewise.
17934 * ubsan.c (instrument_bool_enum_load): Likewise.
17935 * varasm.c (mergeable_string_section): Likewise.
17936 (narrowing_initializer_constant_valid_p): Likewise.
17937 (output_constant): Likewise.
17939 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17940 Alan Hayward <alan.hayward@arm.com>
17941 David Sherwood <david.sherwood@arm.com>
17943 * machmode.h (NARROWEST_INT_MODE): New macro.
17944 * expr.c (alignment_for_piecewise_move): Use it instead of
17945 GET_CLASS_NARROWEST_MODE (MODE_INT).
17946 (push_block): Likewise.
17947 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
17949 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17952 Alan Hayward <alan.hayward@arm.com>
17953 David Sherwood <david.sherwood@arm.com>
17955 * postreload.c (move2add_valid_value_p): Change the type of the
17956 mode parameter to scalar_int_mode.
17957 (move2add_use_add2_insn): Add a mode parameter and use it instead
17959 (move2add_use_add3_insn): Likewise.
17960 (reload_cse_move2add): Update accordingly.
17962 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17963 Alan Hayward <alan.hayward@arm.com>
17964 David Sherwood <david.sherwood@arm.com>
17966 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
17968 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
17969 * optabs.c (expand_unop): Likewise.
17971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17972 Alan Hayward <alan.hayward@arm.com>
17973 David Sherwood <david.sherwood@arm.com>
17975 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
17976 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
17977 (popcount_loc_descriptor): Likewise.
17978 (bswap_loc_descriptor): Likewise.
17979 (rotate_loc_descriptor): Likewise.
17980 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
17981 calling the functions above.
17983 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17984 Alan Hayward <alan.hayward@arm.com>
17985 David Sherwood <david.sherwood@arm.com>
17987 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
17989 (try_combine): Likewise.
17990 (simplify_if_then_else): Likewise.
17991 * cse.c (cse_insn): Likewise.
17992 * dwarf2out.c (mem_loc_descriptor): Likewise.
17993 * emit-rtl.c (gen_lowpart_common): Likewise.
17994 * simplify-rtx.c (simplify_truncation): Likewise.
17995 (simplify_binary_operation_1): Likewise.
17996 (simplify_const_relational_operation): Likewise.
17997 (simplify_ternary_operation): Likewise.
17998 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18000 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18001 Alan Hayward <alan.hayward@arm.com>
18002 David Sherwood <david.sherwood@arm.com>
18004 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18005 * reload.c (push_reload): Likewise.
18006 (find_reloads): Likewise.
18008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18009 Alan Hayward <alan.hayward@arm.com>
18010 David Sherwood <david.sherwood@arm.com>
18012 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18013 (make_compound_operation_int): Likewise.
18014 (change_zero_ext): Likewise.
18015 * expr.c (convert_move): Likewise.
18016 (convert_modes): Likewise.
18017 * fwprop.c (forward_propagate_subreg): Likewise.
18018 * loop-iv.c (get_biv_step_1): Likewise.
18019 * optabs.c (widen_operand): Likewise.
18020 * postreload.c (move2add_valid_value_p): Likewise.
18021 * recog.c (simplify_while_replacing): Likewise.
18022 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18023 (simplify_binary_operation_1): Likewise. Remove redundant
18024 mode equality check.
18026 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18027 Alan Hayward <alan.hayward@arm.com>
18028 David Sherwood <david.sherwood@arm.com>
18030 * combine.c (combine_simplify_rtx): Add checks for
18031 is_a <scalar_int_mode>.
18032 (simplify_if_then_else): Likewise.
18033 (make_field_assignment): Likewise.
18034 (simplify_comparison): Likewise.
18035 * ifcvt.c (noce_try_bitop): Likewise.
18036 * loop-invariant.c (canonicalize_address_mult): Likewise.
18037 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18040 Alan Hayward <alan.hayward@arm.com>
18041 David Sherwood <david.sherwood@arm.com>
18043 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18044 is_a <scalar_int_mode> instead of != BLKmode.
18046 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18047 Alan Hayward <alan.hayward@arm.com>
18048 David Sherwood <david.sherwood@arm.com>
18050 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18051 instead of != VOIDmode.
18052 * combine.c (if_then_else_cond): Likewise.
18053 (change_zero_ext): Likewise.
18054 * dwarf2out.c (mem_loc_descriptor): Likewise.
18055 (loc_descriptor): Likewise.
18056 * rtlanal.c (canonicalize_condition): Likewise.
18057 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18059 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18060 Alan Hayward <alan.hayward@arm.com>
18061 David Sherwood <david.sherwood@arm.com>
18063 * simplify-rtx.c (simplify_binary_operation_1): Use
18064 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18066 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18067 Alan Hayward <alan.hayward@arm.com>
18068 David Sherwood <david.sherwood@arm.com>
18070 * wide-int.h (int_traits<unsigned char>) New class.
18071 (int_traits<unsigned short>) Likewise.
18072 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18073 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18075 * combine.c (set_nonzero_bits_and_sign_copies): Use
18076 is_a <scalar_int_mode>.
18077 (find_split_point): Likewise.
18078 (combine_simplify_rtx): Likewise.
18079 (simplify_logical): Likewise.
18080 (expand_compound_operation): Likewise.
18081 (expand_field_assignment): Likewise.
18082 (make_compound_operation): Likewise.
18083 (extended_count): Likewise.
18084 (change_zero_ext): Likewise.
18085 (simplify_comparison): Likewise.
18086 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18087 (ucompare_loc_descriptor): Likewise.
18088 (minmax_loc_descriptor): Likewise.
18089 (mem_loc_descriptor): Likewise.
18090 (loc_descriptor): Likewise.
18091 * expmed.c (init_expmed_one_mode): Likewise.
18092 * lra-constraints.c (lra_constraint_offset): Likewise.
18093 * optabs.c (prepare_libcall_arg): Likewise.
18094 * postreload.c (move2add_note_store): Likewise.
18095 * reload.c (operands_match_p): Likewise.
18096 * rtl.h (load_extend_op): Likewise.
18097 * rtlhooks.c (gen_lowpart_general): Likewise.
18098 * simplify-rtx.c (simplify_truncation): Likewise.
18099 (simplify_unary_operation_1): Likewise.
18100 (simplify_binary_operation_1): Likewise.
18101 (simplify_const_binary_operation): Likewise.
18102 (simplify_const_relational_operation): Likewise.
18103 (simplify_subreg): Likewise.
18104 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18105 * var-tracking.c (adjust_mems): Likewise.
18106 (prepare_call_arguments): Likewise.
18108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18109 Alan Hayward <alan.hayward@arm.com>
18110 David Sherwood <david.sherwood@arm.com>
18112 * machmode.h (is_int_mode): New fuction.
18113 * combine.c (find_split_point): Use it.
18114 (combine_simplify_rtx): Likewise.
18115 (simplify_if_then_else): Likewise.
18116 (simplify_set): Likewise.
18117 (simplify_shift_const_1): Likewise.
18118 (simplify_comparison): Likewise.
18119 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18120 * cse.c (notreg_cost): Likewise.
18121 (cse_insn): Likewise.
18122 * cselib.c (cselib_lookup_1): Likewise.
18123 * dojump.c (do_jump_1): Likewise.
18124 (do_compare_rtx_and_jump): Likewise.
18125 * dse.c (get_call_args): Likewise.
18126 * dwarf2out.c (rtl_for_decl_init): Likewise.
18127 (native_encode_initializer): Likewise.
18128 * expmed.c (emit_store_flag_1): Likewise.
18129 (emit_store_flag): Likewise.
18130 * expr.c (convert_modes): Likewise.
18131 (store_field): Likewise.
18132 (expand_expr_real_1): Likewise.
18133 * fold-const.c (fold_read_from_constant_string): Likewise.
18134 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18135 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18136 * optabs.c (expand_binop): Likewise.
18137 (expand_unop): Likewise.
18138 (expand_abs_nojump): Likewise.
18139 (expand_one_cmpl_abs_nojump): Likewise.
18140 * simplify-rtx.c (mode_signbit_p): Likewise.
18141 (val_signbit_p): Likewise.
18142 (val_signbit_known_set_p): Likewise.
18143 (val_signbit_known_clear_p): Likewise.
18144 (simplify_relational_operation_1): Likewise.
18145 * tree.c (vector_type_mode): Likewise.
18147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18148 Alan Hayward <alan.hayward@arm.com>
18149 David Sherwood <david.sherwood@arm.com>
18151 * machmode.h (smallest_mode_for_size): Fix formatting.
18152 (smallest_int_mode_for_size): New function.
18153 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18154 instead of smallest_mode_for_size.
18155 * combine.c (make_extraction): Likewise.
18156 * config/arc/arc.c (arc_expand_movmem): Likewise.
18157 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18158 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18159 * config/s390/s390.c (s390_expand_insv): Likewise.
18160 * config/sparc/sparc.c (assign_int_registers): Likewise.
18161 * config/spu/spu.c (spu_function_value): Likewise.
18162 (spu_function_arg): Likewise.
18163 * coverage.c (get_gcov_type): Likewise.
18164 (get_gcov_unsigned_t): Likewise.
18165 * dse.c (find_shift_sequence): Likewise.
18166 * expmed.c (store_bit_field_1): Likewise.
18167 * expr.c (convert_move): Likewise.
18168 (store_field): Likewise.
18169 * internal-fn.c (expand_arith_overflow): Likewise.
18170 * optabs-query.c (get_best_extraction_insn): Likewise.
18171 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18172 * stor-layout.c (layout_type): Likewise.
18173 (initialize_sizetypes): Likewise.
18174 * targhooks.c (default_get_mask_mode): Likewise.
18175 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18177 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18178 Alan Hayward <alan.hayward@arm.com>
18179 David Sherwood <david.sherwood@arm.com>
18181 * machmode.h (opt_mode::else_blk): New function.
18182 (int_mode_for_mode): Declare.
18183 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18184 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18186 * cfgexpand.c (expand_debug_expr): Likewise.
18187 * combine.c (gen_lowpart_or_truncate): Likewise.
18188 (gen_lowpart_for_combine): Likewise.
18189 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18190 * config/avr/avr.c (avr_to_int_mode): Likewise.
18191 (avr_out_plus_1): Likewise.
18192 (avr_out_plus): Likewise.
18193 (avr_out_round): Likewise.
18194 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18195 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18196 (s390_expand_vcond): Likewise.
18197 * config/spu/spu.c (spu_split_immediate): Likewise.
18198 (spu_expand_mov): Likewise.
18199 * dse.c (get_stored_val): Likewise.
18200 * expmed.c (store_bit_field_1): Likewise.
18201 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18203 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18204 (extract_low_bits): Likewise.
18205 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18206 handling rather than repeating the check.
18207 (emit_group_store): Likewise.
18208 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18209 * optabs.c (expand_absneg_bit): Likewise.
18210 (expand_copysign_absneg): Likewise.
18211 (expand_copysign_bit): Likewise.
18212 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18213 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18214 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18215 * var-tracking.c (prepare_call_arguments): Likewise.
18216 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18217 int_mode_for_mode instead of mode_for_size.
18218 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18221 Alan Hayward <alan.hayward@arm.com>
18222 David Sherwood <david.sherwood@arm.com>
18224 * machmode.h (int_mode_for_size): New function.
18225 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18226 instead of mode_for_size.
18227 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18229 * combine.c (expand_field_assignment): Use int_mode_for_size
18230 instead of mode_for_size.
18231 (make_extraction): Likewise.
18232 (simplify_shift_const_1): Likewise.
18233 (simplify_comparison): Likewise.
18234 * dojump.c (do_jump): Likewise.
18235 * dwarf2out.c (mem_loc_descriptor): Likewise.
18236 * emit-rtl.c (init_derived_machine_modes): Likewise.
18237 * expmed.c (flip_storage_order): Likewise.
18238 (convert_extracted_bit_field): Likewise.
18239 * expr.c (copy_blkmode_from_reg): Likewise.
18240 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18241 * internal-fn.c (expand_mul_overflow): Likewise.
18242 * lower-subreg.c (simple_move): Likewise.
18243 * optabs-libfuncs.c (init_optabs): Likewise.
18244 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18245 * tree.c (vector_type_mode): Likewise.
18246 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18247 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18248 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18249 * tree-vect-stmts.c (vectorizable_load): Likewise.
18250 (vectorizable_store): Likewise.
18252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18253 Alan Hayward <alan.hayward@arm.com>
18254 David Sherwood <david.sherwood@arm.com>
18256 * coretypes.h (pod_mode): New type.
18257 (scalar_int_mode_pod): New typedef.
18258 * machmode.h (pod_mode): New class.
18259 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18260 * genmodes.c (emit_mode_int_n): Update accordingly.
18261 * lower-subreg.h (target_lower_subreg): Change type to
18262 scalar_int_mode_pod.
18263 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18264 scalar_int_mode_pod.
18266 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18267 Alan Hayward <alan.hayward@arm.com>
18268 David Sherwood <david.sherwood@arm.com>
18270 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18271 machine_mode to scalar_int_mode.
18272 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18273 (rs6000_option_override_internal): Remove cast to int.
18274 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18275 machine_mode to scalar_int_mode.
18276 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18277 (rs6000_option_override_internal): Remove cast to int.
18278 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18279 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18281 * config/s390/s390.c (s390_expand_builtin): Likewise.
18282 * coretypes.h (scalar_int_mode): New type.
18283 (opt_scalar_int_mode): New typedef.
18284 * machmode.h (scalar_int_mode): New class.
18285 (scalar_int_mode::includes_p): New function.
18286 (byte_mode): Change type to scalar_int_mode.
18287 (word_mode): Likewise.
18288 (ptr_mode): Likewise.
18289 * emit-rtl.c (byte_mode): Likewise.
18290 (word_mode): Likewise.
18291 (ptr_mode): Likewise.
18292 (init_derived_machine_modes): Update accordingly.
18293 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18294 and MODE_PARTIAL_INT.
18295 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18296 opt_scalar_int_mode.
18298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18299 Alan Hayward <alan.hayward@arm.com>
18300 David Sherwood <david.sherwood@arm.com>
18302 * target.def (libgcc_floating_mode_supported_p): Take a
18304 * doc/tm.texi: Regenerate.
18305 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18307 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18308 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18312 Alan Hayward <alan.hayward@arm.com>
18313 David Sherwood <david.sherwood@arm.com>
18315 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18316 * doc/tm.texi: Regenerate.
18317 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18318 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18319 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18320 * targhooks.h (default_floatn_mode): Likewise.
18321 * targhooks.c (default_floatn_mode): Likewise.
18322 * tree.c (build_common_tree_nodes): Update accordingly.
18324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18325 Alan Hayward <alan.hayward@arm.com>
18326 David Sherwood <david.sherwood@arm.com>
18328 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18329 (mode_iterator::iterate_p): Likewise.
18330 (mode_iterator::get_wider): Likewise.
18331 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18333 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18334 Alan Hayward <alan.hayward@arm.com>
18335 David Sherwood <david.sherwood@arm.com>
18337 * coretypes.h (opt_scalar_float_mode): New typedef.
18338 * machmode.h (float_mode_for_size): New function.
18339 * emit-rtl.c (double_mode): Delete.
18340 (init_emit_once): Use float_mode_for_size.
18341 * stor-layout.c (layout_type): Likewise.
18342 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18345 Alan Hayward <alan.hayward@arm.com>
18346 David Sherwood <david.sherwood@arm.com>
18348 * output.h (assemble_real): Take a scalar_float_mode.
18349 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18350 * config/arm/arm.md (consttable_4): Likewise.
18351 (consttable_8): Likewise.
18352 (consttable_16): Likewise.
18353 * config/mips/mips.md (consttable_float): Likewise.
18354 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18355 * varasm.c (assemble_real): Take a scalar_float_mode.
18356 (output_constant_pool_2): Update accordingly.
18357 (output_constant): Likewise.
18359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18360 Alan Hayward <alan.hayward@arm.com>
18361 David Sherwood <david.sherwood@arm.com>
18363 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18364 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18365 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18366 (native_encode_real): Likewise.
18367 (native_interpret_real): Likewise.
18368 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18369 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18372 Alan Hayward <alan.hayward@arm.com>
18373 David Sherwood <david.sherwood@arm.com>
18375 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18376 <scalar_float_mode>. Simplify.
18377 (gen_extend_conv_libfunc): Likewise.
18379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18380 Alan Hayward <alan.hayward@arm.com>
18381 David Sherwood <david.sherwood@arm.com>
18383 * coretypes.h (scalar_float_mode): New type.
18384 * machmode.h (mode_traits::from_int): Use machine_mode if
18385 USE_ENUM_MODES is defined.
18386 (is_a): New function.
18388 (dyn_cast): Likewise.
18389 (scalar_float_mode): New class.
18390 (scalar_float_mode::includes_p): New function.
18391 (is_float_mode): Likewise.
18392 * gdbhooks.py (MachineModePrinter): New class.
18393 (build_pretty_printer): Use it for scalar_float_mode.
18394 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18395 (format_helper::format_helper): Turn into a template.
18396 * genmodes.c (get_mode_class): New function.
18397 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18398 or machine_mode if none.
18399 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18400 as_a <scalar_float_mode>.
18401 * dwarf2out.c (mem_loc_descriptor): Likewise.
18402 (insert_float): Likewise.
18403 (add_const_value_attribute): Likewise.
18404 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18405 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18406 (expand_unop): Update accordingly.
18407 (expand_abs_nojump): Likewise.
18408 (expand_copysign_absneg): Take a scalar_float_mode.
18409 (expand_copysign_bit): Likewise.
18410 (expand_copysign): Update accordingly.
18412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18413 Alan Hayward <alan.hayward@arm.com>
18414 David Sherwood <david.sherwood@arm.com>
18416 * coretypes.h (opt_mode): New class.
18417 * machmode.h (opt_mode): Likewise.
18418 (opt_mode::else_void): New function.
18419 (opt_mode::require): Likewise.
18420 (opt_mode::exists): Likewise.
18421 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18422 (GET_MODE_2XWIDER_MODE): Likewise.
18423 (mode_iterator::get_wider): Update accordingly.
18424 (mode_iterator::get_2xwider): Likewise.
18425 (mode_iterator::get_known_wider): Likewise, turning into a template.
18426 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18427 forcing a wider mode to exist.
18428 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18429 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18430 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18431 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18432 * lower-subreg.c (init_lower_subreg): Likewise.
18433 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18434 on the final iteration.
18435 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18436 a wider mode exists before asking for a move pattern.
18437 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18438 forcing a wider mode to exist.
18439 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18440 returning false if no such mode exists.
18441 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18442 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18443 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18444 Avoid checking for a MODE_INT if we already know the mode is not a
18446 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18447 forcing a wider mode to exist.
18448 (expmed_mult_highpart_optab): Likewise.
18449 (expmed_mult_highpart): Likewise.
18450 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18452 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18453 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18454 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18455 * internal-fn.c (expand_mul_overflow): Update use of
18456 GET_MODE_2XWIDER_MODE.
18457 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18458 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18459 GET_MODE_WIDER_MODE.
18460 (convert_plusminus_to_widen): Likewise.
18461 * tree-switch-conversion.c (array_value_type): Likewise.
18462 * var-tracking.c (emit_note_insn_var_location): Likewise.
18463 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18464 Return false inside rather than outside the loop if no wider mode
18466 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18467 and GET_MODE_2XWIDER_MODE
18468 (can_compare_p): Use else_void.
18469 * gdbhooks.py (OptMachineModePrinter): New class.
18470 (build_pretty_printer): Use it for opt_mode.
18472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18473 Alan Hayward <alan.hayward@arm.com>
18474 David Sherwood <david.sherwood@arm.com>
18476 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18477 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18480 Alan Hayward <alan.hayward@arm.com>
18481 David Sherwood <david.sherwood@arm.com>
18483 * machmode.h (mode_traits): New structure.
18484 (get_narrowest_mode): New function.
18485 (mode_iterator::start): Likewise.
18486 (mode_iterator::iterate_p): Likewise.
18487 (mode_iterator::get_wider): Likewise.
18488 (mode_iterator::get_known_wider): Likewise.
18489 (mode_iterator::get_2xwider): Likewise.
18490 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18491 (FOR_EACH_MODE): Likewise.
18492 (FOR_EACH_MODE_FROM): Likewise.
18493 (FOR_EACH_MODE_UNTIL): Likewise.
18494 (FOR_EACH_WIDER_MODE): Likewise.
18495 (FOR_EACH_2XWIDER_MODE): Likewise.
18496 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18497 * combine.c (simplify_comparison): Likewise
18498 * config/i386/i386.c (type_natural_mode): Likewise.
18499 * cse.c (cse_insn): Likewise.
18500 * dse.c (find_shift_sequence): Likewise.
18501 * emit-rtl.c (init_derived_machine_modes): Likewise.
18502 (init_emit_once): Likewise.
18503 * explow.c (hard_function_value): Likewise.
18504 * expmed.c (extract_fixed_bit_field_1): Likewise.
18505 (extract_bit_field_1): Likewise.
18506 (expand_divmod): Likewise.
18507 (emit_store_flag_1): Likewise.
18508 * expr.c (init_expr_target): Likewise.
18509 (convert_move): Likewise.
18510 (alignment_for_piecewise_move): Likewise.
18511 (widest_int_mode_for_size): Likewise.
18512 (emit_block_move_via_movmem): Likewise.
18513 (copy_blkmode_to_reg): Likewise.
18514 (set_storage_via_setmem): Likewise.
18515 (compress_float_constant): Likewise.
18516 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18517 * optabs-query.c (get_best_extraction_insn): Likewise.
18518 * optabs.c (expand_binop): Likewise.
18519 (expand_twoval_unop): Likewise.
18520 (expand_twoval_binop): Likewise.
18521 (widen_leading): Likewise.
18522 (widen_bswap): Likewise.
18523 (expand_parity): Likewise.
18524 (expand_unop): Likewise.
18525 (prepare_cmp_insn): Likewise.
18526 (prepare_float_lib_cmp): Likewise.
18527 (expand_float): Likewise.
18528 (expand_fix): Likewise.
18529 (expand_sfix_optab): Likewise.
18530 * postreload.c (move2add_use_add2_insn): Likewise.
18531 * reg-stack.c (reg_to_stack): Likewise.
18532 * reginfo.c (choose_hard_reg_mode): Likewise.
18533 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18534 * stor-layout.c (mode_for_size): Likewise.
18535 (smallest_mode_for_size): Likewise.
18536 (mode_for_vector): Likewise.
18537 (finish_bitfield_representative): Likewise.
18538 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18539 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18540 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18541 * var-tracking.c (prepare_call_arguments): Likewise.
18543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18544 Alan Hayward <alan.hayward@arm.com>
18545 David Sherwood <david.sherwood@arm.com>
18547 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18548 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18549 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18550 * machmode.h (mode_size): Move earlier in file.
18551 (mode_precision): Likewise.
18552 (mode_inner): Likewise.
18553 (mode_nunits): Likewise.
18554 (mode_unit_size): Likewise.
18555 (unit_unit_precision): Likewise.
18556 (mode_wider): Likewise.
18557 (mode_2xwider): Likewise.
18558 (machine_mode): New class.
18559 (mode_to_bytes): New function.
18560 (mode_to_bits): Likewise.
18561 (mode_to_precision): Likewise.
18562 (mode_to_inner): Likewise.
18563 (mode_to_unit_size): Likewise.
18564 (mode_to_unit_precision): Likewise.
18565 (mode_to_nunits): Likewise.
18566 (GET_MODE_SIZE): Use mode_to_bytes.
18567 (GET_MODE_BITSIZE): Use mode_to_bits.
18568 (GET_MODE_PRECISION): Use mode_to_precision.
18569 (GET_MODE_INNER): Use mode_to_inner.
18570 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18571 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18572 (GET_MODE_NUNITS): Use mode_to_nunits.
18573 * system.h (ALWAYS_INLINE): New macro.
18574 * config/powerpcspe/powerpcspe-c.c
18575 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18576 int for arg1_mode and arg2_mode.
18578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18579 Alan Hayward <alan.hayward@arm.com>
18580 David Sherwood <david.sherwood@arm.com>
18582 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18583 Prefix mode names with E_ in case statements.
18584 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18585 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18586 (aarch64_split_simd_move): Likewise.
18587 (aarch64_gen_storewb_pair): Likewise.
18588 (aarch64_gen_loadwb_pair): Likewise.
18589 (aarch64_gen_store_pair): Likewise.
18590 (aarch64_gen_load_pair): Likewise.
18591 (aarch64_get_condition_code_1): Likewise.
18592 (aarch64_constant_pool_reload_icode): Likewise.
18593 (get_rsqrte_type): Likewise.
18594 (get_rsqrts_type): Likewise.
18595 (get_recpe_type): Likewise.
18596 (get_recps_type): Likewise.
18597 (aarch64_gimplify_va_arg_expr): Likewise.
18598 (aarch64_simd_container_mode): Likewise.
18599 (aarch64_emit_load_exclusive): Likewise.
18600 (aarch64_emit_store_exclusive): Likewise.
18601 (aarch64_expand_compare_and_swap): Likewise.
18602 (aarch64_gen_atomic_cas): Likewise.
18603 (aarch64_emit_bic): Likewise.
18604 (aarch64_emit_atomic_swap): Likewise.
18605 (aarch64_emit_atomic_load_op): Likewise.
18606 (aarch64_evpc_trn): Likewise.
18607 (aarch64_evpc_uzp): Likewise.
18608 (aarch64_evpc_zip): Likewise.
18609 (aarch64_evpc_ext): Likewise.
18610 (aarch64_evpc_rev): Likewise.
18611 (aarch64_evpc_dup): Likewise.
18612 (aarch64_gen_ccmp_first): Likewise.
18613 (aarch64_gen_ccmp_next): Likewise.
18614 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18615 (alpha_emit_xfloating_libcall): Likewise.
18616 (emit_insxl): Likewise.
18617 (alpha_arg_type): Likewise.
18618 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18619 (arc_preferred_simd_mode): Likewise.
18620 (arc_secondary_reload): Likewise.
18621 (get_arc_condition_code): Likewise.
18622 (arc_print_operand): Likewise.
18623 (arc_legitimate_constant_p): Likewise.
18624 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18625 * config/arc/arc.md (casesi_load): Likewise.
18626 (casesi_compact_jump): Likewise.
18627 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18628 (cc_use_register): Likewise.
18629 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18630 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18631 (arm_init_iwmmxt_builtins): Likewise.
18632 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18633 (neon_expand_vector_init): Likewise.
18634 (arm_attr_length_move_neon): Likewise.
18635 (maybe_get_arm_condition_code): Likewise.
18636 (arm_emit_vector_const): Likewise.
18637 (arm_preferred_simd_mode): Likewise.
18638 (arm_output_iwmmxt_tinsr): Likewise.
18639 (thumb1_output_casesi): Likewise.
18640 (thumb2_output_casesi): Likewise.
18641 (arm_emit_load_exclusive): Likewise.
18642 (arm_emit_store_exclusive): Likewise.
18643 (arm_expand_compare_and_swap): Likewise.
18644 (arm_evpc_neon_vuzp): Likewise.
18645 (arm_evpc_neon_vzip): Likewise.
18646 (arm_evpc_neon_vrev): Likewise.
18647 (arm_evpc_neon_vtrn): Likewise.
18648 (arm_evpc_neon_vext): Likewise.
18649 (arm_validize_comparison): Likewise.
18650 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18651 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18652 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18653 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18654 (c6x_preferred_simd_mode): Likewise.
18655 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18656 (epiphany_rtx_costs): Likewise.
18657 * config/epiphany/predicates.md (proper_comparison_operator):
18659 * config/frv/frv.c (condexec_memory_operand): Likewise.
18660 (frv_emit_move): Likewise.
18661 (output_move_single): Likewise.
18662 (output_condmove_single): Likewise.
18663 (frv_hard_regno_mode_ok): Likewise.
18664 (frv_matching_accg_mode): Likewise.
18665 * config/h8300/h8300.c (split_adds_subs): Likewise.
18666 (h8300_rtx_costs): Likewise.
18667 (h8300_print_operand): Likewise.
18668 (compute_mov_length): Likewise.
18669 (output_logical_op): Likewise.
18670 (compute_logical_op_length): Likewise.
18671 (compute_logical_op_cc): Likewise.
18672 (h8300_shift_needs_scratch_p): Likewise.
18673 (output_a_shift): Likewise.
18674 (compute_a_shift_length): Likewise.
18675 (compute_a_shift_cc): Likewise.
18676 (expand_a_rotate): Likewise.
18677 (output_a_rotate): Likewise.
18678 * config/i386/i386.c (classify_argument): Likewise.
18679 (function_arg_advance_32): Likewise.
18680 (function_arg_32): Likewise.
18681 (function_arg_64): Likewise.
18682 (function_value_64): Likewise.
18683 (ix86_gimplify_va_arg): Likewise.
18684 (ix86_legitimate_constant_p): Likewise.
18685 (put_condition_code): Likewise.
18686 (split_double_mode): Likewise.
18687 (ix86_avx256_split_vector_move_misalign): Likewise.
18688 (ix86_expand_vector_logical_operator): Likewise.
18689 (ix86_split_idivmod): Likewise.
18690 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18691 (ix86_build_const_vector): Likewise.
18692 (ix86_build_signbit_mask): Likewise.
18693 (ix86_match_ccmode): Likewise.
18694 (ix86_cc_modes_compatible): Likewise.
18695 (ix86_expand_branch): Likewise.
18696 (ix86_expand_sse_cmp): Likewise.
18697 (ix86_expand_sse_movcc): Likewise.
18698 (ix86_expand_int_sse_cmp): Likewise.
18699 (ix86_expand_vec_perm_vpermi2): Likewise.
18700 (ix86_expand_vec_perm): Likewise.
18701 (ix86_expand_sse_unpack): Likewise.
18702 (ix86_expand_int_addcc): Likewise.
18703 (ix86_split_to_parts): Likewise.
18704 (ix86_vectorize_builtin_gather): Likewise.
18705 (ix86_vectorize_builtin_scatter): Likewise.
18706 (avx_vpermilp_parallel): Likewise.
18707 (inline_memory_move_cost): Likewise.
18708 (ix86_tieable_integer_mode_p): Likewise.
18709 (x86_maybe_negate_const_int): Likewise.
18710 (ix86_expand_vector_init_duplicate): Likewise.
18711 (ix86_expand_vector_init_one_nonzero): Likewise.
18712 (ix86_expand_vector_init_one_var): Likewise.
18713 (ix86_expand_vector_init_concat): Likewise.
18714 (ix86_expand_vector_init_interleave): Likewise.
18715 (ix86_expand_vector_init_general): Likewise.
18716 (ix86_expand_vector_set): Likewise.
18717 (ix86_expand_vector_extract): Likewise.
18718 (emit_reduc_half): Likewise.
18719 (ix86_emit_i387_round): Likewise.
18720 (ix86_mangle_type): Likewise.
18721 (ix86_expand_round_sse4): Likewise.
18722 (expand_vec_perm_blend): Likewise.
18723 (canonicalize_vector_int_perm): Likewise.
18724 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18725 (expand_vec_perm_1): Likewise.
18726 (expand_vec_perm_interleave3): Likewise.
18727 (expand_vec_perm_even_odd_pack): Likewise.
18728 (expand_vec_perm_even_odd_1): Likewise.
18729 (expand_vec_perm_broadcast_1): Likewise.
18730 (ix86_vectorize_vec_perm_const_ok): Likewise.
18731 (ix86_expand_vecop_qihi): Likewise.
18732 (ix86_expand_mul_widen_hilo): Likewise.
18733 (ix86_expand_sse2_abs): Likewise.
18734 (ix86_expand_pextr): Likewise.
18735 (ix86_expand_pinsr): Likewise.
18736 (ix86_preferred_simd_mode): Likewise.
18737 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18738 * config/i386/sse.md (*andnot<mode>3): Likewise.
18739 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18740 (*<code><mode>3): Likewise.
18741 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18742 (ia64_expand_atomic_op): Likewise.
18743 (ia64_arg_type): Likewise.
18744 (ia64_mode_to_int): Likewise.
18745 (ia64_scalar_mode_supported_p): Likewise.
18746 (ia64_vector_mode_supported_p): Likewise.
18747 (expand_vec_perm_broadcast): Likewise.
18748 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18749 (iq2000_function_arg_advance): Likewise.
18750 (iq2000_function_arg): Likewise.
18751 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18752 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18753 (m68k_libcall_value): Likewise.
18754 (m68k_function_value): Likewise.
18755 (sched_attr_op_type): Likewise.
18756 * config/mcore/mcore.c (mcore_output_move): Likewise.
18757 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18759 (microblaze_function_arg): Likewise.
18760 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18761 (mips_print_operand): Likewise.
18762 (mips_mode_ok_for_mov_fmt_p): Likewise.
18763 (mips_vector_mode_supported_p): Likewise.
18764 (mips_preferred_simd_mode): Likewise.
18765 (mips_expand_vpc_loongson_even_odd): Likewise.
18766 (mips_expand_vec_unpack): Likewise.
18767 (mips_expand_vi_broadcast): Likewise.
18768 (mips_expand_vector_init): Likewise.
18769 (mips_expand_vec_reduc): Likewise.
18770 (mips_expand_msa_cmp): Likewise.
18771 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18772 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18773 (cc_flags_for_mode): Likewise.
18774 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18775 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18776 (nds32_output_casesi_pc_relative): Likewise.
18777 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18778 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18779 (nvptx_gen_unpack): Likewise.
18780 (nvptx_gen_pack): Likewise.
18781 (nvptx_gen_shuffle): Likewise.
18782 (nvptx_gen_wcast): Likewise.
18783 (nvptx_preferred_simd_mode): Likewise.
18784 * config/pa/pa.c (pa_secondary_reload): Likewise.
18785 * config/pa/predicates.md (base14_operand): Likewise.
18786 * config/powerpcspe/powerpcspe-c.c
18787 (altivec_resolve_overloaded_builtin): Likewise.
18788 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18790 (rs6000_preferred_simd_mode): Likewise.
18791 (output_vec_const_move): Likewise.
18792 (rs6000_expand_vector_extract): Likewise.
18793 (rs6000_split_vec_extract_var): Likewise.
18794 (reg_offset_addressing_ok_p): Likewise.
18795 (rs6000_legitimate_offset_address_p): Likewise.
18796 (rs6000_legitimize_address): Likewise.
18797 (rs6000_emit_set_const): Likewise.
18798 (rs6000_const_vec): Likewise.
18799 (rs6000_emit_move): Likewise.
18800 (spe_build_register_parallel): Likewise.
18801 (rs6000_darwin64_record_arg_recurse): Likewise.
18802 (swap_selector_for_mode): Likewise.
18803 (spe_init_builtins): Likewise.
18804 (paired_init_builtins): Likewise.
18805 (altivec_init_builtins): Likewise.
18806 (do_load_for_compare): Likewise.
18807 (rs6000_generate_compare): Likewise.
18808 (rs6000_expand_float128_convert): Likewise.
18809 (emit_load_locked): Likewise.
18810 (emit_store_conditional): Likewise.
18811 (rs6000_output_function_epilogue): Likewise.
18812 (rs6000_handle_altivec_attribute): Likewise.
18813 (rs6000_function_value): Likewise.
18814 (emit_fusion_gpr_load): Likewise.
18815 (emit_fusion_p9_load): Likewise.
18816 (emit_fusion_p9_store): Likewise.
18817 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
18818 (fusion_gpr_mem_load): Likewise.
18819 (fusion_addis_mem_combo_load): Likewise.
18820 (fusion_addis_mem_combo_store): Likewise.
18821 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
18822 (fusion_gpr_mem_load): Likewise.
18823 (fusion_addis_mem_combo_load): Likewise.
18824 (fusion_addis_mem_combo_store): Likewise.
18825 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18827 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
18828 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18829 (rs6000_preferred_simd_mode): Likewise.
18830 (output_vec_const_move): Likewise.
18831 (rs6000_expand_vector_extract): Likewise.
18832 (rs6000_split_vec_extract_var): Likewise.
18833 (reg_offset_addressing_ok_p): Likewise.
18834 (rs6000_legitimate_offset_address_p): Likewise.
18835 (rs6000_legitimize_address): Likewise.
18836 (rs6000_emit_set_const): Likewise.
18837 (rs6000_const_vec): Likewise.
18838 (rs6000_emit_move): Likewise.
18839 (rs6000_darwin64_record_arg_recurse): Likewise.
18840 (swap_selector_for_mode): Likewise.
18841 (paired_init_builtins): Likewise.
18842 (altivec_init_builtins): Likewise.
18843 (rs6000_expand_float128_convert): Likewise.
18844 (emit_load_locked): Likewise.
18845 (emit_store_conditional): Likewise.
18846 (rs6000_output_function_epilogue): Likewise.
18847 (rs6000_handle_altivec_attribute): Likewise.
18848 (rs6000_function_value): Likewise.
18849 (emit_fusion_gpr_load): Likewise.
18850 (emit_fusion_p9_load): Likewise.
18851 (emit_fusion_p9_store): Likewise.
18852 * config/rx/rx.c (rx_gen_move_template): Likewise.
18853 (flags_from_mode): Likewise.
18854 * config/s390/predicates.md (s390_alc_comparison): Likewise.
18855 (s390_slb_comparison): Likewise.
18856 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
18857 (s390_vector_mode_supported_p): Likewise.
18858 (s390_cc_modes_compatible): Likewise.
18859 (s390_match_ccmode_set): Likewise.
18860 (s390_canonicalize_comparison): Likewise.
18861 (s390_emit_compare_and_swap): Likewise.
18862 (s390_branch_condition_mask): Likewise.
18863 (s390_rtx_costs): Likewise.
18864 (s390_secondary_reload): Likewise.
18865 (__SECONDARY_RELOAD_CASE): Likewise.
18866 (s390_expand_cs): Likewise.
18867 (s390_preferred_simd_mode): Likewise.
18868 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
18869 * config/sh/sh.c (sh_print_operand): Likewise.
18870 (dump_table): Likewise.
18871 (sh_secondary_reload): Likewise.
18872 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18873 * config/sh/sh.md (casesi_worker_1): Likewise.
18874 (casesi_worker_2): Likewise.
18875 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
18876 (fcc_comparison_operator): Likewise.
18877 * config/sparc/sparc.c (sparc_expand_move): Likewise.
18878 (emit_soft_tfmode_cvt): Likewise.
18879 (sparc_preferred_simd_mode): Likewise.
18880 (output_cbranch): Likewise.
18881 (sparc_print_operand): Likewise.
18882 (sparc_expand_vec_perm_bmask): Likewise.
18883 (vector_init_bshuffle): Likewise.
18884 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18885 (spu_vector_mode_supported_p): Likewise.
18886 (spu_expand_insv): Likewise.
18887 (spu_emit_branch_or_set): Likewise.
18888 (spu_handle_vector_attribute): Likewise.
18889 (spu_builtin_splats): Likewise.
18890 (spu_builtin_extract): Likewise.
18891 (spu_builtin_promote): Likewise.
18892 (spu_expand_sign_extend): Likewise.
18893 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18894 (tilegx_simd_int): Likewise.
18895 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
18896 (tilepro_simd_int): Likewise.
18897 * config/v850/v850.c (const_double_split): Likewise.
18898 (v850_print_operand): Likewise.
18899 (ep_memory_offset): Likewise.
18900 * config/vax/vax.c (vax_rtx_costs): Likewise.
18901 (vax_output_int_move): Likewise.
18902 (vax_output_int_add): Likewise.
18903 (vax_output_int_subtract): Likewise.
18904 * config/visium/predicates.md (visium_branch_operator): Likewise.
18905 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
18906 (visium_print_operand_address): Likewise.
18907 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18908 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
18909 (xtensa_expand_conditional_branch): Likewise.
18910 (xtensa_copy_incoming_a7): Likewise.
18911 (xtensa_output_literal): Likewise.
18912 * dfp.c (decimal_real_maxval): Likewise.
18913 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18915 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18916 Alan Hayward <alan.hayward@arm.com>
18917 David Sherwood <david.sherwood@arm.com>
18919 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
18920 (mode_nunits_inline): Likewise.
18921 (mode_inner_inline): Likewise.
18922 (mode_unit_size_inline): Likewise.
18923 (mode_unit_precision_inline): Likewise.
18924 (emit_insn_modes_h): Likewise. Also emit a #define of the
18926 (emit_mode_wider): Add an E_ prefix to mode names.
18927 (emit_mode_complex): Likewise.
18928 (emit_mode_inner): Likewise.
18929 (emit_mode_adjustments): Likewise.
18930 (emit_mode_int_n): Likewise.
18931 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
18932 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
18933 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
18934 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
18935 (CRC32_BUILTIN, ENTRY): Likewise.
18936 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
18937 (aarch64_pop_regs): Likewise.
18938 (aarch64_process_components): Likewise.
18939 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
18940 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
18941 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
18942 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
18943 * config/arm/arm.c (arm_init_libfuncs): Likewise.
18944 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
18946 * config/i386/i386-builtin.def (pcmpestr): Likewise.
18947 (pcmpistr): Likewise.
18948 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
18949 * config/mmix/mmix.c (mmix_output_condition): Likewise.
18950 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18952 * config/rl78/rl78.c (mduc_regs): Likewise.
18953 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18954 (htm_expand_builtin): Likewise.
18955 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
18956 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
18957 * config/xtensa/xtensa.c (print_operand): Likewise.
18958 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
18959 (NUM_MODE_VECTOR_INT): Likewise.
18960 * genoutput.c (null_operand): Likewise.
18961 (output_operand_data): Likewise.
18962 * genrecog.c (print_parameter_value): Likewise.
18963 * lra.c (debug_operand_data): Likewise.
18965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18967 * dwarf2out.c (scompare_loc_descriptor_wide)
18968 (scompare_loc_descriptor_narrow): New functions, split out from...
18969 (scompare_loc_descriptor): ...here.
18970 * expmed.c (emit_store_flag_int): New function, split out from...
18971 (emit_store_flag): ...here.
18973 2017-08-30 Richard Biener <rguenther@suse.de>
18975 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
18976 (dwarf2out_early_finish): Move setting of AT_pubnames from
18977 early debug output to early finish.
18979 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18981 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
18982 and -mdata-region to the assembler.
18984 2017-08-30 Richard Biener <rguenther@suse.de>
18986 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
18988 (gen_subprogram_die): Add DW_AT_object_pointer only early.
18989 (dwarf2out_early_global_decl): Only generate a DIE for the
18990 abstract origin if it doesn't already exist or is a declaration DIE.
18991 (resolve_addr): Do not add the linkage name twice when
18992 generating a stub DIE for the DW_TAG_GNU_call_site target.
18994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18996 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18997 Use machine_mode rather than int for arg1_mode.
18999 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19002 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19003 that the second argument of the built-in functions to unpack
19004 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19005 switch statement instead a lot of if statements.
19006 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19007 Allow 64-bit values to be in Altivec registers as well as
19008 traditional floating point registers.
19009 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19011 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19013 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19014 MAX_REGS_PER_ADDRESS == 1.
19016 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19018 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19019 * config/i386/i386.c (ix86_option_override_internal): Simplify
19020 setting of opts->x_flag_entry.
19022 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19023 Jakub Jelinek <jakub@redhat.com>
19024 Richard Biener <rguenther@suse.de>
19026 PR tree-optimization/81503
19027 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19028 folded constant fits in the target type; reorder tests for clarity.
19030 2017-08-29 Martin Liska <mliska@suse.cz>
19032 * passes.def: Include pass_lower_switch.
19033 * stmt.c (dump_case_nodes): Remove and move to
19034 tree-switch-conversion.
19035 (case_values_threshold): Likewise.
19036 (expand_switch_as_decision_tree_p): Likewise.
19037 (emit_case_decision_tree): Likewise.
19038 (expand_case): Likewise.
19039 (balance_case_nodes): Likewise.
19040 (node_has_low_bound): Likewise.
19041 (node_has_high_bound): Likewise.
19042 (node_is_bounded): Likewise.
19043 (emit_case_nodes): Likewise.
19044 (struct simple_case_node): New struct.
19045 (add_case_node): Remove.
19046 (emit_case_dispatch_table): Use vector instead of case_list.
19047 (reset_out_edges_aux): Remove.
19048 (compute_cases_per_edge): Likewise.
19049 (expand_case): Build list of simple_case_node.
19050 (expand_sjlj_dispatch_table): Use it.
19051 * tree-switch-conversion.c (struct case_node): Moved from
19052 stmt.c and adjusted.
19053 (emit_case_nodes): Likewise.
19054 (node_has_low_bound): Likewise.
19055 (node_has_high_bound): Likewise.
19056 (node_is_bounded): Likewise.
19057 (case_values_threshold): Likewise.
19058 (reset_out_edges_aux): Likewise.
19059 (compute_cases_per_edge): Likewise.
19060 (add_case_node): Likewise.
19061 (dump_case_nodes): Likewise.
19062 (balance_case_nodes): Likewise.
19063 (expand_switch_as_decision_tree_p): Likewise.
19064 (emit_jump): Likewise.
19065 (emit_case_decision_tree): Likewise.
19066 (try_switch_expansion): Likewise.
19067 (do_jump_if_equal): Likewise.
19068 (emit_cmp_and_jump_insns): Likewise.
19069 (fix_phi_operands_for_edge): New function.
19070 (record_phi_operand_mapping): Likewise.
19071 (class pass_lower_switch): New pass.
19072 (pass_lower_switch::execute): New function.
19073 (make_pass_lower_switch): Likewise.
19074 (conditional_probability):
19075 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19076 * tree-pass.h: Add make_pass_lower_switch.
19078 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19081 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19084 2017-08-29 Richard Biener <rguenther@suse.de>
19086 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19087 we do not add a DW_AT_inline attribute twice.
19088 (gen_subprogram_die): Remove code setting DW_AT_inline on
19089 DECL_ABSTRACT_P nodes.
19091 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19093 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19094 calls to internal functions.
19095 (gimplify_modify_expr): Likewise.
19096 * tree-call-cdce.c (use_internal_fn): Likewise.
19097 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19098 (convert_to_divmod): Set the nothrow flag.
19099 * tree-if-conv.c (predicate_mem_writes): Likewise.
19100 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19101 (vectorizable_call): Likewise.
19102 (vectorizable_store): Likewise.
19103 (vectorizable_load): Likewise.
19104 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19105 (vect_recog_mask_conversion_pattern): Likewise.
19107 2017-08-29 Martin Liska <mliska@suse.cz>
19110 * gcc.c (driver_handle_option): Add new argument.
19111 * opts-common.c (handle_option): Pass
19112 target_option_override_hook.
19113 * opts-global.c (lang_handle_option): Add new option.
19114 (set_default_handlers): Add new argument.
19115 (decode_options): Likewise.
19116 * opts.c (target_handle_option): Likewise.
19117 (common_handle_option): Call target_option_override_hook.
19118 * opts.h (struct cl_option_handler_func): Add hook for
19119 target option override.
19120 (struct cl_option_handlers): Likewise.
19121 (set_default_handlers): Add new argument.
19122 (decode_options): Likewise.
19123 (common_handle_option): Likewise.
19124 (target_handle_option): Likewise.
19125 * toplev.c (toplev::main): Pass targetm.target_option.override
19128 2017-08-29 Richard Biener <rguenther@suse.de>
19129 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19131 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19132 life to the active subtree.
19134 2017-08-28 Jeff Law <law@redhat.com>
19136 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19137 derive_equivalences.
19138 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19139 Code moved into....
19140 (edge_info::derive_equivalences): New private member function
19142 * tree-ssa-dom.c (class edge_info): Changed from a struct
19143 to a class. Add ctor/dtor, methods and data members.
19144 (edge_info::edge_info): Renamed from allocate_edge_info.
19145 Initialize additional members.
19146 (edge_info::~edge_info): New.
19147 (free_dom_edge_info): Delete the edge info.
19148 (record_edge_info): Use new class & associated member functions.
19149 Tighten forms for testing for edge equivalences.
19150 (record_temporary_equivalences): Iterate over the simple
19151 equivalences rather than assuming there's only one per edge.
19152 (cprop_into_successor_phis): Iterate over the simple
19153 equivalences rather than assuming there's only one per edge.
19154 (optimize_stmt): Use operand_equal_p rather than pointer
19155 equality for mini-DSE code.
19157 2017-08-28 Nathan Sidwell <nathan@acm.org>
19159 * gcc.c (execute): Fold SIGPIPE handling into switch
19160 statement. Adjust internal error message.
19162 2017-08-28 Richard Biener <rguenther@suse.de>
19165 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19166 Do nothing for removed DIEs.
19168 2017-08-28 Richard Biener <rguenther@suse.de>
19170 PR tree-optimization/81977
19171 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19174 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19177 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19178 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19179 using targetm.gen_mem_thread_fence.
19181 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19184 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19185 predicate to register_operand. Reorder operands.
19186 (*btr<mode>): Ditto.
19187 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19188 (*btr<mode>_mask): Ditto.
19190 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19192 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19193 * config/rs6000/xmmintrin.h: New file.
19194 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19196 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19199 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19200 parameter and_insn and return it.
19201 (recombine_lvx_pattern): Insert a copy to ensure availability of
19202 the base register of the copied masking operation at the point of
19203 the instruction replacement.
19204 (recombine_stvx_pattern): Likewise.
19206 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19208 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19209 undocumented switches.
19210 (-mpower9-dform-vector): Likewise.
19211 (-mpower9-dform): Likewise.
19212 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19213 comments to delete references to -mpower9-dform* switches.
19214 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19215 Delete reference to -mpower9-dform* switches, test for
19216 -mpower9-vector instead.
19217 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19218 (OTHER_P9_VECTOR_MASKS): Likewise.
19219 (POWERPC_MASKS): Likewise.
19220 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19221 tests against -mpower9-dform* to -mpower9-vector. Delete code
19222 that checked for -mpower9-dform* consistancy with other options.
19223 Add test for -mpower9-misc to enable other power9 switches.
19224 (rs6000_init_hard_regno_mode_ok): Likewise.
19225 (rs6000_option_override_internal): Likewise.
19226 (rs6000_emit_prologue): Likewise.
19227 (rs6000_emit_epilogue): Likewise.
19228 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19229 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19230 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19232 (emit_fusion_p9_store): Likewise.
19233 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19234 resetting these macros if the assembler does not support ISA 3.0
19236 (TARGET_P9_DFORM_VECTOR): Likewise.
19237 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19238 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19240 2017-08-25 Alan Modra <amodra@gmail.com>
19242 PR rtl-optimization/81747
19243 * cse.c (cse_extended_basic_block): Don't attempt to record
19244 equivalences for degenerate conditional branches that branch
19245 to their fall-through.
19247 2017-08-24 Martin Sebor <msebor@redhat.com>
19249 PR middle-end/81908
19250 * gimple-fold.c (size_must_be_zero_p): New function.
19251 (gimple_fold_builtin_memory_op): Call it.
19253 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19255 * config/rs6000/mm_malloc.h: New file.
19257 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19259 PR tree-optimization/81913
19260 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19261 analysis when either IVs in condition can wrap.
19263 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19265 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19266 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19268 2017-08-24 Richard Biener <rguenther@suse.de>
19271 * targhooks.c (default_target_can_inline_p): Properly
19272 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19273 is present and always compare.
19274 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19275 infer -mfpmath=sse from TARGET_SSE_P.
19276 (ix86_can_inline_p): Properly use target_option_default_node when
19277 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19279 2017-08-24 Richard Biener <rguenther@suse.de>
19282 * dwarf2out.c (output_die): Handle flag_generate_offload like
19284 (output_comp_unit): Likewise.
19285 (gen_array_type_die): Likewise.
19286 (dwarf2out_early_finish): Likewise.
19287 (note_variable_value_in_expr): Likewise.
19288 (dwarf2out_finish): Likewise. Adjust assert.
19289 * cgraphunit.c (symbol_table::compile): Move setting of
19290 flag_generate_offload earlier ...
19291 (symbol_table::finalize_compilation_unit): ... here, before
19292 early debug finalization.
19294 2017-08-24 Richard Biener <rguenther@suse.de>
19296 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19297 and ipa-fnsummary.h.
19298 (ix86_can_inline_p): When ix86_fpmath flags do not match
19299 check whether the callee uses FP math at all.
19301 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19303 PR middle-end/81931
19304 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19305 instead of TYPE_PRECISION.
19307 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19308 Alan Hayward <alan.hayward@arm.com>
19309 David Sherwood <david.sherwood@arm.com>
19311 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19313 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19315 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19317 (ix86_frame::stack_realign_allocate): New field.
19318 (struct machine_frame_state): Modify comments.
19319 (machine_frame_state::sp_realigned_fp_end): New field.
19320 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19321 layout calculation.
19322 (sp_valid_at): Add assertion to assure no attempt to access invalid
19323 offset of a realigned stack.
19324 (fp_valid_at): Likewise.
19325 (choose_baseaddr): Modify comments.
19326 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19327 ix86_expand_prologue.
19328 (ix86_expand_prologue): Modify stack realignment and allocation.
19329 (ix86_expand_epilogue): Modify comments.
19330 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19331 avx2_runtime, avx512f, and avx512f_runtime.
19333 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19335 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19336 (mstackrealign): Do not init to -1.
19337 * config/i386/i386.c (ix86_option_override_internal):
19338 Check opts_set, not opts when setting default value of
19339 opts->x_ix86_force_align_arg_pointer.
19341 2017-08-23 Richard Biener <rguenther@suse.de>
19343 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19344 lang_hooks.decl_printable_name.
19345 * print-rtl-function.c (print_rtx_function): Likewise.
19346 * tree-pretty-print.c (dump_function_header): Likewise.
19348 2017-08-23 Richard Biener <rguenther@suse.de>
19351 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19352 -g0 at compile-time.
19354 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19356 PR middle-end/19706
19357 * doc/sourcebuild.texi (Other hardware attributes):
19360 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19362 PR middle-end/19706
19363 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19364 Fix single-use check.
19366 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19368 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19370 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19372 * doc/install.texi: Modify to add more details on running selected
19375 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19377 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19378 is combined with -mabi=ms.
19379 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19382 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19384 PR tree-optimization/81488
19385 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19386 and cached_basis fields.
19387 (MAX_SPREAD): New constant.
19388 (alloc_cand_and_find_basis): Initialize new fields.
19389 (clear_visited): New function.
19390 (create_phi_basis_1): Rename from create_phi_basis, set visited
19391 and cached_basis fields.
19392 (create_phi_basis): New wrapper function.
19393 (phi_add_costs_1): Rename from phi_add_costs, add spread
19394 parameter, set visited field, short-circuit when limits reached.
19395 (phi_add_costs): New wrapper function.
19396 (record_phi_increments_1): Rename from record_phi_increments, set
19398 (record_phi_increments): New wrapper function.
19399 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19400 (phi_incr_cost): New wrapper function.
19401 (all_phi_incrs_profitable_1): Rename from
19402 all_phi_incrs_profitable, set visited field.
19403 (all_phi_incrs_profitable): New wrapper function.
19405 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19406 Alan Hayward <alan.hayward@arm.com>
19407 David Sherwood <david.sherwood@arm.com>
19409 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19410 that takes the outer and inner modes.
19411 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19412 comparison as the canonical test for a paradoxical subreg.
19413 * combine.c (simplify_set): Use paradoxical_subreg_p.
19414 (make_extraction): Likewise.
19415 (force_to_mode): Likewise.
19416 (rtx_equal_for_field_assignment_p): Likewise.
19417 (gen_lowpart_for_combine): Likewise.
19418 (simplify_comparison): Likewise.
19419 * cse.c (equiv_constant): Likewise.
19420 * expmed.c (store_bit_field_1): Likewise.
19421 * final.c (alter_subreg): Likewise.
19422 * fwprop.c (propagate_rtx): Likewise.
19423 (forward_propagate_subreg): Likewise.
19424 * ira-conflicts.c (ira_build_conflicts): Likewise.
19425 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19426 * lra-constraints.c (curr_insn_transform): Likewise.
19427 (split_reg): Likewise.
19428 * lra-eliminations.c (move_plus_up): Likewise.
19429 (lra_eliminate_regs_1): Likewise.
19430 * recog.c (general_operand): Likewise.
19431 * ree.c (combine_reaching_defs): Likewise.
19432 * reload.c (push_reload): Likewise.
19433 (find_reloads): Likewise.
19434 * reload1.c (elimination_effects): Likewise.
19435 (compute_reload_subreg_offset): Likewise.
19436 (choose_reload_regs): Likewise.
19437 * rtlanal.c (subreg_lsb_1): Likewise.
19438 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19439 (simplify_subreg): Likewise.
19440 * var-tracking.c (track_loc_p): Likewise.
19441 * emit-rtl.c (byte_lowpart_offset): Likewise.
19442 (paradoxical_subreg_p): Delete out-of-line definition.
19444 2017-08-22 Jeff Law <law@redhat.com>
19446 PR tree-optimization/81741
19447 PR tree-optimization/71947
19448 * tree-ssa-dom.c: Include tree-inline.h.
19449 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19450 equivalences if one is more expensive to compute than the other.
19451 * tree-ssa-scopedtables.h (class const_or_copies): Make
19452 record_const_or_copy_raw method private.
19453 (class avail_exprs_stack): New method simplify_binary_operation.
19454 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19455 avail_exprs_stack::simplify_binary_operation as needed.
19456 (avail_exprs_stack::simplify_binary_operation): New function.
19458 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19460 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19461 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19462 (DOT_SYMBOLS): Likewise.
19463 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19464 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19465 (RS6000_ABI_NAME): Likewise.
19466 (TARGET_CMODEL): Likewise.
19467 (TOC_SECTION_ASM_OP): Likewise.
19468 (SET_CMODEL): New macro.
19469 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19471 2017-08-22 Richard Biener <rguenther@suse.de>
19473 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19474 to allow for free-lang-data replacements similar to verify_type_variant.
19476 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19477 Alan Hayward <alan.hayward@arm.com>
19478 David Sherwood <david.sherwood@arm.com>
19480 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19481 VOIDmode for the LABEL_REF.
19483 2017-08-22 Richard Biener <rguenther@suse.de>
19485 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19487 2017-08-22 Richard Biener <rguenther@suse.de>
19489 * common.opt (feliminate-dwarf2-dups): Ignore.
19490 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19491 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19492 same_die_p_wrap, compute_section_prefix,
19493 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19494 (comdat_symbol_id, comdat_symbol_number): Likewise.
19495 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19497 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19498 (output_die): Mark unreachable path unreachable.
19499 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19500 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19501 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19502 (dwarf2out_early_finish): Likewise.
19504 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19506 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19508 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19511 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19512 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19513 (avr_attribute_table) <io, io_low, address>: Initialize
19514 .decl_required with true.
19516 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19518 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19519 undocumented debugging options.
19520 (-mvsx-scalar-double): Likewise.
19521 (-mallow-df-permute): Likewise.
19522 (-mvectorize-builtins): Likewise.
19523 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19524 (rs6000_builtin_vectorized_function): Likewise.
19525 (rs6000_builtin_md_vectorized_function): Likewise.
19526 (rs6000_opt_vars): Likewise.
19528 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19531 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19532 (*btrq_imm): Rename from *btrq.
19533 (*btcq_imm): Rename from *btcq.
19534 (btsc): New code attribute.
19535 (*<btsc><mode>): New insn pattern.
19536 (*btr<mode>): Ditto.
19537 (*<btsc><mode>_mask): New insn_and_split pattern.
19538 (*btr<mode>_mask): Ditto.
19540 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19541 Alan Hayward <alan.hayward@arm.com>
19542 David Sherwood <david.sherwood@arm.com>
19544 * function.c (pad_below): Simplify padding calculation.
19546 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19547 Alan Hayward <alan.hayward@arm.com>
19548 David Sherwood <david.sherwood@arm.com>
19550 * target.def (function_prologue): Remove frame size argument.
19551 (function_epilogue): Likewise.
19552 * doc/tm.texi: Regenerate.
19553 * final.c (final_start_function): Update call to function_prologue.
19554 (final_end_function): Update call to function_epilogue.
19555 (default_function_pro_epilogue): Remove frame size argument.
19556 * output.h (default_function_pro_epilogue): Likewise.
19557 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19558 (arm_output_function_prologue): Likewise.
19559 * config/frv/frv.c (frv_function_prologue): Likewise.
19560 (frv_function_epilogue): Likewise.
19561 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19562 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19563 (ia64_output_function_epilogue): Likewise.
19564 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19565 (m32r_output_function_epilogue): Likewise.
19566 * config/microblaze/microblaze.c (microblaze_function_prologue)
19567 (microblaze_function_epilogue): Likewise.
19568 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19569 (mips_output_function_epilogue): Likewise.
19570 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19571 (mmix_target_asm_function_epilogue): Likewise.
19572 * config/msp430/msp430.c (msp430_start_function): Likewise.
19573 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19574 (nds32_asm_function_epilogue): Likewise.
19575 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19576 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19577 (pa_output_function_epilogue): Likewise.
19578 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19579 (rs6000_output_function_epilogue): Likewise.
19580 * config/rl78/rl78.c (rl78_start_function): Likewise.
19581 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19582 (rs6000_output_function_epilogue): Likewise.
19583 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19584 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19585 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19586 (sparc_asm_function_epilogue): Likewise.
19588 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19590 * tree.h (type_has_mode_precision_p): New function.
19591 * convert.c (convert_to_integer_1): Use it.
19592 * expr.c (expand_expr_real_2): Likewise.
19593 (expand_expr_real_1): Likewise.
19594 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19595 * match.pd: Likewise.
19596 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19597 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19598 * tree-tailcall.c (process_assignment): Likewise.
19599 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19600 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19601 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19602 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19603 (vectorizable_assignment): Likewise.
19604 (vectorizable_shift): Likewise.
19605 (vectorizable_operation): Likewise.
19606 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19608 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19610 * match.pd: Add pow (C, x) simplification.
19612 2017-08-21 Richard Biener <rguenther@suse.de>
19614 PR tree-optimization/81900
19615 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19616 for blocks with abnormal predecessors.
19617 (compute_antic): Do not set visited flag prematurely.
19619 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19622 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19624 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19626 * stor-layout.h (vector_type_mode): Move to...
19627 * tree.h (vector_type_mode): ...here.
19628 * stor-layout.c (vector_type_mode): Move to...
19629 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19631 2017-08-21 Richard Biener <rguenther@suse.de>
19633 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19634 register_external_die hooks.
19635 (debug_false_tree_charstarstar_uhwistar): Declare.
19636 (debug_nothing_tree_charstar_uhwi): Likewise.
19637 * debug.c (do_nothing_debug_hooks): Adjust.
19638 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19639 (debug_nothing_tree_charstar_uhwi): Likewise.
19640 * dbxout.c (dbx_debug_hooks): Adjust.
19641 (xcoff_debug_hooks): Likewise.
19642 * sdbout.c (sdb_debug_hooks): Likewise.
19643 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19644 * dwarf2out.c (macinfo_label_base): New global.
19645 (dwarf2out_register_external_die): New function for the
19646 register_external_die hook.
19647 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19648 (dwarf2_debug_hooks): Use them.
19649 (dwarf2_lineno_debug_hooks): Adjust.
19650 (struct die_struct): Add with_offset flag.
19651 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19652 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19653 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19654 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19655 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19656 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19657 defining section names for the early LTO debug variants.
19658 (reset_indirect_string): New helper.
19659 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19660 (print_dw_val): Add support for offsetted symbol references.
19661 (get_ultimate_context): Split out from is_cxx.
19662 (is_cxx): Use get_ultimate_context.
19663 (is_fortran): Add decl overload.
19664 (compute_comp_unit_symbol): Split out worker from
19665 compute_section_prefix.
19666 (compute_section_prefix): Call compute_comp_unit_symbol and
19667 set comdat_type_p here.
19668 (output_die): Skip DIE symbol output for the LTO added one.
19669 Handle DIE symbol references with offset.
19670 (output_comp_unit): Guard section name mangling properly.
19671 For LTO debug sections emit a symbol at the section beginning
19672 which we use to refer to its DIEs.
19673 (add_abstract_origin_attribute): For DIEs registered via
19674 dwarf2out_register_external_die directly refer to the early
19675 DIE rather than indirectly through the shadow one we created.
19676 Remove obsolete call to dwarf2out_abstract_function for
19677 non-function/block origins.
19678 (gen_array_type_die): When generating early LTO debug do
19679 not emit DW_AT_string_length.
19680 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19681 late when in LTO. As suggested place a gcc_unreachable for
19682 the DECL_ABSTRACT_P case.
19683 (gen_subprogram_die): Avoid another specification DIE
19684 for early built declarations/definitions for the late LTO case.
19685 (gen_variable_die): Add type references for late duplicated VLA dies
19687 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19688 we have the abstract instance already.
19689 (process_scope_var): Adjust decl DIE contexts in LTO which
19690 first puts them in limbo.
19691 (gen_decl_die): Do not generate type DIEs late apart from
19692 types for VLAs or for decls we do not yet have a DIE. Do not
19693 call dwarf2out_abstract_function late.
19694 (dwarf2out_early_global_decl): Make sure to create DIEs
19695 for abstract instances of a decl first.
19696 (dwarf2out_late_global_decl): Adjust comment.
19697 (output_macinfo_op): With multiple macro sections use
19698 macinfo_label_base to distinguish labels.
19699 (output_macinfo): Likewise. Update macinfo_label_base.
19700 Pass in the line info label.
19701 (note_variable_value_in_expr): When generating LTO resolve
19702 all variable values here by generating DIEs as needed.
19703 (init_sections_and_labels): Add early LTO debug flag parameter
19704 and generate different sections and names if set. Add generation
19705 counter for the labels so we can have multiple of them.
19706 (reset_dies): Helper to allow DIEs to be output multiple times.
19707 (dwarf2out_finish): When outputting DIEs to the fat part of an
19708 LTO object first reset DIEs.
19709 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19710 (modified_type_die): Check for decl_ultimate_origin being self
19712 (gen_type_die_with_usage): Likewise.
19713 (gen_typedef_die): Allow decl_ultimate_origin being self.
19714 (set_decl_abstract_flags): Remove.
19715 (set_block_abstract_flags): Likewise.
19716 (dwarf2out_abstract_function): Treat the early generated DIEs
19717 as the abstract copy and only add DW_AT_inline and
19718 DW_AT_artificial here and call set_decl_origin_self.
19719 If the DIE has an abstract origin don't do anything.
19720 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19721 if we have none yet (Go fails to build one, PR78628).
19722 (variably_modified_type_p): Prevent endless recursion for Ada
19723 cyclic pointer types.
19724 * lto-streamer-in.c: Include debug.h.
19725 (dref_queue): New global.
19726 (lto_read_tree_1): Stream in DIE references.
19727 (lto_input_tree): Register DIE references.
19728 (input_function): Stream DECL_DEBUG_ARGS.
19729 * lto-streamer-out.c: Include debug.h.
19730 (lto_write_tree_1): Output DIE references.
19731 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19732 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19733 (output_function): Stream DECL_DEBUG_ARGS.
19734 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19735 Stream DECL_ABSTRACT_ORIGIN.
19736 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19737 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19738 DECL_CONTEXT for file-scope decls.
19739 * lto-streamer.h (struct dref_entry): Declare.
19740 (dref_queue): Likewise.
19741 * cfgexpand.c (pass_expand::execute): Do not call the
19742 outlining_inline_function hook here.
19743 * lto-wrapper.c (debug_obj): New global.
19744 (tool_cleanup): Unlink it if required.
19745 (debug_objcopy): New function.
19746 (run_gcc): Handle early debug sections in the IL files by
19747 extracting them to separate files, partially linkin them and
19748 feeding the result back as result to the linker.
19749 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19750 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19751 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19752 sections into a separate segment.
19753 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19755 (darwin_asm_dwarf_section): Likewise.
19756 (darwin_asm_output_dwarf_offset): Likewise.
19757 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19759 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19760 Alan Hayward <alan.hayward@arm.com>
19761 David Sherwood <david.sherwood@arm.com>
19763 * read-md.h (md_reader::record_potential_iterator_use): Replace
19764 pointer argument with an rtx and an index.
19765 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19766 (apply_mode_iterator): Likewise.
19767 (apply_code_iterator): Likewise.
19768 (apply_int_iterator): Likewise.
19769 (apply_subst_iterator): Likewise.
19770 (record_iterator_use): Likewise.
19771 (record_attribute_use): Likewise.
19772 (md_reader::record_potential_iterator_use): Likewise. Update calls
19773 to record_iterator_use and apply_iterator.
19774 (iterator_use): Replace ptr with x and index.
19775 (attribute_use): Likewise.
19776 (apply_attribute_uses): Update calls to apply_iterator.
19777 (apply_iterators): Likewise. Update initialization of iterator_use.
19778 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19779 and record_potential_iterator_use.
19780 (rtx_reader::read_rtx_operand): Likewise.
19782 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19783 Alan Hayward <alan.hayward@arm.com>
19784 David Sherwood <david.sherwood@arm.com>
19786 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19789 2017-08-21 Richard Biener <rguenther@suse.de>
19791 PR middle-end/81884
19792 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19793 at struct end conservatively when comparing common bases.
19795 2017-08-21 Richard Biener <rguenther@suse.de>
19797 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
19798 (mem_ref_in_stmt): Remove.
19799 (determine_max_movement): Use ref index to get at the reference.
19800 (invariantness_dom_walker::before_dom_children): Deal with
19801 lim data already initialized.
19802 (gather_mem_refs_stmt): Initialize lim data and record ref index.
19804 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19806 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
19807 (TARGET_ISA_ROUND): Ditto.
19808 (TARGET_ROUND): Ditto.
19809 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
19810 * config/i386/i386.md: Ditto.
19811 * config/i386/sse.md: Ditto.
19812 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
19813 with OPTION_MASK_ISA_SSE4_1.
19815 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19818 * doc/extend.texi (x86 Built-in Functions): Correct the name of
19819 __builtin_ia32_lzcnt_u16.
19821 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
19824 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
19825 (rs6000_set_current_function): Rewrite function to use it.
19827 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
19830 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
19831 and TYPE_WARN_IF_NOT_ALIGN.
19832 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
19833 (handle_warn_if_not_align): New.
19834 (place_union_field): Call handle_warn_if_not_align.
19835 (place_field): Call handle_warn_if_not_align.
19836 Copy TYPE_WARN_IF_NOT_ALIGN.
19837 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
19838 (layout_type): Likewise.
19839 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
19841 (tree_decl_common): Add warn_if_not_align.
19842 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
19843 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
19844 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
19845 (DECL_WARN_IF_NOT_ALIGN): Likewise.
19846 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
19847 * doc/extend.texi: Document warn_if_not_aligned attribute.
19848 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
19850 2017-08-17 Martin Liska <mliska@suse.cz>
19853 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
19854 (get_data_dependence): Use it as pointer type.
19855 (distribute_loop): Likewise.
19857 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19859 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
19860 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
19861 (p8_vmrgow_<mode>_direct): New define_insn.
19862 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
19863 handle endianness for vmrgew and vmrgow permute patterns.
19865 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19867 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
19868 * config/rs6000/rs6000-cpus.def: Remove comment.
19869 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
19870 (POWERPC_MASKS): Likewise.
19871 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
19872 use of TARGET_VSX_TIMODE.
19873 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
19874 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
19875 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
19876 (rs6000_option_override_internal): Remove dead code.
19877 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
19878 (rs6000_legitimize_reload_address): Likewise.
19879 (rs6000_legitimate_address_p): Likewise.
19880 (rs6000_opt_masks): Delete "vsx-timode".
19881 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
19882 from function comment.
19883 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
19884 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
19885 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
19887 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
19888 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
19889 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
19890 (VSX_TI): Likewise.
19892 (define_peephole2): Likewise.
19894 2017-08-17 Martin Sebor <msebor@redhat.com>
19897 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
19898 past the end of an array.
19899 (test_pp_format): Add test cases.
19901 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
19903 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
19904 missing ECF_NOTHROW flags.
19906 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19909 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
19910 operands residing in integer registers.
19911 (*vsx_le_perm_load_<mode>): Likewise.
19912 (*vsx_le_perm_store_<mode>): Likewise.
19913 (define_peephole2): Add peepholes to optimize the above.
19915 2017-08-17 Marek Polacek <polacek@redhat.com>
19917 PR middle-end/81814
19918 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
19919 to mimic what shorten_compare did. Change the return type to bool.
19920 (fold_cond_expr_with_comparison): Update call to
19921 operand_equal_for_comparison_p.
19922 (fold_ternary_loc): Likewise.
19924 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
19926 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
19928 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
19930 2017-08-17 Richard Biener <rguenther@suse.de>
19932 * tree-ssa-structalias.c (solve_graph): When propagating
19933 to successors update the graphs succ edges and avoid duplicate work.
19935 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
19938 * config/i386/i386.c (ix86_option_override_internal): Save target
19939 specific options after ix86_stack_protector_guard_reg was changed.
19941 2017-08-17 Richard Biener <rguenther@suse.de>
19943 PR tree-optimization/81827
19944 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
19945 (new_var_info): Initialize it conservatively.
19946 (get_call_vi): Mark register vars.
19947 (new_scalar_tmp_constraint_exp): Likewise.
19948 (handle_rhs_call): Likewise.
19949 (handle_const_call): Likewise.
19950 (create_function_info_for): Likewise.
19951 (solve_constraints): Sort varinfos to separate register from
19952 non-register vars to pack points-to solution bitmaps during
19955 2017-08-17 Marek Polacek <polacek@redhat.com>
19957 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
19959 2017-08-17 Richard Biener <rguenther@suse.de>
19961 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
19962 to true when overflow is undefined and we saturated the result.
19964 2017-08-17 Alan Modra <amodra@gmail.com>
19967 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
19968 Don't use store multiple if only one reg needs saving.
19969 (interesting_frame_related_regno): New function.
19970 (rs6000_frame_related): Don't emit frame info for regs that
19972 (rs6000_emit_epilogue): Likewise.
19974 2017-08-16 Nathan Sidwell <nathan@acm.org>
19976 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
19977 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
19978 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
19979 * tree.c (free_lang_data_in_type): Use else-if chain. Always
19980 clear TYPE_LANG_1. Remove obsolete member-function stripping.
19981 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
19982 (verify_type): Adjust for TYPE_BINFO move.
19983 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
19984 process TYPE_BINFO directly.
19985 (hash_tree): Likewise.
19986 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19988 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19991 2017-08-16 David Malcolm <dmalcolm@redhat.com>
19993 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
19995 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
19998 * config/i386/i386.md (*anddi_1_btr): Change predicates of
19999 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20000 Add ix86_binary_operator_ok to insn constraint.
20001 (*iordi_1_bts): Ditto.
20002 (*xordi_1_btc): Ditto.
20003 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20004 Update corresponding peephole2 pattern.
20008 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20010 PR tree-optimization/81832
20011 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20012 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20014 2017-08-16 Marek Polacek <polacek@redhat.com>
20017 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20018 regarding size_zero_node.
20020 2017-08-16 Martin Liska <mliska@suse.cz>
20023 * config.gcc: Respect previously set extra_objs in case
20026 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20028 PR tree-optimization/81835
20029 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20030 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20031 not depend on the phi.
20033 2017-08-16 Alan Modra <amodra@gmail.com>
20035 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20038 2017-08-16 Alan Modra <amodra@gmail.com>
20040 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20041 (save_reg_p): ..into this. Update all callers.
20042 (first_reg_to_save): Simplify.
20044 2017-08-16 Alan Modra <amodra@gmail.com>
20046 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20049 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20053 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20054 constant count if that count is less than 32.
20056 2017-08-15 Nathan Sidwell <nathan@acm.org>
20058 * gcc.c (execute): Emit friendlier message if inferior is killed
20059 by an external cause.
20061 2017-08-15 Richard Biener <rguenther@suse.de>
20063 PR tree-optimization/81790
20064 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20065 CONSTRUCTORs from simplifying and VN.
20067 2017-08-14 Martin Sebor <msebor@redhat.com>
20069 * builtin-attrs.def: Add comments.
20071 2017-08-14 Martin Sebor <msebor@redhat.com>
20074 * doc/extend.texi (attribute nonstring): Document new attribute.
20076 2017-08-14 Martin Sebor <msebor@redhat.com>
20079 * tree-diagnostic.c (default_tree_printer): Handle %G.
20080 * gimple-pretty-print.h (percent_G_format): Declare new function.
20081 * gimple-pretty-print.c (percent_G_format): Define.
20082 * tree-pretty-print.c (percent_K_format): Add argument.
20084 2017-08-14 Martin Sebor <msebor@redhat.com>
20086 PR translation/79998
20087 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20088 Remove a stray space.
20090 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20093 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20094 (*iordi_1_bts): Ditto.
20095 (*xordi_1_btc): Ditto.
20097 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20100 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20101 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20103 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20104 quoted strings, and make more translator-friendly.
20105 (darwin_rs6000_override_options): Likewise.
20106 (rs6000_option_override_internal): Likewise.
20107 (rs6000_return_in_memory): Fix overlong line.
20108 (init_cmulative_args): Use quoted strings, and make more
20109 translator-friendly.
20110 (rs6000_pass_by_reference): Fix overlong line.
20111 (def_builtin): Use quoted strings.
20112 (altivec_expand_predicate_builtin): Use quoted strings, and make
20113 more translator-friendly.
20114 (htm_expand_builtin): Use quoted strings.
20115 (cpu_expand_builtin): Use quoted strings, and make more
20116 translator-friendly.
20117 (altivec_expand_builtin): Likewise.
20118 (paired_expand_predicate_builtin): Likewise.
20119 (rs6000_invalid_builtin): Likewise.
20120 (builtin_function_type): Use quoted strings.
20121 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20122 more translator-friendly.
20123 (rs6000_trampoline_init): Likewise.
20124 (rs6000_handle_altivec_attribute): Likewise.
20125 (rs6000_inner_target_options): Use quoted strings.
20126 (rs6000_disable_incompatible_switches): Likewise.
20127 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20128 strings, and make more translator-friendly.
20129 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20131 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20133 PR tree-optimization/81799
20134 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20135 cond_expr to simple gimple operand.
20137 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20139 PR middle-end/46932
20140 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20142 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20146 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20147 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20148 TARGET_GASISR_PROLOGUES.
20149 * config/avr/avr.c (avr_option_override): Same.
20150 (avr_pass_pre_proep::execute): Same.
20152 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20155 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20156 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20158 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20160 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20161 define_insn_and_split. Split to a memory load from 0 in
20162 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20163 using PTR mode iterator.
20164 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20165 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20166 (*add_tp_<mode>): Redefine as define_insn_and_split.
20167 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20168 address space. Merge with *add_tp_x32 using PTR mode iterator.
20169 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20170 Split to an add with a memory load from 0 in
20171 DEFAULT_TLS_SEG_REG address space.
20173 2017-08-12 Andrew Pinski <apinski@cavium.com>
20175 * config/aarch64/aarch64-option-extensions.def (rdma):
20176 Fix feature string to what Linux prints out in /proc/cpuinfo.
20178 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20181 * dwarf2out.c (modified_type_die): For C typedef types that have
20182 an ultimate origin, process the ultimate origin instead of the
20184 (gen_typedef_die): Assert that input DECLs have no ultimate
20186 (gen_type_die_with_usage): For typedef variants that have an
20187 ultimate origin, just call gen_decl_die on the original DECL.
20188 (process_scope_var): Avoid creating DIEs for local typedefs and
20189 concrete static variables.
20191 2017-08-12 Alan Modra <amodra@gmail.com>
20195 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20196 match gnu-user.h startfile.
20197 (ENDFILE_LINUX_SPEC): Similarly.
20199 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20202 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20204 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20206 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20207 * config/aarch64/aarch64.md (mov<mode>): Change.
20208 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20209 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20210 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20212 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20214 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20215 for storage order barriers.
20217 2017-08-11 Martin Liska <mliska@suse.cz>
20220 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20221 variables of void type.
20223 2017-08-11 Martin Liska <mliska@suse.cz>
20225 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20226 TARGET_SUPPORTS_ALIASES.
20227 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20228 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20229 (optimize_weakref): Likewise.
20230 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20231 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20232 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20234 2017-08-11 Martin Liska <mliska@suse.cz>
20237 * config/i386/i386.c (make_resolver_func): Do complete
20238 refactoring of the function.
20240 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20243 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20244 * config/i386/i386.c (ix86_stack_protect_guard): Use
20245 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20246 * doc/invoke.texi (x86 Options): Document
20247 -mstack-protector-guard-symbol= option.
20249 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20251 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20252 * config/i386/i386.c (ix86_split_stack_guard): New function.
20253 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20254 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20255 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20256 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20257 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20258 (split_stack_space_check): Call ix86_split_stack_guard.
20260 2017-08-10 Martin Sebor <msebor@redhat.com>
20262 * print-tree.c (print_node): Print location using the established
20264 Replace spaces with colons.
20265 (debug_raw, debug): Ditto.
20267 2017-08-10 Martin Sebor <msebor@redhat.com>
20270 * pretty-print.c (pp_format): Correct the handling of %s precision.
20272 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20275 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20277 (ix86_finalize_stack_frame_flags): This. Also clear
20278 frame_pointer_needed if -fno-omit-frame-pointer is used without
20280 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20281 with ix86_finalize_stack_frame_flags.
20282 (ix86_expand_epilogue): Likewise.
20283 (ix86_expand_split_stack_prologue): Likewise.
20284 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20286 2017-08-10 Martin Liska <mliska@suse.cz>
20289 * c-attribs.c (handle_target_attribute):
20290 Report warning for an empty string argument of target attribute.
20292 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20295 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20297 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20298 or DECL_NONLOCAL labels.
20299 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20300 or DECL_NONLOCAL labels here.
20302 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20304 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20305 to indicate when early gimple folding has been disabled.
20306 (rs6000_gimple_fold_builtin): Add debug content.
20307 (rs6000_invalid_builtin): Fix whitespace.
20308 (rs6000_expand_builtin): Fix whitespace.
20309 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20311 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20314 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20315 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20317 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20319 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20320 (qdf24xx): Likewise.
20321 * config/aarch64/aarch64.md: Include falkor.md.
20322 * config/aarch64/falkor.md: New.
20324 2017-08-09 Marek Polacek <polacek@redhat.com>
20327 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20328 * diagnostic.c (emit_diagnostic): Add a comment.
20329 (emit_diagnostic_valist): New function.
20331 2017-08-09 Marek Polacek <polacek@redhat.com>
20334 * input.c (make_location): New overload.
20335 * input.h (make_location): Declare.
20337 2017-08-08 Alan Modra <amodra@gmail.com>
20338 H.J. Lu <hongjiu.lu@intel.com>
20341 * gcc.c (NO_PIE_SPEC): Delete.
20342 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20344 (LINK_PIE_SPEC): ..to here.
20345 (LINK_COMMAND_SPEC): Support -no-pie.
20346 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20347 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20349 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20350 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20351 (ENDFILE_CRTEND_SPEC): Similarly.
20353 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20356 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20357 (mstack-protector-guard-offset=): Ditto.
20358 * config/i386/i386.c (ix86_option_override): Handle
20359 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20361 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20362 ix86_stack_protect_guard_offset variables.
20363 (TARGET_STACK_PROTECT_GUARD): Always define.
20364 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20365 and -mstack-protector-guard-offset= options.
20367 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20369 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20370 boundary case for the last candidate.
20372 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20374 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20375 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20377 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20379 PR middle-end/19706
20380 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20381 * config/aarch64/aarch64-builtins.c
20382 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20383 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20384 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20386 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20387 Andrew Pinski <pinskia@gmail.com>
20389 PR middle-end/19706
20390 * internal-fn.def (XORSIGN): New.
20391 * optabs.def (xorsign_optab): New.
20392 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20393 (convert_expand_mult_copysign): New.
20394 (pass_optimize_widening_mul::execute): Call
20395 convert_expand_mult_copysign.
20397 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20399 PR tree-optimization/81354
20400 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20401 Insert on edges rather than explicitly creating landing pads.
20402 (analyze_candidates_and_replace): Commit edge inserts.
20404 2017-08-08 Richard Biener <rguenther@suse.de>
20406 PR middle-end/81719
20407 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20408 (expand_simple_operations): Also look through ADDR_EXPRs with
20409 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20411 2017-08-08 Richard Biener <rguenther@suse.de>
20413 PR tree-optimization/81723
20414 * tree-vect-slp.c (struct bst_traits): New hash traits.
20415 (bst_fail): New global.
20416 (vect_build_slp_tree_2): New worker, split out from ...
20417 (vect_build_slp_tree): ... this now wrapping it with using
20418 bst_fail set to cache SLP tree build fails. Properly handle
20420 (vect_analyze_slp_instance): Allocate and free bst_fail.
20422 2017-08-08 Martin Liska <mliska@suse.cz>
20425 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20426 LABEL_DECLs that can be from a different function.
20428 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20430 PR tree-optimization/81744
20431 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20432 loop's number of iterations.
20434 2017-08-08 Martin Liska <mliska@suse.cz>
20436 * asan.c: Include header files.
20437 * attribs.c (build_decl_attribute_variant): New function moved
20439 (build_type_attribute_qual_variant): Likewise.
20440 (cmp_attrib_identifiers): Likewise.
20441 (simple_cst_list_equal): Likewise.
20442 (omp_declare_simd_clauses_equal): Likewise.
20443 (attribute_value_equal): Likewise.
20444 (comp_type_attributes): Likewise.
20445 (build_type_attribute_variant): Likewise.
20446 (lookup_ident_attribute): Likewise.
20447 (remove_attribute): Likewise.
20448 (merge_attributes): Likewise.
20449 (merge_type_attributes): Likewise.
20450 (merge_decl_attributes): Likewise.
20451 (merge_dllimport_decl_attributes): Likewise.
20452 (handle_dll_attribute): Likewise.
20453 (attribute_list_equal): Likewise.
20454 (attribute_list_contained): Likewise.
20455 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20456 (lookup_attribute_by_prefix): Likewise.
20457 * bb-reorder.c: Include header files.
20458 * builtins.c: Likewise.
20459 * calls.c: Likewise.
20460 * cfgexpand.c: Likewise.
20461 * cgraph.c: Likewise.
20462 * cgraphunit.c: Likewise.
20463 * convert.c: Likewise.
20464 * dwarf2out.c: Likewise.
20465 * final.c: Likewise.
20466 * fold-const.c: Likewise.
20467 * function.c: Likewise.
20468 * gimple-expr.c: Likewise.
20469 * gimple-fold.c: Likewise.
20470 * gimple-pretty-print.c: Likewise.
20471 * gimple.c: Likewise.
20472 * gimplify.c: Likewise.
20473 * hsa-common.c: Likewise.
20474 * hsa-gen.c: Likewise.
20475 * internal-fn.c: Likewise.
20476 * ipa-chkp.c: Likewise.
20477 * ipa-cp.c: Likewise.
20478 * ipa-devirt.c: Likewise.
20479 * ipa-fnsummary.c: Likewise.
20480 * ipa-inline.c: Likewise.
20481 * ipa-visibility.c: Likewise.
20483 * lto-cgraph.c: Likewise.
20484 * omp-expand.c: Likewise.
20485 * omp-general.c: Likewise.
20486 * omp-low.c: Likewise.
20487 * omp-offload.c: Likewise.
20488 * omp-simd-clone.c: Likewise.
20489 * opts-global.c: Likewise.
20490 * passes.c: Likewise.
20491 * predict.c: Likewise.
20492 * sancov.c: Likewise.
20493 * sanopt.c: Likewise.
20494 * symtab.c: Likewise.
20495 * toplev.c: Likewise.
20496 * trans-mem.c: Likewise.
20497 * tree-chkp.c: Likewise.
20498 * tree-eh.c: Likewise.
20499 * tree-into-ssa.c: Likewise.
20500 * tree-object-size.c: Likewise.
20501 * tree-parloops.c: Likewise.
20502 * tree-profile.c: Likewise.
20503 * tree-ssa-ccp.c: Likewise.
20504 * tree-ssa-live.c: Likewise.
20505 * tree-ssa-loop.c: Likewise.
20506 * tree-ssa-sccvn.c: Likewise.
20507 * tree-ssa-structalias.c: Likewise.
20508 * tree-ssa.c: Likewise.
20509 * tree-streamer-in.c: Likewise.
20510 * tree-vectorizer.c: Likewise.
20511 * tree-vrp.c: Likewise.
20512 * tsan.c: Likewise.
20513 * ubsan.c: Likewise.
20514 * varasm.c: Likewise.
20515 * varpool.c: Likewise.
20516 * tree.c: Remove functions moved to attribs.[ch].
20517 * tree.h: Likewise.
20518 * config/aarch64/aarch64.c: Add attrs.h header file.
20519 * config/alpha/alpha.c: Likewise.
20520 * config/arc/arc.c: Likewise.
20521 * config/arm/arm.c: Likewise.
20522 * config/avr/avr.c: Likewise.
20523 * config/bfin/bfin.c: Likewise.
20524 * config/c6x/c6x.c: Likewise.
20525 * config/cr16/cr16.c: Likewise.
20526 * config/cris/cris.c: Likewise.
20527 * config/darwin.c: Likewise.
20528 * config/epiphany/epiphany.c: Likewise.
20529 * config/fr30/fr30.c: Likewise.
20530 * config/frv/frv.c: Likewise.
20531 * config/ft32/ft32.c: Likewise.
20532 * config/h8300/h8300.c: Likewise.
20533 * config/i386/winnt.c: Likewise.
20534 * config/ia64/ia64.c: Likewise.
20535 * config/iq2000/iq2000.c: Likewise.
20536 * config/lm32/lm32.c: Likewise.
20537 * config/m32c/m32c.c: Likewise.
20538 * config/m32r/m32r.c: Likewise.
20539 * config/m68k/m68k.c: Likewise.
20540 * config/mcore/mcore.c: Likewise.
20541 * config/microblaze/microblaze.c: Likewise.
20542 * config/mips/mips.c: Likewise.
20543 * config/mmix/mmix.c: Likewise.
20544 * config/mn10300/mn10300.c: Likewise.
20545 * config/moxie/moxie.c: Likewise.
20546 * config/msp430/msp430.c: Likewise.
20547 * config/nds32/nds32-isr.c: Likewise.
20548 * config/nds32/nds32.c: Likewise.
20549 * config/nios2/nios2.c: Likewise.
20550 * config/nvptx/nvptx.c: Likewise.
20551 * config/pa/pa.c: Likewise.
20552 * config/pdp11/pdp11.c: Likewise.
20553 * config/powerpcspe/powerpcspe.c: Likewise.
20554 * config/riscv/riscv.c: Likewise.
20555 * config/rl78/rl78.c: Likewise.
20556 * config/rx/rx.c: Likewise.
20557 * config/s390/s390.c: Likewise.
20558 * config/sh/sh.c: Likewise.
20559 * config/sol2.c: Likewise.
20560 * config/sparc/sparc.c: Likewise.
20561 * config/spu/spu.c: Likewise.
20562 * config/stormy16/stormy16.c: Likewise.
20563 * config/tilegx/tilegx.c: Likewise.
20564 * config/tilepro/tilepro.c: Likewise.
20565 * config/v850/v850.c: Likewise.
20566 * config/vax/vax.c: Likewise.
20567 * config/visium/visium.c: Likewise.
20568 * config/xtensa/xtensa.c: Likewise.
20570 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20573 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20574 constraints since the -mupper-regs-* switches have been
20576 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20577 into a vector from a double word element that was extracted from
20578 another vector, and eliminate extra XXPERMDI instructions.
20579 (vsx_concat_<mode>_2): Likewise.
20580 (vsx_concat_<mode>_3): Likewise.
20581 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20582 concat to allow optimizing inserts from previous extracts.
20584 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20586 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20587 memory reference to a SSP offset in TLS address space.
20588 (ix86_print_operand) <case '@'>: Remove.
20589 (ix86_print_operand_punct_valid_p): Remove '@' code.
20590 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20591 UNSPEC_SP_TLS_TEST.
20592 (stack_tls_protect_set_<mode>): Remove.
20593 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20594 (stack_tls_protect_test_<mode>): Remove.
20595 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20597 2017-08-07 Olivier Hainque <hainque@adacore.com>
20600 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20602 2017-08-07 Douglas Rupp <rupp@adacore.com>
20604 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20605 variable was referenced as multidir in command.
20607 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20610 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20613 2017-08-07 Martin Liska <mliska@suse.cz>
20615 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20616 * config/rl78/rl78.c: Add include of attribs.h.
20617 * config/sh/sh.c: Likewise.
20618 * config/v850/v850.c: Likewise.
20620 2017-08-07 Tom de Vries <tom@codesourcery.com>
20622 PR middle-end/78266
20623 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20625 2017-08-07 Martin Liska <mliska@suse.cz>
20627 * config/mips/mips.c: Include attribs.h.
20629 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20632 * doc/invoke.texi: Document change in behvaior for -Ofast for
20635 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20637 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20639 (aarch64_pop_regs): Likewise.
20640 (aarch64_gen_load_pair): Likewise.
20641 (aarch64_save_callee_saves): Likewise.
20642 (aarch64_restore_callee_saves): Likewise.
20644 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20646 * config/i386/i386.c: Revert the last change.
20648 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20651 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20653 (ix86_finalize_stack_frame_flags): This. Also clear
20654 frame_pointer_needed if -fno-omit-frame-pointer is used without
20656 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20657 with ix86_finalize_stack_frame_flags.
20658 (ix86_expand_epilogue): Likewise.
20659 (ix86_expand_split_stack_prologue): Likewise.
20661 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20664 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20665 to P_AES for Westmere.
20667 2017-08-07 Jonathan Yong <10walls@gmail.com>
20669 * config/i386/mingw.opt (fset-stack-executable): Removed.
20670 * config/i386/cygming.opt (fset-stack-executable): Moved
20672 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20674 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20676 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20678 2017-08-07 Marek Polacek <polacek@redhat.com>
20680 PR middle-end/81737
20681 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20683 2017-08-07 Martin Liska <mliska@suse.cz>
20685 * attribs.h (canonicalize_attr_name): New function.
20686 (cmp_attribs): Move from c-format.c and adjusted.
20687 (is_attribute_p): Moved from tree.h.
20688 * tree-inline.c: Add new includes.
20689 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20690 (private_is_attribute_p): Remove.
20691 (private_lookup_attribute): Likewise.
20692 (private_lookup_attribute_by_prefix): Simplify.
20693 (remove_attribute): Use is_attribute_p.
20694 * tree.h: Remove removed declarations.
20696 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20698 PR middle-end/81698
20699 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20700 instead of computing it in the function. Formatting fix.
20701 (expand_case): Don't rely on default_edge being the first edge,
20702 clear it if removing it, pass default_edge to
20703 emit_case_dispatch_table.
20704 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20707 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20709 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20710 insn in the function, emit NOP after the insn.
20712 2017-08-06 Tom de Vries <tom@codesourcery.com>
20714 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20717 2017-08-06 Tom de Vries <tom@codesourcery.com>
20719 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20722 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20724 PR tree-optimization/57371
20725 * match.pd: New pattern.
20727 2017-08-04 Marek Polacek <polacek@redhat.com>
20729 PR middle-end/81695
20730 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20731 perform the computation in offset_int.
20733 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20735 PR tree-optimization/81136
20736 * tree-vectorizer.h: Include tree-hash-traits.h.
20737 (vec_base_alignments): New typedef.
20738 (vec_info): Add a base_alignments field.
20739 (vect_record_base_alignments): Declare.
20740 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20742 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20743 (create_data_ref): Add an is_conditional_in_stmt argument.
20744 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20745 the is_conditional_in_stmt field.
20746 (data_ref_loc): Add an is_conditional_in_stmt field.
20747 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20748 (find_data_references_in_stmt): Update call to create_data_ref.
20749 (graphite_find_data_references_in_stmt): Likewise.
20750 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20751 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20752 (vect_record_base_alignment): New function.
20753 (vect_record_base_alignments): Likewise.
20754 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20755 for nested statements even if we fail to compute a misalignment.
20756 Use pooled base alignments for unconditional references.
20757 (vect_find_same_alignment_drs): Compare base addresses instead
20759 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20760 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20762 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20764 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20765 Add an explicit name for the enum. Use auto_vec for slp_instances
20766 and grouped_stores.
20767 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20769 (_bb_vec_info): Add a constructor and destructor.
20770 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20771 (destroy_loop_vec_info): Delete.
20772 (vect_destroy_datarefs): Likewise.
20773 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20774 (vec_info::vec_info): New function.
20775 (vec_info::~vec_info): Likewise.
20776 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20777 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20778 destroy_loop_vec_info.
20779 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20780 (_loop_vec_info::_loop_vec_info): ...this.
20781 (destroy_loop_vec_info): Replace with...
20782 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20783 the stmt_vec_infos. Leave handling of vec_info information to its
20784 destructor. Remove explicit vector releases.
20785 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20786 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20787 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20788 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20789 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20790 (destroy_bb_vec_info): Replace with...
20791 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20792 information to its destructor.
20793 (vect_slp_analyze_bb_1): Use new and delete instead of
20794 new_bb_vec_info and destroy_bb_vec_info.
20795 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
20798 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20800 * tree-data-ref.h (subscript): Add access_fn field.
20801 (data_dependence_relation): Add could_be_independent_p.
20802 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
20803 (same_access_functions): Move to tree-data-ref.c.
20804 * tree-data-ref.c (ref_contains_union_access_p): New function.
20805 (access_fn_component_p): Likewise.
20806 (access_fn_components_comparable_p): Likewise.
20807 (dr_analyze_indices): Add a reference to access_fn_component_p.
20808 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
20810 (constant_access_functions): Likewise.
20811 (add_other_self_distances): Likewise.
20812 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
20813 (initialize_data_dependence_relation): Use XCNEW and remove
20814 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
20815 of access functions that have the same type. Allow the
20816 subsequence to end with different bases in some circumstances.
20817 Record the chosen access functions in SUB_ACCESS_FN.
20818 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
20819 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
20820 (subscript_dependence_tester_1): Likewise dra and drb.
20821 (build_classic_dist_vector): Update calls accordingly.
20822 (subscript_dependence_tester): Likewise.
20823 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
20824 DDR_COULD_BE_INDEPENDENT_P.
20825 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
20826 comp_alias_ddrs instead of may_alias_ddrs.
20827 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
20829 (vect_analyze_data_ref_dependence): Use it if
20830 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
20831 distance vectors if that fails.
20832 (dependence_distance_ge_vf): New function.
20833 (vect_prune_runtime_alias_test_list): Use it. Don't clear
20834 LOOP_VINFO_MAY_ALIAS_DDRS.
20836 2017-08-04 Richard Biener <rguenther@suse.de>
20838 PR middle-end/81705
20839 * fold-const.c (fold_binary_loc): Properly restrict
20840 minus_var0 && minus_var1 case when associating undefined overflow
20843 2017-08-04 Tom de Vries <tom@codesourcery.com>
20845 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
20847 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20849 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20850 Don't start diagnostic messages with a capital letter.
20851 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20853 (rs6000_invalid_builtin): Likewise.
20854 (rs6000_trampoline_init): Likewise.
20856 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20859 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
20860 after setting changeable df flags.
20862 2017-08-03 Richard Biener <rguenther@suse.de>
20864 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
20865 up if the use is in USE - X.
20867 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
20869 * toplev.c (dumpfile.h): New include.
20870 (internal_error_reentered): New static function. Use it...
20871 (internal_error_function): ...here to handle reentered internal_error.
20873 2017-08-03 Richard Biener <rguenther@suse.de>
20875 PR middle-end/81148
20876 * fold-const.c (split_tree): Add minus_var and minus_con
20877 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
20878 here but always use minus_*.
20879 (associate_trees): Assert we never associate with MINUS_EXPR
20880 and NULL first operand. Do not recurse for PLUS_EXPR operands
20881 when associating as MINUS_EXPR either.
20882 (fold_binary_loc): Track minus_var and minus_con.
20884 2017-08-03 Tom de Vries <tom@codesourcery.com>
20887 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
20888 ACCEL_COMPILER, apply finish_options on
20889 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
20891 2017-08-03 Tom de Vries <tom@codesourcery.com>
20894 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
20895 function_entry_patch_area_size > 0.
20897 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20900 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
20901 instead of 10??LU, perform unit multiplication in wide_int,
20902 don't change alloc_object_size_limit if the limit is larger
20905 PR tree-optimization/81655
20906 PR tree-optimization/81588
20907 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
20908 the case when ranges[i].low and high are 1 for unsigned type with
20911 PR middle-end/81052
20912 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
20913 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
20915 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20917 * tree-vrp.h: Add include guard.
20919 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
20922 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
20923 (ud2): New insn pattern.
20924 * config/i386/i386.c (ix86_expand_epilogue):
20925 For naked functions, generate ud2 instead of trap insn.
20927 2017-08-02 Marek Polacek <polacek@redhat.com>
20930 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
20932 2017-08-02 Tom de Vries <tom@codesourcery.com>
20933 Cesar Philippidis <cesar@codesourcery.com>
20935 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
20936 Add missing edge probabilities.
20938 2017-08-02 Tamar Christina <tamar.christina@arm.com>
20940 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
20941 Correct endianness.
20943 2017-08-02 Jakub Jelinek <jakub@redhat.com>
20945 PR middle-end/79499
20946 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
20947 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
20948 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
20949 prologue_seq sequences - if any.
20951 2017-08-02 Richard Biener <rguenther@suse.de>
20953 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
20954 via vectors if supported, integer extracts via punning if supported
20955 or otherwise vector extracts.
20957 2017-08-02 Richard Biener <rguenther@suse.de>
20959 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
20961 (bitmap_insert_into_set): ... this.
20963 2017-08-02 Richard Biener <rguenther@suse.de>
20965 PR tree-optimization/81633
20967 2015-08-17 Alan Hayward <alan.hayward@arm.com>
20969 PR tree-optimization/71752
20970 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
20972 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
20974 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
20975 (machine_function::call_ms2sysv_pad_out): Remove field.
20976 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
20977 (ix86_compute_frame_layout): Likewise.
20979 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
20982 * config/i386/i386.c (ix86_set_func_type): Disallow naked
20983 attribute with interrupt attribute.
20985 2017-08-01 Andrew Pinski <apinski@cavium.com>
20987 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
20988 BIT_INSERT_EXPR's operand 1
20989 to see if the types precision matches.
20991 2017-08-01 Martin Liska <mliska@suse.cz>
20993 PR middle-end/70140
20994 * builtins.c (expand_builtin_memcpy_args): Remove.
20995 (expand_builtin_memcpy): Call newly added function
20996 expand_builtin_memory_copy_args.
20997 (expand_builtin_memcpy_with_bounds): Likewise.
20998 (expand_builtin_mempcpy): Remove last argument.
20999 (expand_builtin_mempcpy_with_bounds): Likewise.
21000 (expand_builtin_memory_copy_args): New function created from
21001 expand_builtin_mempcpy_args with small modifications.
21002 (expand_builtin_mempcpy_args): Remove.
21003 (expand_builtin_stpcpy): Remove unused argument.
21004 (expand_builtin): Likewise.
21005 (expand_builtin_with_bounds): Likewise.
21007 2017-08-01 Martin Liska <mliska@suse.cz>
21010 Make mempcpy more optimal (PR middle-end/70140).
21012 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21015 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21016 __builtin_vec_cmpne verify both arguments are compatible vectors
21017 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21018 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21019 move computation of aligned to after checking the argument types.
21023 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21024 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21027 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21028 Doug Rupp <rupp@adacore.com>
21029 Olivier Hainque <hainque@adacore.com>
21031 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21032 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21034 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21035 for TARGET_OS_CPP_BUILTIN.
21036 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21037 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21039 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21040 passing required abi options to the assembler for EABI configurations.
21041 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21042 of .text.hot and .text.unlikely sections for kernel modules when
21043 using ARM style exceptions.
21044 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21045 options. Add EXTRA_CC1_SPEC.
21046 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21048 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21050 (ARM_TARGET2_DWARF_FORMAT): Define.
21051 * config/arm/t-vxworks: Adjust multilib control to removal of the
21052 Diab command line options.
21054 2017-08-01 Martin Liska <mliska@suse.cz>
21056 PR gcov-profile/81561
21057 * gcov.c (unblock): Make unblocking safe as we need to preserve
21058 index correspondence of blocks and block_lists.
21060 2017-08-01 Richard Biener <rguenther@suse.de>
21062 PR tree-optimization/81181
21063 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21064 (compute_antic): ... end of iteration here.
21066 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21068 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21069 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21070 (ftree-slp-vectorize): Likewise.
21071 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21072 can no longer be set independent of flag_tree_loop_vectorize.
21073 * omp-general.c (emp_max_vf): Likewise.
21074 * opts.c (enable_fdo_optimizations): Remove references to
21075 flag_tree_vectorize, these are now implicit.
21076 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21077 and leave it for the options machinery.
21079 2017-08-01 Martin Liska <mliska@suse.cz>
21081 PR middle-end/70140
21082 * builtins.c (expand_builtin_memcpy_args): Remove.
21083 (expand_builtin_memcpy): Call newly added function
21084 expand_builtin_memory_copy_args.
21085 (expand_builtin_memcpy_with_bounds): Likewise.
21086 (expand_builtin_mempcpy): Remove last argument.
21087 (expand_builtin_mempcpy_with_bounds): Likewise.
21088 (expand_builtin_memory_copy_args): New function created from
21089 expand_builtin_mempcpy_args with small modifications.
21090 (expand_builtin_mempcpy_args): Remove.
21091 (expand_builtin_stpcpy): Remove unused argument.
21092 (expand_builtin): Likewise.
21093 (expand_builtin_with_bounds): Likewise.
21095 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21098 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21099 print "ds:" only for immediates in generic address space.
21101 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21104 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21105 (ix86_function_ok_for_sibcall): Return false for naked functions.
21107 2017-08-01 Richard Biener <rguenther@suse.de>
21109 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21110 (compute_antic): Seed worklist with exit block predecessors.
21111 * cfganal.c (dfs_find_deadend): For a cycle return the source
21112 of the edge closing it.
21114 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21116 * config/aarch64/aarch64.c
21117 (aarch64_can_const_movi_rtx_p): Move 0 check.
21119 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21121 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21122 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21124 * match.pd: Ditto in address comparison pattern.
21126 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21128 PR tree-optimization/81627
21129 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21130 closed ssa form for store-store chain.
21132 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21134 PR tree-optimization/81620
21135 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21136 for store-store chain.
21138 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21140 PR tree-optimization/81588
21141 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21142 ranges[i].in_p, invert comparison code ccode. For >/>=,
21143 swap rhs1 and rhs2 and comparison code unconditionally,
21144 for </<= don't do that. Don't swap rhs1/rhs2 again if
21145 ranges[i].in_p, instead invert comparison code ccode if
21146 opcode or oe->rank is BIT_IOR_EXPR.
21149 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21150 a direct optab to conversion optab.
21151 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21152 with GET_MODE_INNER as last argument instead of optab_handler.
21153 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21154 vector extraction if possible and optab is available.
21155 * expr.c (store_constructor): Use convert_optab_handler instead
21156 of optab_handler. Use vector initialization from smaller
21157 vectors if possible and optab is available.
21158 * tree-vect-stmts.c (vectorizable_load): Likewise.
21159 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21160 now have two modes.
21161 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21162 of vec_init from half-sized vectors with the same element mode.
21163 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21164 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21165 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21166 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21167 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21168 after mode in gen_vec_extract* calls.
21169 (vec_extract<mode>): Renamed to ...
21170 (vec_extract<mode><ssescalarmodelower>): ... this.
21171 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21172 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21173 element mode after mode in gen_vec_init* calls.
21174 (VEC_INIT_HALF_MODE): New mode iterator.
21175 (vec_init<mode>): Renamed to ...
21176 (vec_init<mode><ssescalarmodelower>): ... this.
21177 (vec_init<mode><ssehalfvecmodelower>): New expander.
21178 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21179 (vec_extractv2sfsf): ... this.
21180 (vec_initv2sf): Renamed to ...
21181 (vec_initv2sfsf): ... this.
21182 (vec_extractv2si): Renamed to ...
21183 (vec_extractv2sisi): ... this.
21184 (vec_initv2si): Renamed to ...
21185 (vec_initv2sisi): ... this.
21186 (vec_extractv4hi): Renamed to ...
21187 (vec_extractv4hihi): ... this.
21188 (vec_initv4hi): Renamed to ...
21189 (vec_initv4hihi): ... this.
21190 (vec_extractv8qi): Renamed to ...
21191 (vec_extractv8qiqi): ... this.
21192 (vec_initv8qi): Renamed to ...
21193 (vec_initv8qiqi): ... this.
21194 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21195 (vec_init<mode>): Renamed to ...
21196 (vec_init<mode><VEC_base_l>): ... this.
21197 (vec_extract<mode>): Renamed to ...
21198 (vec_extract<mode><VEC_base_l>): ... this.
21199 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21200 (vec_initv2sfsf): ... this.
21201 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21202 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21203 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21204 element mode after mode in gen_vec_init* calls.
21205 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21206 (vec_init<mode><Vel>): ... this.
21207 (vec_extract<mode>): Renamed to ...
21208 (vec_extract<mode><Vel>): ... this.
21209 * config/aarch64/iterators.md (Vel): New mode attribute.
21210 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21211 Add element mode after mode in gen_vec_extract* calls.
21212 * config/s390/vector.md (non_vec_l): New mode attribute.
21213 (vec_extract<mode>): Renamed to ...
21214 (vec_extract<mode><non_vec_l>): ... this.
21215 (vec_init<mode>): Renamed to ...
21216 (vec_init<mode><non_vec_l>): ... this.
21217 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21218 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21220 * config/arm/iterators.md (V_elem_l): New mode attribute.
21221 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21222 (vec_extract<mode><V_elem_l>): ... this.
21223 (vec_extractv2di): Renamed to ...
21224 (vec_extractv2didi): ... this.
21225 (vec_init<mode>): Renamed to ...
21226 (vec_init<mode><V_elem_l>): ... this.
21227 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21228 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21229 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21230 Add element mode after gen_vec_extract* calls.
21231 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21232 (vec_init<mode><unitmode>): ... this.
21233 (vec_extract<mode>): Renamed to ...
21234 (vec_extract<mode><unitmode>): ... this.
21235 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21236 (vec_init<mode><unitmode>): ... this.
21237 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21238 (vec_initv2sfsf): ... this.
21239 (vec_extractv2sf): Renamed to ...
21240 (vec_extractv2sfsf): ... this.
21241 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21242 Add element mode after gen_vec_extract* calls.
21243 * config/mips/mips.md (unitmode): New mode iterator.
21244 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21245 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21246 * config/spu/spu.md (inner_l): New mode attribute.
21247 (vec_init<mode>): Renamed to ...
21248 (vec_init<mode><inner_l>): ... this.
21249 (vec_extract<mode>): Renamed to ...
21250 (vec_extract<mode><inner_l>): ... this.
21251 * config/sparc/sparc.md (veltmode): New mode iterator.
21252 (vec_init<VMALL:mode>): Renamed to ...
21253 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21254 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21255 (vec_initv2sisi): ... this.
21256 (vec_initv2sf): Renamed to ...
21257 (vec_initv2sfsf): ... this.
21258 (vec_extractv2sf): Renamed to ...
21259 (vec_extractv2sfsf): ... this.
21260 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21261 (vec_init<mode>): Renamed to ...
21262 (vec_init<mode><VEC_base_l>): ... this.
21263 (vec_extract<mode>): Renamed to ...
21264 (vec_extract<mode><VEC_base_l>): ... this.
21265 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21266 (vec_initv2sfsf): ... this.
21267 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21268 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21269 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21270 gen_vec_init* calls.
21272 2017-08-01 Richard Biener <rguenther@suse.de>
21274 PR tree-optimization/81297
21275 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21276 TREE_OVERFLOW from INTEGER_CSTs.
21278 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21280 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21282 2017-07-31 Carl Love <cel@us.ibm.com>
21284 * config/rs6000/rs6000-c: Add support for built-in functions
21285 vector signed char vec_xl_be (signed long long, signed char *);
21286 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21287 vector signed int vec_xl_be (signed long long, signed int *);
21288 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21289 vector signed long long vec_xl_be (signed long long, signed long long *);
21290 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21291 vector signed short vec_xl_be (signed long long, signed short *);
21292 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21293 vector double vec_xl_be (signed long long, double *);
21294 vector float vec_xl_be (signed long long, float *);
21295 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21296 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21297 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21299 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21300 (altivec_expand_builtin): Add switch statement to call
21301 altivec_expand_xl_be for each builtin.
21302 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21303 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21304 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21305 __builtin_vsx_le_be_v16qi.
21306 * doc/extend.texi: Update the built-in documentation file for the
21307 new built-in functions.
21309 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21312 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21314 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21316 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21318 * config.gcc: Add z14.
21319 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21320 CPU model numbers for z13s and z14.
21321 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21323 * config/s390/s390-opts.h (enum processor_type): Rename
21324 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21325 * config/s390/s390.c (processor_table): Add field for CPU name to
21326 be passed to Binutils.
21327 (s390_asm_output_machine_for_arch): Use the new field in
21328 processor_table for Binutils.
21329 (s390_expand_builtin): Replace arch12 with z14.
21330 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21331 (s390_get_sched_attrmask): Likewise.
21332 (s390_get_unit_mask): Likewise.
21333 * config/s390/s390.opt: Add z14 to processor_type enum.
21335 2017-07-31 Martin Jambor <mjambor@suse.cz>
21338 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21339 regardless of optimization level.
21341 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21342 Martin Liska <mliska@suse.cz>
21344 * predict.def: Remove old comment and adjust probability.
21345 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21346 PREDICT statements.
21348 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21351 * config/i386/i386.c (ix86_function_naked): New function.
21352 (ix86_can_use_return_insn_p): Return false for naked functions.
21353 (ix86_expand_prologue): Skip prologue for naked functions.
21354 (ix86_expand_epilogue): Skip epilogue for naked functions
21355 and emit trap instruction.
21356 (ix86_warn_func_return): New function.
21357 (ix86_attribute_table): Add "naked" attribute specification.
21358 (TARGET_WARN_FUNC_RETURN): Define.
21359 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21361 2017-07-31 Martin Liska <mliska@suse.cz>
21363 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21364 (dump_gimple_bb_header): Always dump BB info.
21365 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21367 2017-07-31 Martin Liska <mliska@suse.cz>
21370 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21371 also with current_function_decl non-null equality.
21373 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21376 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21377 change type to the element type, instead add eltype variable and
21378 use it where we are interested in the element type.
21380 PR tree-optimization/81603
21381 * ipa-polymorphic-call.c
21382 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21383 offset arithmetic in offset_int, bail out if the resulting bit offset
21384 doesn't fit into shwi.
21386 2017-07-31 Martin Liska <mliska@suse.cz>
21388 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21389 (gimplify_save_expr): Fix comment.
21391 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21394 * config/i386/i386.c (ix86_function_arg): Update arguments for
21396 (ix86_compute_frame_layout): Set the initial stack offset to
21397 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21398 INCOMING_FRAME_SP_OFFSET.
21399 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21400 stack before exception handler returns.
21401 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21402 the 'ERROR_CODE' for exception handler.
21404 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21406 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21407 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21408 (ASM_OUTPUT_REG_POP): Ditto.
21409 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21410 instead of asm_fprintf to output pure string.
21412 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21414 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21415 to imported_module_or_decl hook.
21416 (debug_nothing_tree_tree_tree_bool): Remove.
21417 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21418 * debug.c (do_nothing_debug_hooks): Use
21419 debug_nothing_tree_tree_tree_bool_bool instead of
21420 debug_nothing_tree_tree_tree_bool.
21421 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21422 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21423 * sdbout.c (sdb_debug_hooks): Likewise.
21424 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21425 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21427 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21428 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21429 attribute, don't add anything.
21431 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21433 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21434 (fold_build2_stat_loc): Likewise.
21435 (fold_build3_stat_loc): Likewise.
21436 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21437 (fold_build1_loc): Remove macro.
21438 (fold_build2_loc): Likewise.
21439 (fold_build3_loc): Likewise.
21441 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21443 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21444 (gimple_build_debug_bind_source_stat): Likewise.
21445 * gimple.h (gimple_build_debug_bind): Remove macro.
21446 (gimple_build_debug_bind_source): Likewise.
21448 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21450 * bitmap.c (bitmap_alloc): Adjust.
21451 (bitmap_gc_alloc): Likewise.
21452 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21454 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21456 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21457 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21458 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21459 (bitmap_gc_alloc_stat): Likewise.
21460 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21462 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21464 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21465 * rtl.h (shallow_copy_rtx): Remove macro.
21467 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21469 * emit-rtl.c (gen_raw_REG): Adjust.
21470 * gengenrtl.c (gendef): Likewise.
21471 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21472 * rtl.h (rtx_alloc): Remove macro.
21474 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21476 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21477 (build_tree_list_stat): Likewise.
21478 * tree.h (build_tree_list): Remove macro.
21479 (build_tree_list_vec): Likewise.
21481 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21483 * tree.c (make_vector_stat): Remove _stat from name.
21484 (build_vector_stat): Likewise.
21485 * tree.h (make_vector_stat): Remove macro.
21486 (build_vector_stat): Likewise.
21488 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21490 * tree.h (build_var_debug_value): Remove prototype.
21492 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21494 * tree.c (tree_cons_stat): Remove _stat from name.
21495 * tree.h (tree_cons): Remove macro.
21497 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21499 * tree.c (build_vl_exp_stat): Remove _stat from name.
21500 * tree.h (build_vl_exp): Remove macro.
21502 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21504 * tree.c (build_decl_stat): Remove _stat from name.
21505 * tree.h (build_decl): Remove macro.
21507 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21509 * gimple.c (gimple_build_with_ops_stat): Adjust.
21510 (gimple_alloc_stat): Remove _stat from name.
21511 * gimple.h (gimple_alloc): Remove macro.
21513 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21515 * tree.c (make_tree_vec_stat): Remove _stat from name.
21516 (grow_tree_vec_stat): Likewise.
21517 * tree.h (make_tree_vec_stat): Adjust prototype.
21518 (grow_tree_vec_stat): Likewise.
21519 (make_tree_vec): Remove macro.
21520 (grow_tree_vec): Likewise.
21522 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21524 * fold-const.c (fold_build1_stat_loc): Adjust.
21525 (fold_build2_stat_loc): Likewise.
21526 (fold_build3_stat_loc): Likewise.
21527 * tree.c (build0_stat): Remove _stat from name.
21528 (build1_stat): Likewise.
21529 (build2_stat): Likewise.
21530 (build3_stat): Likewise.
21531 (build4_stat): Likewise.
21532 (build5_stat): Likewise.
21533 * tree.h (build1_loc): Remove macro, and rename _stat function
21535 (build2_loc): Likewise.
21536 (build3_loc): Likewise.
21537 (build4_loc): Likewise.
21538 (build5_loc): Likewise.
21540 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21542 * tree.c (make_int_cst_stat): Remove _stat from name.
21543 * tree.h (make_int_cst_stat): Adjust prototype.
21544 (make_int_cst): Remove macro.
21546 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21548 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21549 * tree.h (make_tree_binfo_stat): Adjust prototype.
21550 (make_tree_binfo): Remove.
21552 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21554 * tree.c (copy_node_stat): Rename to copy_node.
21555 (build_distinct_type_copy): Adjust.
21556 * tree.h (copy_node_stat): Adjust prototype.
21557 (copy_node): Remove macro.
21559 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21561 * tree.c (make_node_stat): rename to make_node.
21562 (build_tree_list_stat): Adjust.
21563 (build0_stat): Likewise.
21564 (build2_stat): Likewise.
21565 (build3_stat): Likewise.
21566 (build4_stat): Likewise.
21567 (build5_stat): Likewise.
21568 (build_decl_stat): Likewise.
21569 * tree.h (make_node_stat): Adjust prototype.
21570 (make_node): remove macro.
21572 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21574 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21575 (PPC_FEATURE2_SCV): Likewise.
21576 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21578 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21580 * config/aarch64/aarch64.c
21581 (aarch64_internal_mov_immediate): Add new special pattern.
21582 * config/aarch64/aarch64.md (*movdi_aarch64):
21583 Add reg/32bit const mov case.
21585 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21586 Richard Sandiford <richard.sandiford@linaro.org>
21588 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21589 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21590 Add integer and movi cases.
21591 (movi-split-hf-df-sf split, fp16): New.
21592 (enabled): Added TARGET_FP_F16INST.
21593 * config/aarch64/iterators.md (GPF_HF): New.
21594 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21596 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21598 * config/aarch64/aarch64.c
21599 (aarch64_simd_container_mode): Add prototype.
21600 (aarch64_expand_mov_immediate): Add HI support.
21601 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21602 (aarch64_can_const_movi_rtx_p): New.
21603 (aarch64_preferred_reload_class):
21604 Remove restrictions of using FP registers for certain SIMD operations.
21605 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21606 (aarch64_valid_floating_const): Add integer move validation.
21607 (aarch64_simd_imm_scalar_p): Remove.
21608 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21609 (aarch64_legitimate_constant_p): Expand list of supported cases.
21610 * config/aarch64/aarch64-protos.h
21611 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21612 (aarch64_reinterpret_float_as_int): New.
21613 (aarch64_simd_imm_scalar_p): Remove.
21614 * config/aarch64/constraints.md (Uvi): New.
21615 (Dd): Split into Ds and new Dd.
21616 * config/aarch64/aarch64.md (*movsi_aarch64):
21618 (*movdi_aarch64): Add SIMD mov case.
21620 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21622 * tree-predcom.c: (struct chain): Handle store-store chain in which
21623 stores for elimination only store loop invariant values.
21624 (execute_pred_commoning_chain): Ditto.
21625 (prepare_initializers_chain_store_elim): Ditto.
21626 (prepare_finalizers): Ditto.
21627 (is_inv_store_elimination_chain): New function.
21628 (initialize_root_vars_store_elim_1): New function.
21630 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21632 * tree-predcom.c: Revise general description of the pass.
21633 (enum chain_type): New enum type for store elimination.
21634 (struct chain): New field supporting store elimination.
21635 (struct component): Ditto.
21636 (dump_chain): Dump store-stores chain.
21637 (release_chain): Release resources.
21638 (split_data_refs_to_components): Compute and create component
21639 contains only stores for elimination.
21640 (get_chain_last_ref_at): New function.
21641 (make_invariant_chain): Initialization.
21642 (make_rooted_chain): Specify chain type in parameter and record it.
21643 (add_looparound_copies): Skip for store-stores chain.
21644 (determine_roots_comp): Compute type of chain and pass it to
21646 (initialize_root_vars_store_elim_2): New function.
21647 (finalize_eliminated_stores): New function.
21648 (remove_stmt): Handle store for elimination.
21649 (execute_pred_commoning_chain): Execute predictive commoning on
21650 store-store chains.
21651 (determine_unroll_factor): Skip unroll for store-stores chain.
21652 (prepare_initializers_chain_store_elim): New function.
21653 (prepare_initializers_chain): Hanlde store-store chain.
21654 (prepare_finalizers_chain, prepare_finalizers): New function.
21655 (tree_predictive_commoning_loop): Return integer value indicating
21656 if loop is unrolled or lcssa form is corrupted.
21657 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21659 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21661 * tree-predcom.c (initialize_root): Delete.
21662 (execute_pred_commoning_chain): Initialize root vars and replace
21663 reference of non-combined chain directly, rather than call above
21666 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21668 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21669 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21671 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21673 * tree-predcom.c (struct chain): New field init_seq.
21674 (release_chain): Release init_seq.
21675 (prepare_initializers_chain): Record intialization stmts in above
21677 (insert_init_seqs): New function.
21678 (tree_predictive_commoning_loop): Call insert_init_seqs.
21680 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21682 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21684 2017-07-28 Richard Biener <rguenther@suse.de>
21686 * match.pd: Remove superfluous :c.
21687 * genmatch.c (simplify::id): Add member.
21688 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21690 (current_id): New global.
21691 (dt_node::parent): Move from ...
21692 (dt_operand::parent): ... here. Add for_id member.
21693 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21694 (decision_tree::find_node): Relax order requirement when
21695 merging DT_TRUE nodes to ones inbetween the current simplify
21696 and the one we try to merge with. Add diagnostic whenever
21697 we need to enforce pattern order by not merging.
21698 (decision_tree::insert): Set current_id.
21699 (decision_tree::print_node): Dump parent node and for_id.
21700 (parser::last_id): Add member.
21701 (parser::push_simplify): Assign unique id.
21702 (parser::parser): Initialize last_id.
21704 2017-07-28 Martin Liska <mliska@suse.cz>
21707 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21708 gimple_build_debug_bind.
21710 2017-07-28 Richard Biener <rguenther@suse.de>
21712 PR tree-optimization/81502
21713 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21715 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21717 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21718 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21719 for BIT_FIELD_REF args.
21720 * fold-const.c (make_bit_field_ref): Likewise.
21721 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21723 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21726 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21727 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21728 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21729 Or it into SANITIZER_UNDEFINED.
21730 * ubsan.c: Include gimple-fold.h and varasm.h.
21731 (ubsan_expand_ptr_ifn): New function.
21732 (instrument_pointer_overflow): New function.
21733 (maybe_instrument_pointer_overflow): New function.
21734 (instrument_object_size): Formatting fix.
21735 (pass_ubsan::execute): Call instrument_pointer_overflow
21736 and maybe_instrument_pointer_overflow.
21737 * internal-fn.c (expand_UBSAN_PTR): New function.
21738 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21739 * sanitizer.def (__ubsan_handle_pointer_overflow,
21740 __ubsan_handle_pointer_overflow_abort): New builtins.
21741 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21742 * internal-fn.def (UBSAN_PTR): New internal function.
21743 * opts.c (sanitizer_opts): Add pointer-overflow.
21744 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21745 * fold-const.c (build_range_check): Compute pointer range check in
21746 integral type if pointer arithmetics would be needed. Formatting
21749 2017-07-28 Martin Liska <mliska@suse.cz>
21752 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21753 parameters that are of a variable-length.
21755 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21757 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21759 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21760 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21761 (CRT_CALL_STATIC_FUNCTION): Likewise.
21762 (ASM_DEFAULT_SPEC): New define.
21763 (ASM_SPEC32): Likewise.
21764 (ASM_SPEC64): Likewise.
21765 (ASM_SPEC_COMMON): Likewise.
21766 (ASM_SPEC): Likewise.
21767 (INVALID_64BIT): Likewise.
21768 (LINK_OS_DEFAULT_SPEC): Likewise.
21769 (LINK_OS_SPEC32): Likewise.
21770 (LINK_OS_SPEC64): Likewise.
21771 (POWERPC_LINUX): Likewise.
21772 (PTRDIFF_TYPE): Likewise.
21773 (RESTORE_FP_PREFIX): Likewise.
21774 (RESTORE_FP_SUFFIX): Likewise.
21775 (SAVE_FP_PREFIX): Likewise.
21776 (SAVE_FP_SUFFIX): Likewise.
21777 (SIZE_TYPE): Likewise.
21778 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21779 (TARGET_64BIT): Likewise.
21780 (TARGET_64BIT): Likewise.
21781 (TARGET_AIX): Likewise.
21782 (WCHAR_TYPE_SIZE): Likewise.
21783 (WCHAR_TYPE): Undefine.
21784 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21785 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21786 (CPP_OS_RTEMS_SPEC): Delete.
21787 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21788 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21790 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21792 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21794 PR tree-optimization/81578
21795 * tree-parloops.c (build_new_reduction): Bail out if
21796 reduction_code isn't one of the standard OpenMP reductions.
21797 Move the details printing after that decision.
21799 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21801 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
21802 related to reload_in_progress.
21803 (splat_input_operand): Likewise.
21804 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
21806 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
21808 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
21809 (TARGET_INSTANTIATE_DECLS): Likewise.
21810 (legitimate_indexed_address_p): Delete reload_in_progress code.
21811 (rs6000_debug_legitimate_address_p): Likewise.
21812 (rs6000_eliminate_indexed_memrefs): Likewise.
21813 (rs6000_emit_le_vsx_store): Likewise.
21814 (rs6000_emit_move_si_sf_subreg): Likewise.
21815 (rs6000_emit_move): Likewise.
21816 (register_to_reg_type): Likewise.
21817 (rs6000_pre_atomic_barrier): Likewise.
21818 (rs6000_machopic_legitimize_pic_address): Likewise.
21819 (rs6000_allocate_stack_temp): Likewise.
21820 (rs6000_address_for_fpconvert): Likewise.
21821 (rs6000_address_for_altivec): Likewise.
21822 (rs6000_secondary_memory_needed_rtx): Delete function.
21823 (rs6000_check_sdmode): Likewise.
21824 (rs6000_alloc_sdmode_stack_slot): Likewise.
21825 (rs6000_instantiate_decls): Likewise.
21826 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
21827 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
21828 Delete reload_in_progress.
21829 (*vec_reload_and_plus_<mptrsize>): Likewise.
21830 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
21831 (vsx_div_v2di): Likewise.
21832 (vsx_udiv_v2di): Likewise.
21834 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21836 * config/rs6000/rs6000.opt (mlra): Replace with stub.
21837 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
21838 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
21839 (rs6000_debug_reg_global): Delete print of LRA status.
21840 (rs6000_option_override_internal): Delete dead LRA related code.
21841 (rs6000_lra_p): Delete function.
21842 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
21844 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
21846 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
21847 * config/riscv/rtems.h: New file.
21849 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21850 Sudakshina Das <sudi.das@arm.com>
21852 * config/aarch64/aarch64.md
21853 (define_split for and<mode>3nr_compare): Move
21854 non aarch64_logical_operand to a register.
21855 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
21856 register immediate operand to a register.
21857 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
21859 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21861 PR middle-end/81564
21862 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
21864 2017-07-27 Richard Biener <rguenther@suse.de>
21866 PR tree-optimization/81573
21867 PR tree-optimization/81494
21868 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
21869 multi defuse cycle case.
21871 2017-07-27 Richard Biener <rguenther@suse.de>
21873 PR tree-optimization/81571
21874 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
21877 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
21879 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
21880 earlier and only if MASK_FPU is set. Adjust formatting.
21882 2017-07-27 Martin Liska <mliska@suse.cz>
21884 * opt-functions.awk: Add validation of value of Init.
21885 * optc-gen.awk: Pass new argument.
21887 2017-07-27 Martin Liska <mliska@suse.cz>
21889 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
21890 Fix wrong condition.
21892 2017-07-27 Martin Liska <mliska@suse.cz>
21894 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
21895 BBs and edges seen by autoFDO.
21897 2017-07-27 Richard Biener <rguenther@suse.de>
21899 PR tree-optimization/81502
21900 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
21901 with incompatible but same sized type.
21902 (execute_update_addresses_taken): Likewise.
21904 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
21906 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
21907 flag_tree_loop_vectorize rather than flag_tree_vectorize.
21909 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21912 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
21913 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
21914 Change s_operand to memory_operand.
21916 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
21918 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
21919 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
21920 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
21921 Emit instructions rather than returning an expression. Handle TFmode
21922 and KFmode by casting to TImode.
21923 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
21924 (rs6000_emit_le_vsx_store): Likewise.
21925 * config/rs6000/vsx.md (VSX_TI): New iterator.
21926 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
21927 (*vsx_le_undo_permute_<mode>): Likewise.
21928 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
21929 emit the split sequence.
21930 (*vsx_le_perm_store_<mode>): Likewise.
21932 2017-07-27 Jakub Jelinek <jakub@redhat.com>
21934 PR tree-optimization/81555
21935 PR tree-optimization/81556
21936 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
21937 if true, force CHANGED for the recursive invocation.
21938 (reassociate_bb): Remember original length of ops array, pass
21939 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
21941 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
21942 attributes for noipa attribute. For naked attribute use
21943 lookup_attribute first before lookup_attribute_spec.
21944 * final.c (rest_of_handle_final): Disable IPA RA for functions with
21946 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
21947 for functions with noipa attribute.
21948 (cgraph_externally_visible_p): Return true for functions with noipa
21950 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
21951 for functions with noipa attribute.
21952 * doc/extend.texi: Document noipa function attribute.
21953 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
21954 also for functions with noipa attribute.
21955 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
21957 2017-07-26 Andrew Pinski <apinski@cavium.com>
21959 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
21960 vec_unalign_load_cost and vec_unalign_store_cost.
21962 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
21964 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
21965 -mvsx-small-integer option.
21966 (ISA_3_0_MASKS_IEEE): Likewise.
21967 (OTHER_VSX_VECTOR_MASKS): Likewise.
21968 (POWERPC_MASKS): Likewise.
21969 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
21970 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
21971 code, only testing for DImode being allowed in non-VSX floating
21973 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
21974 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
21976 (rs6000_option_override_internal): Delete -mvsx-small-integer.
21977 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
21978 TARGET_P8_VECTOR test.
21979 (rs6000_secondary_reload_simple_move): Likewise.
21980 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
21981 since TARGET_P9_VECTOR was already tested.
21982 (rs6000_opt_masks): Remove -mvsx-small-integer.
21983 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
21984 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21986 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
21987 test for TARGET_VEXTRACTUB was used, and that uses
21989 (p9 extract splitter): Likewise.
21990 (vsx_extract_<mode>_di_p9): Likewise.
21991 (vsx_extract_<mode>_store_p9): Likewise.
21992 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
21993 for TARGET_P9_VECTOR was used. Delete code that is now dead with
21994 the elimination of TARGET_VSX_SMALL_INTEGER.
21995 (vsx_extract_<mode>_p8): Likewise.
21996 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
21997 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
21998 (vsx_set_<mode>_p9): Likewise.
21999 (vsx_set_v4sf_p9): Likewise.
22000 (vsx_set_v4sf_p9_zero): Likewise.
22001 (vsx_insert_extract_v4sf_p9): Likewise.
22002 (vsx_insert_extract_v4sf_p9_2): Likewise.
22003 * config/rs6000/rs6000.md (sign extend splitter): Change
22004 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22005 (floatsi<mode>2_lfiwax_mem): Likewise.
22006 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22007 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22008 since a test for TARGET_P9_VECTOR was used.
22009 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22010 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22011 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22012 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22013 TARGET_P8_VECTOR test.
22014 (fix_trunc<mode>si2_stfiwx): Likewise.
22015 (fix_trunc<mode>si2_internal): Likewise.
22016 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22017 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22019 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22020 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22021 TARGET_P8_VECTOR test.
22022 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22023 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22024 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22026 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22027 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22028 since a test for TARGET_P9_VECTOR was used.
22029 (splitter for loading small constants): Likewise.
22031 2017-07-26 Andrew Pinski <apinski@cavium.com>
22033 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22036 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22039 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22040 (fp_valid_at): Likewise.
22042 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22044 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22045 (qdf24xx_addrcost_table): Likewise.
22046 (cortexa57_tunings): Update to use generic_branch_cost.
22047 (cortexa72_tunings): Likewise.
22048 (cortexa73_tunings): Likewise.
22049 (qdf24xx_tunings): Likewise.
22051 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22053 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22054 (thunderx2t99_branch_cost): Likewise.
22055 (cortexa35_tunings): Update to use generic_branch_cost.
22056 (cortexa53_tunings): Likewise.
22057 (cortexa57_tunings): Likewise.
22058 (cortexa72_tunings): Likewise.
22059 (cortexa73_tunings): Likewise.
22060 (thunderx2t99_tunings): Likewise.
22062 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22064 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22065 (sparc_option_override): Honour MASK_FSMULD.
22066 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22067 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22068 * config/sparc/sparc.opt (mfsmuld): New option.
22069 * doc/invoke.texi (mfsmuld): Document option.
22071 2017-07-26 Marek Polacek <polacek@redhat.com>
22073 PR middle-end/70992
22074 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22076 2017-07-26 Richard Biener <rguenther@suse.de>
22078 * gimple-match-head.c (do_valueize): Return OP if valueize
22080 (get_def): New helper to get at the def stmt of a SSA name
22081 if valueize allows.
22082 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22083 do_valueize to get at the def stmt.
22084 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22086 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22088 PR middle-end/46932
22089 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22091 2017-07-26 Martin Liska <mliska@suse.cz>
22094 * function.c (expand_function_start): Make expansion of
22095 nonlocal_goto_save_area after parm_birth_insn.
22097 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22099 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22100 from all CPU target flags enable members.
22102 2017-07-26 Richard Biener <rguenther@suse.de>
22104 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22105 (decision_tree::gen): Make 'type' const.
22106 (write_predicate): Likewise.
22108 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22110 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22111 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22112 (rs6000_option_override_internal): Likewise.
22113 (rs6000_expand_vector_set): Likewise.
22114 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22115 (TARGET_UPPER_REGS_SF): Likewise.
22116 (TARGET_UPPER_REGS_DI): Likewise.
22117 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22118 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22119 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22120 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22121 (Splitters for DI constants in Altivec registers): Likewise.
22122 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22123 (vsx_set_v4sf_p9): Likewise.
22124 (vsx_set_v4sf_p9_zero): Likewise.
22125 (vsx_insert_extract_v4sf_p9): Likewise.
22126 (vsx_insert_extract_v4sf_p9_2): Likewise.
22128 2017-07-25 Carl Love <cel@us.ibm.com>
22130 * doc/extend.texi: Update the built-in documentation file for the
22131 existing built-in functions
22132 vector signed char vec_cnttz (vector signed char);
22133 vector unsigned char vec_cnttz (vector unsigned char);
22134 vector signed short vec_cnttz (vector signed short);
22135 vector unsigned short vec_cnttz (vector unsigned short);
22136 vector signed int vec_cnttz (vector signed int);
22137 vector unsigned int vec_cnttz (vector unsigned int);
22138 vector signed long long vec_cnttz (vector signed long long);
22139 vector unsigned long long vec_cnttz (vector unsigned long long);
22141 2017-07-25 Andrew Pinski <apinski@cavium.com>
22143 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22144 accesses where the use is for the first operand of a BIT_INSERT.
22146 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22149 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22150 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22152 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22154 * config/i386/gstabs.h: Delete.
22155 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22157 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22159 * config/i386/i386.c (ix86_decompose_address): Do not check for
22160 register RTX when looking at index_reg or base_reg.
22161 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22163 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22165 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22166 to update EH info here.
22168 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22170 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22172 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22174 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22176 2017-07-25 Torsten Duwe <duwe@suse.de>
22178 * common.opt: Introduce -fpatchable-function-entry
22179 command line option, and its variables function_entry_patch_area_size
22180 and function_entry_patch_area_start.
22181 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22182 including a two-value parser.
22183 * target.def (print_patchable_function_entry): New target hook.
22184 * targhooks.h (default_print_patchable_function_entry): New function.
22185 * targhooks.c (default_print_patchable_function_entry): Likewise.
22186 * toplev.c (process_options): Switch off IPA-RA if
22187 patchable function entries are being generated.
22188 * varasm.c (assemble_start_function): Look at the
22189 patchable-function-entry command line switch and current
22190 function attributes and maybe generate NOP instructions by
22191 calling the print_patchable_function_entry hook.
22192 * doc/extend.texi: Document patchable_function_entry attribute.
22193 * doc/invoke.texi: Document -fpatchable_function_entry
22194 command line option.
22195 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22197 * doc/tm.texi: Re-generate.
22199 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22202 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22203 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22205 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22207 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22209 2017-07-25 Richard Biener <rguenther@suse.de>
22211 PR tree-optimization/81455
22212 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22213 not walk in cycles when looking for guards.
22215 2017-07-25 Richard Biener <rguenther@suse.de>
22217 PR tree-optimization/81529
22218 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22219 when optimizing backedge uses.
22221 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22223 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22225 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22226 to dl_section_ref. On AIX, append an expression to subtract
22227 the size of the section length to dl_section_ref.
22229 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22231 * configure.ac: If any of the config.* scripts fail, exit 1.
22232 * configure: Regenerate.
22234 2017-07-25 Richard Biener <rguenther@suse.de>
22236 PR middle-end/81546
22237 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22238 of immediate uses, be more verbose on errors.
22240 2017-07-25 Richard Biener <rguenther@suse.de>
22242 PR tree-optimization/81510
22243 * tree-vect-loop.c (vect_is_simple_reduction): When the
22244 reduction stmt is not inside the loop bail out.
22246 2017-07-25 Richard Biener <rguenther@suse.de>
22248 PR tree-optimization/81303
22249 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22250 profitability check against LOOP_VINFO_NITERSM1.
22252 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22254 * domwalk.c (cmp_bb_postorder): Simplify.
22255 (sort_bbs_postorder): New function. Use it...
22256 (dom_walker::walk): ...here to optimize common cases.
22258 2017-07-25 Martin Liska <mliska@suse.cz>
22261 * ipa-visibility.c (function_and_variable_visibility): Make the
22262 redirection just on target that supports aliasing.
22263 Fix GNU coding style.
22265 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22268 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22270 * config/arm/rtems.h: Likewise.
22271 * config/bfin/rtems.h: Likewise.
22272 * config/i386/rtemself.h: Likewise.
22273 * config/lm32/rtems.h: Likewise.
22274 * config/m32c/rtems.h: Likewise.
22275 * config/m68k/rtemself.h: Likewise.
22276 * config/microblaze/rtems.h: Likewise.
22277 * config/mips/rtems.h: Likewise.
22278 * config/moxie/rtems.h: Likewise.
22279 * config/nios2/rtems.h: Likewise.
22280 * config/powerpcspe/rtems.h: Likewise.
22281 * config/rs6000/rtems.h: Likewise.
22282 * config/rtems.h: Likewise.
22283 * config/sh/rtems.h: Likewise.
22284 * config/sh/rtemself.h: Likewise.
22285 * config/sparc/rtemself.h: Likewise.
22287 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22290 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22291 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22292 * tree-ssa-structalias.c (alias_get_name): Same.
22294 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22297 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22298 instructions if no du chain is found.
22300 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22302 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22304 2017-07-25 Richard Biener <rguenther@suse.de>
22306 PR middle-end/81505
22307 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22310 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22312 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22313 upper-regs options.
22314 (ISA_2_7_MASKS_SERVER): Likewise.
22315 (ISA_3_0_MASKS_IEEE): Likewise.
22316 (OTHER_P8_VECTOR_MASKS): Likewise.
22317 (OTHER_VSX_VECTOR_MASKS): Likewise.
22318 (POWERPC_MASKS): Likewise.
22319 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22320 duplicate list of options.
22321 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22322 explicit -mupper-regs options.
22323 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22324 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22325 alias for -mupper-regs-df.
22326 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22327 (rs6000_init_hard_regno_mode_ok): Likewise.
22328 (rs6000_option_override_internal): Likewise.
22329 (rs6000_opt_masks): Likewise.
22330 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22331 options in terms of whether -mvsx or -mpower8-vector was used.
22332 (TARGET_UPPER_REGS_DI): Likewise.
22333 (TARGET_UPPER_REGS_SF): Likewise.
22334 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22335 -mupper-regs-* options.
22337 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22339 * passes.c (emergency_dump_function): Print some empty lines and a
22340 header before the RTL dump.
22342 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22344 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22346 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22349 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22350 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22352 2017-07-24 Carl Love <cel@us.ibm.com>
22354 * config/rs6000/rs6000-c.c: Add support for built-in functions
22355 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22356 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22357 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22358 vec_extract_fp_from_shortl): Add defines for the two builtins.
22359 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22360 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22362 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22363 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22364 * doc/extend.texi: Update the built-in documentation file for the
22365 new built-in function.
22367 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22370 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22372 * doc/generic.texi: Likewise.
22373 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22374 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22376 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22378 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22379 (aarch64_mls_elt_merge<mode>): Likewise.
22381 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22383 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22384 having __cxa_atexit.
22386 2017-07-23 Michael Collison <michael.collison@arm.com>
22388 * config/arm/arm.c (arm_option_override): Deprecate
22389 use of -mstructure-size-boundary.
22390 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22391 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22393 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22396 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22397 Reduce cost estimate for direct moves.
22399 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22402 * config/i386/i386.c (ix86_option_override_internal): Disable
22403 BMI, BMI2 and TBM instructions for -m16.
22405 2017-07-21 Carl Love <cel@us.ibm.com>
22407 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22408 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22409 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22410 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22411 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22412 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22413 VMULOSW): New enum "unspec" values.
22414 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22415 altivec_vmulosw): New patterns.
22416 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22417 VMULOSW): Add definitions.
22419 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22421 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22422 (qdf24xx): Likewise.
22423 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22424 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22425 (AARCH64_FL_V8_1): Renumber.
22426 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22427 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22428 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22429 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22430 rdma to feature modifiers list.
22432 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22434 PR middle-end/56727
22435 * ipa-visibility (function_and_variable_visibility): Convert
22436 recursive PLT call to direct call if appropriate.
22438 2017-07-21 Andrew Pinski <apinski@cavium.com>
22440 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22441 operand 1 to see if the types precision matches.
22442 * fold-const.c (operand_equal_p): Likewise.
22444 2017-07-21 Richard Biener <rguenther@suse.de>
22446 PR tree-optimization/81303
22447 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22448 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22449 (vect_peeling_hash_get_lowest_cost): Adjust.
22450 (vect_enhance_data_refs_alignment): Likewise. Use
22451 vect_get_peeling_costs_all_drs to compute the penalty for no
22452 peeling to match up costs.
22454 2017-07-21 Richard Biener <rguenther@suse.de>
22456 PR tree-optimization/81500
22457 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22458 we didn't identify a reduction path.
22460 2017-07-21 Tom de Vries <tom@codesourcery.com>
22461 Cesar Philippidis <cesar@codesourcery.com>
22463 PR gcov-profile/81442
22464 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22467 2017-07-21 Tom de Vries <tom@codesourcery.com>
22470 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22472 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22473 nvptx_override_options_after_change.
22475 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22477 * dwarf2out.c (output_file_names): Avoid double testing for
22478 dwarf_version >= 5.
22480 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22482 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22484 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22486 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22488 (try_crossjump_to_edge): Do not punt on partitioned functions.
22490 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22492 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22493 Put all BBs reachable only via paths crossing cold region to cold
22495 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22497 2016-07-21 Richard Biener <rguenther@suse.de>
22499 PR tree-optimization/81303
22500 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22501 into account prologue and epilogue iterations when raising
22502 min_profitable_iters to sth at least covering one vector iteration.
22504 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22506 * config/arm/arm.c (arm_test_cpu_arch_dat):
22509 2017-07-20 Nathan Sidwell <nathan@acm.org>
22511 Remove TYPE_METHODS.
22512 * tree.h (TYPE_METHODS): Delete.
22513 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22514 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22515 (dbxout_type_methods): Scan TYPE_FIELDS.
22516 (dbxout_type): Don't check TYPE_METHODS here.
22517 * function.c (use_register_for_decl): Always ignore register for
22518 class types when not optimizing.
22519 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22520 * tree.c (free_lang_data_in_type): Stitch out member functions and
22521 templates from TYPE_FIELDS.
22522 (build_distinct_type_copy, verify_type_variant,
22523 verify_type): Member fns are on TYPE_FIELDS.
22524 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22525 * tree-pretty-print.c (dump_generic_node): Likewise.
22527 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22530 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22531 V2TImode and V4TImode.
22532 (ix86_expand_vector_extract): Likewise.
22533 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22534 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22535 (ssescalarmode): Handle V4TImode and V2TImode.
22536 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22537 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22538 (VEXTRACTI128_MODE): New mode iterator.
22539 (splitter for *vec_extractv?ti first element): New.
22540 (VEC_INIT_MODE): New mode iterator.
22541 (vec_init<mode>): Consolidate 3 expanders into one using
22542 VEC_INIT_MODE mode iterator.
22544 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22546 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22547 non_spilled_static_chain_regno_p.
22549 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22551 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22553 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22555 * bb-reorder.c (connect_traces): Allow copying of blocks within
22558 2017-07-20 Richard Biener <rguenther@suse.de>
22560 * gimple.h (gimple_phi_result): Add gphi * overload.
22561 (gimple_phi_result_ptr): Likewise.
22562 (gimple_phi_arg): Likewise. Adjust index assert to only
22563 allow actual argument accesses rather than all slots available
22565 (gimple_phi_arg_def): Add gphi * overload.
22566 * tree-phinodes.c (make_phi_node): Initialize only actual
22568 (resize_phi_node): Clear memory not covered by old node,
22569 do not initialize excess argument slots.
22570 (reserve_phi_args_for_new_edge): Initialize new argument slot
22573 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22575 PR tree-optimization/81388
22577 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22579 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22580 by removing computation of may_be_zero.
22582 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22583 Tom de Vries <tom@codesourcery.com>
22585 PR middle-end/81030
22586 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22587 when gimple level profile disagrees with what RTL expander did.
22589 2017-07-20 Richard Biener <rguenther@suse.de>
22591 PR tree-optimization/61171
22592 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22593 (vect_analyze_stmt): Add slp instance parameter.
22594 (vectorizable_reduction): Likewise.
22595 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22596 (vect_is_simple_reduction): Deal with chains not detected
22597 as SLP reduction chain, specifically not properly associated
22598 chains containing a mix of plus/minus.
22599 (get_reduction_op): Remove.
22600 (get_initial_defs_for_reduction): Simplify, pass in whether
22601 this is a reduction chain, pass in the SLP node for the PHIs.
22602 (vect_create_epilog_for_reduction): Get the SLP instance as
22604 (vectorizable_reduction): Get the SLP instance as arg.
22605 During analysis remember the SLP node with the PHIs in the
22606 instance. Simplify getting at the vectorized reduction PHIs.
22607 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22608 through SLP instance.
22609 (vect_slp_analyze_operations): Likewise.
22610 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22611 (vect_transform_stmt): Likewise.
22613 2017-07-20 Tom de Vries <tom@codesourcery.com>
22615 PR tree-optimization/81489
22616 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22617 read of phi arg location to before loop that modifies phi.
22619 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22621 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22624 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22626 PR middle-end/81331
22627 * except.c (execute): Fix ordering issue.
22629 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22631 PR rtl-optimization/81423
22632 * combine.c (make_compound_operation_int): Don't try to optimize
22633 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22635 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22637 PR rtl-optimization/81423
22638 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22639 with a constant that is -1 in the truncated to mode.
22641 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22643 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22644 (determine_unlikely_bbs): ... here.
22645 * predict.h (propagate_unlikely_bbs_forward): Declare.
22646 * cfgexpand.c (pass_expand::execute): Use it.
22647 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22649 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22650 propagate_unlikely_bbs_forward.
22652 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22654 PR middle-end/81331
22655 * except.c (maybe_add_nop_after_section_switch): New function.
22658 2017-07-19 Tom de Vries <tom@codesourcery.com>
22660 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22662 2017-07-19 Tom de Vries <tom@codesourcery.com>
22664 * gimple.h (gimple_phi_arg): Make assert more strict.
22666 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22668 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22669 * config/rs6000/mmintrin.h: New file.
22670 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22672 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22674 PR tree-optimization/81346
22675 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22677 2017-07-19 Tom de Vries <tom@codesourcery.com>
22679 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22681 2017-07-19 Tom de Vries <tom@codesourcery.com>
22683 * config/nvptx/nvptx-modes.def: Add V2DImode.
22684 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22685 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22686 (nvptx_output_mov_insn): Handle lack of mov.b128.
22687 (nvptx_print_operand): Handle 'H' and 'L' codes.
22688 (nvptx_vector_mode_supported): Allow V2DImode.
22689 (nvptx_preferred_simd_mode): New function.
22690 (nvptx_data_alignment): New function.
22691 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22692 nvptx_preferred_simd_mode.
22693 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22695 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22697 2017-07-19 Tom de Vries <tom@codesourcery.com>
22699 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22700 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22701 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22702 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22703 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22704 (mov<VECIM>_insn): New define_insn.
22705 (define_expand "mov<VECIM>): New define_expand.
22707 2017-07-19 Tom de Vries <tom@codesourcery.com>
22709 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22711 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22713 PR tree-optimization/81346
22714 * fold-const.h (fold_div_compare, range_check_type): Declare.
22715 * fold-const.c (range_check_type): New function.
22716 (build_range_check): Use range_check_type.
22717 (fold_div_compare): No longer static, rewritten into
22718 a match.pd helper function.
22719 (fold_comparison): Don't call fold_div_compare here.
22720 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22721 as helper function.
22723 2017-07-19 Nathan Sidwell <nathan@acm.org>
22725 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22726 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22727 * tree.c (find_decls_types_r, verify_type): Use
22728 TYPE_{MIN,MAX}_VALUE_RAW.
22729 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22730 (hash_tree): Likewise.
22731 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22733 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22736 2017-07-18 Tom de Vries <tom@codesourcery.com>
22738 PR middle-end/81464
22739 * omp-expand.c (expand_omp_for_static_chunk): Handle
22740 equal-argument loop exit phi.
22742 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22745 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22746 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22747 operand 2 predicate.
22748 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22749 operand 2 predicate.
22750 (ror,rol -> rorx splitters): Use const_int_operand as
22751 operand 2 predicate.
22753 2017-06-18 Richard Biener <rguenther@suse.de>
22755 PR tree-optimization/81410
22756 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22757 the gap in the ! slp_perm SLP case after each group.
22759 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22761 PR middle-end/81463
22762 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22765 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22767 PR middle-end/81462
22768 * predict.c (set_even_probabilities): Cleanup; do not affect
22769 probabilities that are already known.
22770 (combine_predictions_for_bb): Call even when count is set.
22772 2017-07-18 Nathan Sidwell <nathan@acm.org>
22774 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22777 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22780 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22781 optimization for loop niter analysis.
22783 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22786 * config/avr/avr.c (avr_optimize_casesi): Don't use
22787 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22789 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22791 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22792 body_cost_vec from _vect_peel_extended_info.
22793 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
22794 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
22797 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22799 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
22801 2017-07-18 Richard Biener <rguenther@suse.de>
22803 PR tree-optimization/80620
22804 PR tree-optimization/81403
22805 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
22806 info when re-using a VN table entry.
22808 2017-07-18 Richard Biener <rguenther@suse.de>
22810 PR tree-optimization/81418
22811 * tree-vect-loop.c (vectorizable_reduction): Properly compute
22812 vectype_in. Verify that with lane-reducing reduction operations
22813 we have a single def-use cycle.
22815 2017-07-17 Carl Love <cel@us.ibm.com>
22817 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
22819 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22820 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22821 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22822 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22823 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22824 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22825 VMULOSW): New enum "unspec" values.
22826 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22827 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22828 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22829 altivec_vmulosw): New patterns.
22830 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22831 VMULOSW): Add definitions.
22833 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
22835 * config/alpha/alpha.c: Include predict.h.
22837 2017-07-17 Yury Gribov <tetra2005@gmail.com>
22839 * tree-vrp.c (compare_assert_loc): Fix comparison function
22840 to return predictable results.
22842 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22844 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
22846 (subdi3): Likewise.
22847 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
22848 * doc/invoke.texi (mexpand-adddi): Update text.
22850 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22852 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
22853 that also clobbers the CC register. The old expand code is moved
22855 (*arc_clzsi2): ... here.
22856 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
22857 the CC register. The old expand code is moved to ...
22858 (arc_ctzsi2): ... here.
22860 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22862 * config/arc/arc.opt (mindexed-loads): Use initial value
22863 TARGET_INDEXED_LOADS_DEFAULT.
22864 (mauto-modify-reg): Use initial value
22865 TARGET_AUTO_MODIFY_REG_DEFAULT.
22866 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22867 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22868 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22869 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22871 2017-07-17 Martin Liska <mliska@suse.cz>
22874 * opts.c (finish_options): Do not allow -fgnu-tm
22875 w/ -fsanitize={kernel-,}address. Say sorry.
22877 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22880 * tree-loop-distribution.c (classify_partition): Only assert on
22881 numer of iterations.
22882 (merge_dep_scc_partitions): Delete prameter. Update function call.
22883 (distribute_loop): Remove code handling loop with unknown niters.
22884 (pass_loop_distribution::execute): Skip loop with unknown niters.
22886 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22889 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
22890 function sort_partitions_by_post_order.
22892 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22894 PR tree-optimization/81374
22895 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
22896 the max index of basic blocks, rather than number of basic blocks.
22898 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22900 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
22902 (arc_legitimate_pic_operand_p): Likewise.
22903 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
22905 (arc_needs_pcl_p): Likewise.
22906 (arc_legitimate_pc_offset_p): Likewise.
22907 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
22908 function is also used in constrains.md.
22909 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
22910 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
22911 PLUS. Only return true/false in known cases, otherwise assert.
22912 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
22913 is already called in arc_legitimate_constant_p.
22914 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
22916 (LEGITIMATE_PIC_OPERAND_P): Use
22917 arc_raw_symbolic_reference_mentioned_p function.
22918 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
22923 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22924 Andrew Burgess <andrew.burgess@embecosm.com>
22926 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
22927 (arc_return_address_register): New function.
22928 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
22929 (arc_handle_fndecl_attribute): Add naked attribute.
22930 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22931 (TARGET_WARN_FUNC_RETURN): Likewise.
22932 (arc_allocate_stack_slots_for_args): New function.
22933 (arc_warn_func_return): Likewise.
22934 (machine_function): Change type fn_type.
22935 (arc_compute_function_type): Consider new naked function type,
22936 change function return type.
22937 (arc_must_save_register): Adapt to handle new
22938 arc_compute_function_type's return type.
22939 (arc_expand_prologue): Likewise.
22940 (arc_expand_epilogue): Likewise.
22941 (arc_return_address_regs): Delete.
22942 (arc_return_address_register): New function.
22943 (arc_epilogue_uses): Use above function.
22944 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
22945 (arc_function_type): Change encoding, add naked type.
22946 (ARC_INTERRUPT_P): Change to handle the new encoding.
22947 (ARC_FAST_INTERRUPT_P): Likewise.
22948 (ARC_NORMAL_P): Define.
22949 (ARC_NAKED_P): Likewise.
22950 (arc_compute_function_type): Delete prototype.
22951 * config/arc/arc.md (in_ret_delay_slot): Use
22952 arc_return_address_register function.
22953 (simple_return): Likewise.
22954 (p_return_i): Likewise.
22956 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22958 PR tree-optimization/81428
22959 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
22960 can't be built for those types.
22962 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22964 Remove stuff dead since r239246.
22966 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
22967 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
22968 (avr_inform_devices): Remove dead stuff.
22970 2017-07-17 Tamar Christina <tamar.christina@arm.com>
22972 * config/arm/arm_neon.h: Fix softp typo.
22974 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22976 PR tree-optimization/81365
22977 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
22978 aggregate moves onto bb predecessor edges, make sure there are no
22979 loads that could alias the lhs in between the start of bb and the
22982 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22985 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
22986 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
22987 [LSHIFTRT, outer_code = TRUNCATE]: Same.
22989 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22991 PR tree-optimization/81396
22992 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
22993 (init_symbolic_number): Initialize it to 1.
22994 (perform_symbolic_merge): Add n_ops from both operands into the new
22996 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
22997 without base_addr as useless if they need more than one operation.
22998 (bswap_replace): Handle !bswap case for NULL base_addr.
23000 2017-07-17 Tom de Vries <tom@codesourcery.com>
23003 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23006 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23008 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23009 conditional builtin define __FIX_LEON3FT_B2BST.
23011 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23013 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23014 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23017 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23019 PR rtl-optimization/81424
23020 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23021 to remove potential trapping from operands if -fnon-call-exceptions.
23023 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23025 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23026 profile_proability for scalling.
23027 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23029 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23031 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23033 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23035 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23036 fixpoint arithmetics.
23038 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23040 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23041 fixpoint arithmetics.
23043 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23045 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23046 fixpoint arithmetics.
23048 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23050 * profile-count.h (profile_probability::from_reg_br_prob_note,
23051 profile_probability::to_reg_br_prob_note): New functions.
23052 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23053 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23054 * predict.c (probability_reliable_p): Update.
23055 (edge_probability_reliable_p): Update.
23056 (br_prob_note_reliable_p): Update.
23057 (invert_br_probabilities): Update.
23058 (add_reg_br_prob_note): New function.
23059 (combine_predictions_for_insn): Update.
23060 * asan.c (asan_clear_shadow): Update.
23061 * cfgbuild.c (compute_outgoing_frequencies): Update.
23062 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23063 (update_br_prob_note): Update.
23064 (rtl_verify_edges): Update.
23065 (purge_dead_edges): Update.
23066 (fixup_reorder_chain): Update.
23067 * emit-rtl.c (try_split): Update.
23068 * ifcvt.c (cond_exec_process_insns): Update.
23069 (cond_exec_process_if_block): Update.
23070 (dead_or_predicable): Update.
23071 * internal-fn.c (expand_addsub_overflow): Update.
23072 (expand_neg_overflow): Update.
23073 (expand_mul_overflow): Update.
23074 * loop-doloop.c (doloop_modify): Update.
23075 * loop-unroll.c (compare_and_jump_seq): Update.
23076 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23077 * predict.h: Update.
23078 * reorg.c (mostly_true_jump): Update.
23080 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23081 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23082 * config/arc/arc.c: (emit_unlikely_jump): Update.
23083 * config/arm/arm.c: (emit_unlikely_jump): Update.
23084 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23085 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23086 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23087 (ix86_print_operand): Update.
23088 (ix86_split_fp_branch): Update.
23089 (predict_jump): Update.
23090 * config/ia64/ia64.c (ia64_print_operand): Update.
23091 * config/mmix/mmix.c (mmix_print_operand): Update.
23092 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23093 (rs6000_expand_split_stack_prologue): Update.
23094 * config/rs6000/rs6000.c: Update.
23095 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23096 (s390_expand_vec_movstr): Update.
23097 (s390_expand_cs_tdsi): Update.
23098 (s390_expand_split_stack_prologue): Update.
23099 * config/sh/sh.c (sh_print_operand): Update.
23100 (expand_cbranchsi4): Update.
23101 (expand_cbranchdi4): Update.
23102 * config/sparc/sparc.c (output_v9branch): Update.
23103 * config/spu/spu.c (get_branch_target): Update.
23104 (ea_load_store_inline): Update.
23105 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23106 * config/tilepro/tilepro.c: Update.
23108 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23110 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23111 (gimplify_save_expr): Likewise.
23113 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23115 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23117 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23119 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23121 * timevar.def (TV_IPA_FNSUMMARY): Define.
23123 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23125 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23126 to back store errata sensitive sequence from being generated.
23127 (sqrtdf2_fix): Likewise.
23129 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23131 * tree-ssa-threadupdate.c (compute_path_counts,
23132 update_joiner_offpath_counts): Use profile_probability.
23134 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23137 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23139 * config/arm/arm-c.c (arm_cpu_builtins): Define
23140 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23142 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23144 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23145 array entries to represent __ieee128 versions of the
23146 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23147 scalar_extract_sig, and scalar_insert_exp built-in functions.
23148 (altivec_resolve_overloaded_builtin): Add special case handling
23149 for the __builtin_scalar_insert_exp function, as represented by
23150 the P9V_BUILTIN_VEC_VSIEDP constant.
23151 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23152 exponent support for __ieee128 argument.
23153 (VSESQP): Add scalar extract signature support for __ieee128
23155 (VSTDCNQP): Add scalar test negative support for __ieee128
23157 (VSIEQP): Add scalar insert exponent support for __int128 argument
23158 with __ieee128 result.
23159 (VSIEQPF): Add scalar insert exponent support for __ieee128
23160 argument with __ieee128 result.
23161 (VSTDCQP): Add scalar test data class support for __ieee128
23163 (VSTDCNQP): Add overload support for scalar test negative with
23164 __ieee128 argument.
23165 (VSTDCQP): Add overload support for scalar test data class
23166 __ieee128 argument.
23167 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23168 UNSPEC_VSX_SXSIGDP.
23169 (UNSPEC_VSX_SIEXPQP): New constant.
23170 (xsxexpqp): New insn for VSX scalar extract exponent quad
23172 (xsxsigqp): New insn for VSX scalar extract significand quad
23174 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23175 precision with floating point argument.
23176 (xststdcqp): New expand for VSX scalar test data class quad
23178 (xststdcnegqp): New expand for VSX scalar test negative quad
23180 (xststdcqp): New insn to match expansions for VSX scalar test data
23181 class quad precision and VSX scalar test negative quad precision.
23182 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23183 special case operand checking to enforce that second operand of
23184 VSX scalar test data class with quad precision argument is a 7-bit
23186 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23187 prototypes and descriptions of __ieee128 versions of
23188 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23189 scalar_test_data_class, and scalar_test_neg built-in functions.
23191 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23193 PR tree-optimization/81162
23194 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23195 replace a negate with an add.
23197 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23199 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23201 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23203 * config/arm/arm-c.c (arm_cpu_builtins): Define
23204 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23206 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23208 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23209 (armv8-r): Set ARM Cortex-R52 as default CPU.
23210 * config/arm/arm-tables.opt: Regenerate.
23211 * config/arm/arm-tune.md: Regenerate.
23212 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23214 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23215 extension for -mcpu=cortex-r52.
23217 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23219 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23220 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23221 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23222 (fp-armv8): Define it as FP_ARMv8 only.
23223 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23224 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23226 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23227 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23228 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23229 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23230 than TARGET_FPU_ARMV8.
23231 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23232 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23233 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23235 * config/arm/neon.md (neon_vrint): Likewise.
23236 (neon_vcvt): Likewise.
23237 (neon_<fmaxmin_op><mode>): Likewise.
23238 (<fmaxmin><mode>3): Likewise.
23239 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23240 * config/arm/predicates.md (arm_cond_move_operator): Check against
23241 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23243 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23245 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23246 to top of function.
23248 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23250 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23251 loop in comment with memset.
23253 2017-07-14 Martin Liska <mliska@suse.cz>
23255 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23256 * dwarf2out.c (is_java): Remove the function.
23257 (output_pubname): Remove usage of the function.
23258 (lower_bound_default): Remove usage of DW_LANG_Java.
23259 (gen_compile_unit_die): Likewise.
23260 * gcc.c: Remove compiler defaults for .java and .zip files.
23261 * gimple-expr.c (remove_suffix): Change as there's no longer
23262 extension than 4-letter one.
23263 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23264 (gimplify_save_expr): Likewise.
23265 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23266 as it's possible even for other languages than Java.
23267 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23268 * lto-opts.c (lto_write_options): Remove reference to Java.
23269 * opts.c (strip_off_ending): Update file extension handling.
23270 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23271 * tree-eh.c (lower_resx): Likewise.
23272 * tree.c (free_lang_data_in_type): Remove dead code.
23273 (find_decls_types_r): Likewise.
23274 (build_common_builtin_nodes): Remove Java from a comment.
23275 (verify_type): Remove dead code.
23276 * varasm.c (assemble_external): Remove Java from a comment.
23278 2017-07-14 Martin Liska <mliska@suse.cz>
23280 * opts.c (finish_options): Add quotes.
23281 (common_handle_option): Likewise.
23283 2017-07-14 Martin Liska <mliska@suse.cz>
23285 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23286 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23287 Remove N_SO_PASCAL.
23288 * dwarf2out.c (lower_bound_default): Do not handle
23290 (gen_compile_unit_die): Likewise.
23291 * gcc.c: Remove default extension binding for GNU Pascal.
23292 * stmt.c: Remove Pascal language from a comment.
23293 * xcoffout.c: Likewise.
23295 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23298 * diagnostic-show-locus.c (fixit_cmp): New function.
23299 (layout::layout): Sort m_fixit_hints.
23300 (column_range::column_range): Assert that the values are valid.
23301 (struct char_span): New struct.
23302 (correction::overwrite): New method.
23303 (struct source_line): New struct.
23304 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23305 calls in terms of classes source_line and char_span, and
23306 correction::overwrite.
23307 (selftest::test_overlapped_fixit_printing_2): New function.
23308 (selftest::diagnostic_show_locus_c_tests): Call it.
23310 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23312 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23313 early if there is no lhs.
23315 2017-07-13 Martin Liska <mliska@suse.cz>
23317 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23318 (gen_reference_type_die): Likewise.
23319 * stor-layout.c: Remove Pascal-related comment.
23321 2017-07-13 Martin Liska <mliska@suse.cz>
23323 * opts.c (finish_options): Add quotes to error messages.
23324 (parse_sanitizer_options): Likewise.
23326 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23328 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23330 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23332 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23334 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23336 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23338 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23340 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23343 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23344 provides the hardware capability bits, define the macro
23345 __BUILTIN_CPU_SUPPORTS__.
23346 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23347 if GLIBC does not provide the hardware capability bits. Add a
23348 gcc_unreachable call if the built-in cpu function is neither
23349 __builtin_cpu_is nor __builtin_cpu_supports.
23350 (rs6000_get_function_versions_dispatcher): Change the warning
23351 that an old GLIBC is used which does not export the capability
23352 bits to be an error.
23353 * doc/extend.texi (target_clones attribute): Document the
23354 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23355 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23356 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23357 the macros defined by GCC if the newer GLIBC is available.
23359 2017-07-12 Jeff Law <law@redhat.com>
23361 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23362 remaining includes slightly.
23363 * config/riscv/riscv-builtins.c: Include profile-count.h.
23365 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23368 * config/avr/avr.c (avr_set_current_function): In diagnostic
23369 messages: Quote keywords and (parts of) identifiers.
23370 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23373 2017-07-12 Carl Love <cel@us.ibm.com>
23375 * config/rs6000/rs6000-c.c: Add support for built-in functions
23376 vector bool char vec_revb (vector bool char);
23377 vector bool short vec_revb (vector short char);
23378 vector bool int vec_revb (vector bool int);
23379 vector bool long long vec_revb (vector bool long long);
23380 * doc/extend.texi: Update the built-in documentation file for the
23381 new built-in functions.
23383 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23385 * config/s390/s390.md: Remove movcc splitter.
23387 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23389 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23390 load/store on condition.
23392 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23395 * config/avr/avr.c (avr_encode_section_info)
23396 [progmem && !TREE_READONLY]: Error if progmem object needs
23399 2017-07-11 Michael Collison <michael.collison@arm.com>
23401 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23404 2017-07-11 Carl Love <cel@us.ibm.com>
23406 * config/rs6000/rs6000-c.c: Add support for builtins
23407 vector unsigned int vec_parity_lsbb (vector signed int);
23408 vector unsigned int vec_parity_lsbb (vector unsigned int);
23409 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23410 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23411 vector unsigned long long vec_parity_lsbb (vector signed long long);
23412 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23413 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23414 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23415 * doc/extend.texi: Update the built-in documentation file for the
23416 new built-in functions.
23418 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23420 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23421 (layout::m_primary_loc): New field.
23422 (layout::layout): Initialize new field. Move location filtering
23423 logic from here to...
23424 (layout::maybe_add_location_range): ...this new method. Add
23425 support for filtering to just the lines already specified by other
23427 (layout::will_show_line_p): New method.
23428 (gcc_rich_location::add_location_if_nearby): New method.
23429 (selftest::test_add_location_if_nearby): New test function.
23430 (selftest::diagnostic_show_locus_c_tests): Call it.
23431 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23434 2017-07-11 Tom de Vries <tom@codesourcery.com>
23436 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23437 (bb_first_real_insn): New function.
23438 (nvptx_single): Add extra initialization of broadcasted condition
23441 2017-07-11 Nathan Sidwell <nathan@acm.org>
23443 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23445 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23447 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23448 Binutils doc as TEXI will mess them up.
23449 * doc/invoke.texi (AVR Options): Same here.
23451 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23453 * config/sparc/sparc.opt (mfix-ut700): New option.
23454 (mfix-gr712rc): Likewise.
23455 (sparc_fix_b2bst): New variable.
23456 * doc/invoke.texi (SPARC options): Document them.
23457 (ARM options): Fix warnings.
23458 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23459 instructions to prevent sequences that can trigger the store-store
23460 errata for certain LEON3FT processors.
23461 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23462 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23463 * config/sparc/sparc.md (fix_b2bst): New attribute.
23464 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23466 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23469 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23471 (*rsqrtsf2_sse): Ditto.
23473 (div<mode>3): Macroize insn from divdf3 and divsf3
23474 using MODEF mode iterator.
23476 2017-07-10 Martin Sebor <msebor@redhat.com>
23478 PR tree-optimization/80397
23479 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23480 instead of testing for equality to INTEGER_TYPE.
23482 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23484 * config.gcc: Remove uclibc from arc target spec.
23486 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23488 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23490 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23493 * lto-wrapper.c (remove_option): New function.
23494 (merge_and_complain): Merge PIC/PIE options more realistically.
23496 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23498 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23502 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23503 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23504 * config.in: Regenerate.
23505 * configure: Regenerate.
23506 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23507 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23508 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23509 (TARGET_GASISR_PROLOGUES): ...target mask.
23510 * common/config/avr/avr-common.c
23511 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23512 Set -mgas-isr-prologues.
23513 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23514 INSERT_PASS_BEFORE for it.
23515 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23516 * config/avr/avr.c (avr_option_override)
23517 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23518 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23519 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23520 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23521 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23523 (make_avr_pass_pre_proep): New function.
23524 (emit_push_sfr) <treg>: Add argument to function and use it
23525 instead of TMP_REG.
23526 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23527 and set machine->gasisr.yes.
23528 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23529 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23530 __gcc_isr.n_pushed to .L__stack_usage.
23531 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23532 (avr_asm_final_postscan_insn): ...this new static function.
23533 * config/avr/avr.h (machine_function)
23534 <is_no_gccisr, use_L__stack_usage>: New fields.
23535 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23536 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23537 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23538 (gasisr, *gasisr): New expander and insn.
23539 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23540 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23541 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23543 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23545 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23548 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23550 Move jump-tables out of .text again.
23553 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23554 (ASM_OUTPUT_ADDR_VEC): New function.
23555 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23556 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23557 INSN_ADDRESSes as asm comment.
23558 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23559 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23560 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23561 * config/avr/avr.md (*tablejump): Adjust comment.
23562 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23563 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23565 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23566 (avr_output_addr_vec): New proto.
23567 (avr_log_t) <insn_addresses>: New field.
23569 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23572 * config/i386/i386.c (ix86_function_arg_advance): Set
23573 outgoing_args_on_stack to true if there are outgoing arguments
23575 (ix86_function_arg): Likewise.
23576 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23577 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23578 * config/i386/i386.h (machine_function): Add
23579 outgoing_args_on_stack.
23581 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23583 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23584 supporting pthreds.
23585 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23587 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23589 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23590 (REAL_H): Remove $(MACHMODE_H).
23591 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23593 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23594 $(MACHMODE_H) and double-int.h.
23595 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23597 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23600 2017-07-07 Andrew Pinski <apinski@cavium.com>
23602 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23603 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23605 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23607 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23608 Add warning if GCC was not configured to link against a GLIBC that
23609 exports the hardware capability bits.
23610 (make_resolver_func): Make resolver function private and not a
23611 COMDAT function. Create the name with clone_function_name instead
23612 of make_unique_name.
23615 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23616 correct operand in doing the split.
23618 2017-07-07 Carl Love <cel@us.ibm.com>
23620 * config/rs6000/rs6000-c: Add support for built-in function
23621 vector unsigned short vec_pack_to_short_fp32 (vector float,
23623 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23624 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23625 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23626 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23627 (convert_4f32_8i16): Add define_expand.
23628 * doc/extend.texi: Update the built-in documentation file for the
23629 new built-in function.
23631 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23633 * config/sparc/m8.md: New file.
23634 * config/sparc/sparc.md: Include m8.md.
23636 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23638 * config/sparc/sparc.opt: New option -mvis4b.
23639 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23640 (sparc_option_override): Handle VIS4B.
23641 (enum sparc_builtins): Define
23642 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23643 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23644 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23645 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23646 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23647 (check_constant_argument): New function.
23648 (sparc_vis_init_builtins): Define builtins
23649 __builtin_vis_dictunpack{8,16,32},
23650 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23651 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23652 __builtin_vis_fpcmpde{8,16,32}shl and
23653 __builtin_vis_fpcmpur{8,16,32}shl.
23654 (sparc_expand_builtin): Check that the constant operands to
23655 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23656 constant and in range.
23657 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23659 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23660 (SPARC_IMM5_P): Likewise.
23661 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23662 (enabled): Handle vis4b.
23663 (UNSPEC_DICTUNPACK): New unspec.
23664 (UNSPEC_FPCMPSHL): Likewise.
23665 (UNSPEC_FPUCMPSHL): Likewise.
23666 (UNSPEC_FPCMPDESHL): Likewise.
23667 (UNSPEC_FPCMPURSHL): Likewise.
23668 (cpu_feature): New CPU feature `vis4b'.
23669 (dictunpack{8,16,32}): New insns.
23670 (FPCSMODE): New mode iterator.
23671 (fpcscond): New code iterator.
23672 (fpcsucond): Likewise.
23673 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23674 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23675 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23676 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23677 * config/sparc/constraints.md: Define constraints `q' for unsigned
23678 2-bit integer constants and `t' for unsigned 5-bit integer
23680 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23682 (imm5_operand_dictunpack16): Likewise.
23683 (imm5_operand_dictunpack32): Likewise.
23684 (imm2_operand): Likewise.
23685 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23686 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23687 ditunpack* and fpcmp*shl builtins.
23689 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23691 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23692 * config.in: Add HAVE_AS_SPARC6 define.
23693 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23695 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23697 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23698 (CPP_CPU_SPEC): Handle m8.
23699 (ASM_CPU_SPEC): Likewise.
23700 * config/sparc/sparc-opts.h (enum processor_type): Add
23702 * config/sparc/sparc.c (m8_costs): New struct.
23703 (sparc_option_override): Handle TARGET_CPU_m8.
23704 (sparc32_initialize_trampoline): Likewise.
23705 (sparc64_initialize_trampoline): Likewise.
23706 (sparc_issue_rate): Likewise.
23707 (sparc_register_move_cost): Likewise.
23708 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23709 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23710 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23711 (CPP_CPU_SPEC): Handle M8.
23712 (ASM_CPU_SPEC): Likewise.
23713 (AS_M8_FLAG): Define.
23714 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23715 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23716 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23718 * configure: Regenerate.
23719 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23722 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23724 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23726 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23727 ("*movdi_insn_sp32"): Do not set v3pipe.
23728 ("*movsi_insn"): Likewise.
23729 ("*movdi_insn_sp64"): Likewise.
23730 ("*movsf_insn"): Likewise.
23731 ("*movdf_insn_sp32"): Likewise.
23732 ("*movdf_insn_sp64"): Likewise.
23733 ("*zero_extendsidi2_insn_sp64"): Likewise.
23734 ("*sign_extendsidi2_insn"): Likewise.
23735 ("*mov<VM32:mode>_insn"): Likewise.
23736 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23737 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23738 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23739 ("<vlop:code><VL:mode>3"): Likewise.
23740 ("*not_<vlop:code><VL:mode>3"): Likewise.
23741 ("*nand<VL:mode>_vis"): Likewise.
23742 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23743 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23744 ("one_cmpl<VL:mode>2"): Likewise.
23745 ("faligndata<VM64:mode>_vis"): Likewise.
23746 ("alignaddrsi_vis"): Likewise.
23747 ("alignaddrdi_vis"): Likweise.
23748 ("alignaddrlsi_vis"): Likewise.
23749 ("alignaddrldi_vis"): Likewise.
23750 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23751 ("bmaskdi_vis"): Likewise.
23752 ("bmasksi_vis"): Likewise.
23753 ("bshuffle<VM64:mode>_vis"): Likewise.
23754 ("cmask8<P:mode>_vis"): Likewise.
23755 ("cmask16<P:mode>_vis"): Likewise.
23756 ("cmask32<P:mode>_vis"): Likewise.
23757 ("pdistn<P:mode>_vis"): Likewise.
23758 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23760 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23762 * config/sparc/sparc.md ("subtype"): New insn attribute.
23763 ("*wrgsr_sp64"): Set insn subtype.
23764 ("*rdgsr_sp64"): Likewise.
23765 ("alignaddrsi_vis"): Likewise.
23766 ("alignaddrdi_vis"): Likewise.
23767 ("alignaddrlsi_vis"): Likewise.
23768 ("alignaddrldi_vis"): Likewise.
23769 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23770 ("fexpand_vis"): Likewise.
23771 ("fpmerge_vis"): Likewise.
23772 ("faligndata<VM64:mode>_vis"): Likewise.
23773 ("bshuffle<VM64:mode>_vis"): Likewise.
23774 ("cmask8<P:mode>_vis"): Likewise.
23775 ("cmask16<P:mode>_vis"): Likewise.
23776 ("cmask32<P:mode>_vis"): Likewise.
23777 ("fchksm16_vis"): Likewise.
23778 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23779 ("fmean16_vis"): Likewise.
23780 ("fp<plusminus_insn>64_vis"): Likewise.
23781 ("<plusminus_insn>v8qi3"): Likewise.
23782 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23783 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23784 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23785 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23786 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23787 ("*movqi_insn"): Likewise.
23788 ("*movhi_insn"): Likewise.
23789 ("*movsi_insn"): Likewise.
23790 ("movsi_pic_gotdata_op"): Likewise.
23791 ("*movdi_insn_sp32"): Likewise.
23792 ("*movdi_insn_sp64"): Likewise.
23793 ("movdi_pic_gotdata_op"): Likewise.
23794 ("*movsf_insn"): Likewise.
23795 ("*movdf_insn_sp32"): Likewise.
23796 ("*movdf_insn_sp64"): Likewise.
23797 ("*zero_extendhisi2_insn"): Likewise.
23798 ("*zero_extendqihi2_insn"): Likewise.
23799 ("*zero_extendqisi2_insn"): Likewise.
23800 ("*zero_extendqidi2_insn"): Likewise.
23801 ("*zero_extendhidi2_insn"): Likewise.
23802 ("*zero_extendsidi2_insn_sp64"): Likewise.
23803 ("ldfsr"): Likewise.
23804 ("prefetch_64"): Likewise.
23805 ("prefetch_32"): Likewise.
23806 ("tie_ld32"): Likewise.
23807 ("tie_ld64"): Likewise.
23808 ("*tldo_ldub_sp32"): Likewise.
23809 ("*tldo_ldub1_sp32"): Likewise.
23810 ("*tldo_ldub2_sp32"): Likewise.
23811 ("*tldo_ldub_sp64"): Likewise.
23812 ("*tldo_ldub1_sp64"): Likewise.
23813 ("*tldo_ldub2_sp64"): Likewise.
23814 ("*tldo_ldub3_sp64"): Likewise.
23815 ("*tldo_lduh_sp32"): Likewise.
23816 ("*tldo_lduh1_sp32"): Likewise.
23817 ("*tldo_lduh_sp64"): Likewise.
23818 ("*tldo_lduh1_sp64"): Likewise.
23819 ("*tldo_lduh2_sp64"): Likewise.
23820 ("*tldo_lduw_sp32"): Likewise.
23821 ("*tldo_lduw_sp64"): Likewise.
23822 ("*tldo_lduw1_sp64"): Likewise.
23823 ("*tldo_ldx_sp64"): Likewise.
23824 ("*mov<VM32:mode>_insn"): Likewise.
23825 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23826 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23828 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23830 * config/sparc/sparc.md ("type"): New insn type viscmp.
23831 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
23833 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
23834 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
23835 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23836 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
23838 ("n7_vis_logical_11cycle"): Likewise.
23839 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
23840 * config/sparc/niagara2.md ("niag3_vis": Likewise.
23841 * config/sparc/niagara.md ("niag_vis"): Likewise.
23842 * config/sparc/ultra3.md ("us3_fga"): Likewise.
23843 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
23845 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23847 * config/sparc/sparc.md: New instruction type `bmask'.
23848 (bmaskdi_vis): Use the `bmask' type.
23849 (bmasksi_vis): Likewise.
23850 * config/sparc/ultra3.md (us3_array): Likewise.
23851 * config/sparc/niagara7.md (n7_array): Likewise.
23852 * config/sparc/niagara4.md (n4_array): Likewise.
23853 * config/sparc/niagara2.md (niag2_vis): Likewise.
23854 (niag3_vis): Likewise.
23855 * config/sparc/niagara.md (niag_vis): Likewise.
23857 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23859 * ipa-comdats.c: Remove optimize check from gate.
23860 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
23861 for functions not optimized.
23862 (ipa_fn_summary_read): Skip optimize check.
23863 (ipa_fn_summary_write): Likewise.
23864 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
23866 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
23868 (can_inline_edge_p): Check flag_pcc_struct_return for match.
23869 (check_callers): Give up on caller which is not optimized.
23870 (inline_small_functions): Likewise.
23871 (ipa_inline): Do not give up when not optimizing.
23872 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
23873 away unoptimizes cdtors.
23874 (whole_program_function_and_variable_visibility): Do
23875 ipa_discover_readonly_nonaddressable_vars in LTO mode.
23876 * ipa.c (process_references): Do not check optimize.
23877 (symbol_table::remove_unreachable_nodes): Update optimize check.
23878 (set_writeonly_bit): Update optimize check.
23879 (pass_ipa_cdtor_merge::gate): Do not check optimize.
23880 (pass_ipa_single_use::gate): Remove.
23882 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23884 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
23885 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
23886 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
23887 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
23888 permute_load, permute_store, adjust_extract, adjust_splat,
23889 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
23890 replace_swap_with_copy, dump_swap_insn_table,
23891 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
23892 recombine_lvx_pattern, recombine_stvx_pattern,
23893 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
23894 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
23895 to file rs6000-p8swap.c.
23896 * config/rs6000/rs6000-p8swap.c: New file.
23897 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
23898 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
23899 and rs6000*-*-* targets.
23901 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23903 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
23905 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23907 * lto-wrapper.c (merge_and_complain): Do not merge
23908 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
23909 fsigned_zeros, ftrapping_math, fwrapv.
23910 (append_compiler_options): Do not track these options.
23911 (append_linker_options): Likewie
23913 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23915 * cgraphunit.c (cgraph_node::finalize_function): When
23916 !flag_toplevel_reorde set no_reorder flag.
23917 (varpool_node::finalize_decl): Likewise.
23918 (symbol_table::compile): Drop no toplevel reorder path.
23920 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23922 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
23923 edges; zero probability is not better than uninitialized.
23925 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23927 * asan.h (asan_sanitize_allocas_p): Declare.
23928 * asan.c (asan_sanitize_allocas_p): New function.
23929 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
23930 (handle_builtin_alloca): Likewise.
23931 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
23932 if !asan_sanitize_allocas_p.
23933 * params.def (asan-instrument-allocas): Add new option.
23934 * params.h (ASAN_PROTECT_ALLOCAS): Define.
23935 * opts.c (common_handle_option): Disable allocas sanitization for
23938 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23940 * asan.c: Include gimple-fold.h.
23941 (get_last_alloca_addr): New function.
23942 (handle_builtin_stackrestore): Likewise.
23943 (handle_builtin_alloca): Likewise.
23944 (asan_emit_allocas_unpoison): Likewise.
23945 (get_mem_refs_of_builtin_call): Add new parameter, remove const
23946 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
23947 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
23948 (instrument_builtin_call): Pass gimple iterator to
23949 get_mem_refs_of_builtin_call.
23950 (last_alloca_addr): New global.
23951 * asan.h (asan_emit_allocas_unpoison): Declare.
23952 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
23953 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
23954 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
23955 if function calls alloca.
23956 * gimple-fold.c (replace_call_with_value): Remove static keyword.
23957 * gimple-fold.h (replace_call_with_value): Declare.
23958 * internal-fn.c: Include asan.h.
23959 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
23960 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
23962 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23964 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
23965 (C_SELFTEST_FLAGS): New.
23966 (CPP_SELFTEST_FLAGS): New.
23967 (SELFTEST_DEPS): New, from deps of s-selftest.
23968 (C_SELFTEST_DEPS): New, from deps of s-selftest.
23969 (CPP_SELFTEST_DEPS): New.
23970 (selftest): Add dependency on s-selftest-c++.
23971 (s-selftest): Rename to...
23972 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
23973 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
23974 than SELFTEST_FLAGS.
23975 (selftest-gdb): Rename to...
23976 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
23978 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
23979 (selftest-valgrind): Rename to...
23980 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
23982 (selftest-valgrind): Reintroduce as an alias for
23983 selftest-c-valgrind.
23984 (s-selftest-c++): New.
23985 (selftest-c++-gdb): New.
23986 (selftest-c++-valgrind): New.
23988 2017-07-06 Olivier Hainque <hainque@adacore.com>
23990 * gcc.c (process_command): When deciding if undefined variables
23991 should be ignored when processing specs, accept "gcc -v" as well.
23993 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23995 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
23996 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
23998 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24000 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24001 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24002 * config/arm/arm-tables.opt: Regenerate.
24003 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24005 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24007 2017-07-06 Carl Love <cel@us.ibm.com>
24009 * ChangeLog: Clean up from mid air collision
24011 2017-07-06 Carl Love <cel@us.ibm.com>
24013 * config/rs6000/rs6000-c.c: Add support for built-in functions
24014 vector signed int vec_subc (vector signed int, vector signed int);
24015 vector signed __int128 vec_subc (vector signed __int128,
24016 vector signed __int128);
24017 vector unsigned __int128 vec_subc (vector unsigned __int128,
24018 vector unsigned __int128);
24019 vector signed int vec_sube (vector signed int, vector signed int,
24020 vector signed int);
24021 vector unsigned int vec_sube (vector unsigned int,
24022 vector unsigned int,
24023 vector unsigned int);
24024 vector signed __int128 vec_sube (vector signed __int128,
24025 vector signed __int128,
24026 vector signed__int128);
24027 vector unsigned __int128 vec_sube (vector unsigned __int128,
24028 vector unsigned __int128,
24029 vector unsigned __int128);
24030 vector signed int vec_subec (vector signed int, vector signed int,
24031 vector signed int);
24032 vector unsigned int vec_subec (vector unsigned int,
24033 vector unsigned int,
24034 vector unsigned int);
24035 vector signed __int128 vec_subec (vector signed __int128,
24036 vector signed __int128,
24037 vector signed__int128);
24038 vector unsigned __int128 vec_subec (vector unsigned __int128,
24039 vector unsigned __int128,
24040 vector unsigned __int128);
24041 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24042 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24043 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24044 BU_ALTIVEC_OVERLOAD_X definitions.
24045 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24046 * doc/extend.texi: Update the built-in documentation file for the new
24047 built-in functions.
24049 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24052 * diagnostic-show-locus.c (layout::layout): Use start and finish
24053 spelling location for the start and finish of each range.
24054 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24055 Add unused aspect param.
24056 * input.c (expand_location_1): Add "aspect" param, and use it
24057 to access the correct part of the location.
24058 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24060 (expand_location_to_spelling_point): Likewise.
24061 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24062 param, and pass it to expand_location_1.
24064 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24066 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24067 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24068 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24069 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24070 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24071 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24072 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24073 _mm_maskz_getmant_ss): New intrinsics.
24074 (__builtin_ia32_getexpss128_mask): Changed to ...
24075 __builtin_ia32_getexpss128_round ... this.
24076 (__builtin_ia32_getexpsd128_mask): Changed to ...
24077 __builtin_ia32_getexpsd128_round ... this.
24078 * config/i386/i386-builtin-types.def
24079 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24080 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24081 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24082 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24083 __builtin_ia32_getmantss_mask_round): New builtins.
24084 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24085 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24086 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24087 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24088 * config/i386/sse.md
24089 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24090 avx512f_sgetexp<mode><mask_scalar_name>
24091 <round_saeonly_scalar_name> ... this.
24092 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24093 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24094 vgetexp<ssescalarmodesuffix>
24095 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24096 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24097 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24098 avx512f_vgetmant<mode><mask_scalar_name>
24099 <round_saeonly_scalar_name> ... this.
24100 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24101 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24102 vgetmant<ssescalarmodesuffix>
24103 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24104 %0<mask_scalar_operand4>, %1, %2
24105 <round_saeonly_scalar_mask_op4>, %3} ... this.
24106 * config/i386/subst.md (mask_scalar_operand4,
24107 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24108 round_saeonly_scalar_nimm_predicate): New subst attributes.
24110 2017-07-06 Julia Koval <julia.koval@intel.com>
24112 * config/i386/i386.c (ix86_erase_embedded_rounding):
24113 Remove code for old rounding pattern.
24115 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24117 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24119 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24121 * doc/sourcebuild.texi (Test Directives, Variants of
24122 dg-require-support): Add documentation for dg-require-stack-check.
24124 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24126 * config/i386/subst.md (mask_scalar, round_scalar,
24127 round_saeonly_scalar): New meta-templates.
24128 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24129 round_scalar_mask_operand3, round_scalar_mask_op3,
24130 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24131 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24132 round_saeonly_scalar_constraint,
24133 round_saeonly_scalar_prefix): New subst attribute.
24134 * config/i386/sse.md
24135 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24136 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24137 <round_scalar_name> ... this.
24138 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24139 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24140 <round_scalar_name> ... this.
24141 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24142 <sse>_vm<code><mode>3<mask_scalar_name>
24143 <round_saeonly_scalar_name> ... this.
24144 (v<plusminus_mnemonic><ssescalarmodesuffix>
24145 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24146 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24147 v<plusminus_mnemonic><ssescalarmodesuffix>
24148 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24149 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24150 (v<multdiv_mnemonic><ssescalarmodesuffix>
24151 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24152 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24153 v<multdiv_mnemonic><ssescalarmodesuffix>
24154 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24155 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24156 (v<maxmin_float><ssescalarmodesuffix>
24157 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24158 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24159 v<maxmin_float><ssescalarmodesuffix>
24160 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24161 %0<mask_scalar_operand3>, %1, %<iptr>2
24162 <round_saeonly_scalar_mask_op3>} ... this.
24164 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24166 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24167 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24169 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24170 Alan Hayward <alan.hayward@arm.com>
24171 David Sherwood <david.sherwood@arm.com>
24173 * combine.c (simplify_if_then_else): Remove "enum" before
24175 * compare-elim.c (can_eliminate_compare): Likewise.
24176 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24178 (aarch64_lookup_simd_builtin_type): Likewise.
24179 (aarch64_simd_builtin_type): Likewise.
24180 (aarch64_init_simd_builtin_types): Likewise.
24181 (aarch64_simd_expand_args): Likewise.
24182 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24184 (aarch64_reverse_mask): Likewise.
24185 (aarch64_simd_emit_reg_reg_move): Likewise.
24186 (aarch64_gen_adjusted_ldpstp): Likewise.
24187 (aarch64_ccmp_mode_to_code): Likewise.
24188 (aarch64_operands_ok_for_ldpstp): Likewise.
24189 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24190 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24192 (aarch64_min_divisions_for_recip_mul): Likewise.
24193 (aarch64_reassociation_width): Likewise.
24194 (aarch64_get_condition_code_1): Likewise.
24195 (aarch64_simd_emit_reg_reg_move): Likewise.
24196 (aarch64_simd_attr_length_rglist): Likewise.
24197 (aarch64_reverse_mask): Likewise.
24198 (aarch64_operands_ok_for_ldpstp): Likewise.
24199 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24200 (aarch64_gen_adjusted_ldpstp): Likewise.
24201 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24203 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24204 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24205 (arm_lookup_simd_builtin_type): Likewise.
24206 (arm_simd_builtin_type): Likewise.
24207 (arm_init_simd_builtin_types): Likewise.
24208 (arm_expand_builtin_args): Likewise.
24209 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24210 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24211 (ft32_setup_incoming_varargs): Likewise.
24212 (ft32_function_arg): Likewise.
24213 (ft32_function_arg_advance): Likewise.
24214 (ft32_pass_by_reference): Likewise.
24215 (ft32_arg_partial_bytes): Likewise.
24216 (ft32_valid_pointer_mode): Likewise.
24217 (ft32_addr_space_pointer_mode): Likewise.
24218 (ft32_addr_space_legitimate_address_p): Likewise.
24219 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24221 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24222 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24223 (iamcu_alignment): Likewise.
24224 (canonicalize_vector_int_perm): Likewise.
24225 (ix86_noce_conversion_profitable_p): Likewise.
24226 (ix86_mpx_bound_mode): Likewise.
24227 (ix86_operands_ok_for_move_multiple): Likewise.
24228 * config/microblaze/microblaze-protos.h
24229 (microblaze_expand_conditional_branch_reg): Likewise.
24230 * config/microblaze/microblaze.c
24231 (microblaze_expand_conditional_branch_reg): Likewise.
24232 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24234 (rs6000_reassociation_width): Likewise.
24235 (rs6000_invalid_binary_op): Likewise.
24236 (fusion_p9_p): Likewise.
24237 (emit_fusion_p9_load): Likewise.
24238 (emit_fusion_p9_store): Likewise.
24239 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24241 (riscv_hard_regno_mode_ok_p): Likewise.
24242 (riscv_address_insns): Likewise.
24243 (riscv_split_symbol): Likewise.
24244 (riscv_legitimize_move): Likewise.
24245 (riscv_function_value): Likewise.
24246 (riscv_hard_regno_nregs): Likewise.
24247 (riscv_expand_builtin): Likewise.
24248 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24249 (riscv_build_integer): Likewise.
24250 (riscv_split_integer): Likewise.
24251 (riscv_legitimate_constant_p): Likewise.
24252 (riscv_cannot_force_const_mem): Likewise.
24253 (riscv_regno_mode_ok_for_base_p): Likewise.
24254 (riscv_valid_base_register_p): Likewise.
24255 (riscv_valid_offset_p): Likewise.
24256 (riscv_valid_lo_sum_p): Likewise.
24257 (riscv_classify_address): Likewise.
24258 (riscv_legitimate_address_p): Likewise.
24259 (riscv_address_insns): Likewise.
24260 (riscv_load_store_insns): Likewise.
24261 (riscv_force_binary): Likewise.
24262 (riscv_split_symbol): Likewise.
24263 (riscv_force_address): Likewise.
24264 (riscv_legitimize_address): Likewise.
24265 (riscv_move_integer): Likewise.
24266 (riscv_legitimize_const_move): Likewise.
24267 (riscv_legitimize_move): Likewise.
24268 (riscv_address_cost): Likewise.
24269 (riscv_subword): Likewise.
24270 (riscv_output_move): Likewise.
24271 (riscv_canonicalize_int_order_test): Likewise.
24272 (riscv_emit_int_order_test): Likewise.
24273 (riscv_function_arg_boundary): Likewise.
24274 (riscv_pass_mode_in_fpr_p): Likewise.
24275 (riscv_pass_fpr_single): Likewise.
24276 (riscv_pass_fpr_pair): Likewise.
24277 (riscv_get_arg_info): Likewise.
24278 (riscv_function_arg): Likewise.
24279 (riscv_function_arg_advance): Likewise.
24280 (riscv_arg_partial_bytes): Likewise.
24281 (riscv_function_value): Likewise.
24282 (riscv_pass_by_reference): Likewise.
24283 (riscv_setup_incoming_varargs): Likewise.
24284 (riscv_print_operand): Likewise.
24285 (riscv_elf_select_rtx_section): Likewise.
24286 (riscv_save_restore_reg): Likewise.
24287 (riscv_for_each_saved_reg): Likewise.
24288 (riscv_register_move_cost): Likewise.
24289 (riscv_hard_regno_mode_ok_p): Likewise.
24290 (riscv_hard_regno_nregs): Likewise.
24291 (riscv_class_max_nregs): Likewise.
24292 (riscv_memory_move_cost): Likewise.
24293 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24294 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24295 (rl78_addr_space_address_mode): Likewise.
24296 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24298 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24299 (rs6000_reassociation_width): Likewise.
24300 (rs6000_invalid_binary_op): Likewise.
24301 (fusion_p9_p): Likewise.
24302 (emit_fusion_p9_load): Likewise.
24303 (emit_fusion_p9_store): Likewise.
24304 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24305 (ok_for_simple_move_operands): Likewise.
24306 (ok_for_simple_move_strict_operands): Likewise.
24307 (ok_for_simple_arith_logic_operands): Likewise.
24308 (visium_legitimize_reload_address): Likewise.
24309 (visium_select_cc_mode): Likewise.
24310 (output_cbranch): Likewise.
24311 (visium_split_double_move): Likewise.
24312 (visium_expand_copysign): Likewise.
24313 (visium_expand_int_cstore): Likewise.
24314 (visium_expand_fp_cstore): Likewise.
24315 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24316 (visium_function_arg): Likewise.
24317 (visium_function_arg_advance): Likewise.
24318 (visium_libcall_value): Likewise.
24319 (visium_setup_incoming_varargs): Likewise.
24320 (visium_legitimate_constant_p): Likewise.
24321 (visium_legitimate_address_p): Likewise.
24322 (visium_legitimize_address): Likewise.
24323 (visium_secondary_reload): Likewise.
24324 (visium_register_move_cost): Likewise.
24325 (visium_memory_move_cost): Likewise.
24326 (prepare_move_operands): Likewise.
24327 (ok_for_simple_move_operands): Likewise.
24328 (ok_for_simple_move_strict_operands): Likewise.
24329 (ok_for_simple_arith_logic_operands): Likewise.
24330 (visium_function_value_1): Likewise.
24331 (rtx_ok_for_offset_p): Likewise.
24332 (visium_legitimize_reload_address): Likewise.
24333 (visium_split_double_move): Likewise.
24334 (visium_expand_copysign): Likewise.
24335 (visium_expand_int_cstore): Likewise.
24336 (visium_expand_fp_cstore): Likewise.
24337 (visium_split_cstore): Likewise.
24338 (visium_select_cc_mode): Likewise.
24339 (visium_split_cbranch): Likewise.
24340 (output_cbranch): Likewise.
24341 (visium_print_operand_address): Likewise.
24342 * expmed.c (flip_storage_order): Likewise.
24343 * expmed.h (emit_cstore): Likewise.
24344 (flip_storage_order): Likewise.
24345 * genrecog.c (validate_pattern): Likewise.
24346 * hsa-gen.c (gen_hsa_addr): Likewise.
24347 * internal-fn.c (expand_arith_overflow): Likewise.
24348 * ira-color.c (allocno_copy_cost_saving): Likewise.
24349 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24350 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24351 (process_invariant_for_inheritance): Likewise.
24352 * lra-eliminations.c (move_plus_up): Likewise.
24353 * omp-low.c (lower_oacc_reductions): Likewise.
24354 * simplify-rtx.c (simplify_subreg): Likewise.
24355 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24356 (TARGET_CHKP_BOUND_MODE): Likewise..
24357 * targhooks.c (default_chkp_bound_mode): Likewise.
24358 (default_setup_incoming_vararg_bounds): Likewise.
24359 * targhooks.h (default_chkp_bound_mode): Likewise.
24360 (default_setup_incoming_vararg_bounds): Likewise.
24361 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24362 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24363 (have_whole_vector_shift): Likewise.
24364 * tree-vect-stmts.c (vectorizable_load): Likewise.
24365 * doc/tm.texi: Regenerate.
24367 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24369 Graceful degrade if Binutils PR21472 is not available.
24372 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24373 .rodata in flash test fails.
24374 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24375 * confgure: Regenerate.
24376 * config.in: Regenerate.
24377 * config/avr/avr.c (avr_asm_named_section)
24378 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24379 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24380 (avr_asm_init_sections): Same.
24382 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24384 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24385 (fma<VH:mode>4_intrinsic): Likewise.
24386 (*fmsub<VCVTF:mode>4): Likewise.
24387 (*fmsub<VH:mode>4_intrinsic): Likewise.
24389 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24392 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24393 Don't depend on "optimize > 0".
24394 (out_movhi_r_mr, out_movqi_mr_r): Same.
24395 (out_movhi_mr_r, out_movqi_r_mr): Same.
24396 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24397 io_address_operand on "optimize > 0".
24399 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24401 * tree-loop-distribution.c: Add general explanantion on the pass.
24402 (generate_loops_for_partition): Mark distributed loop.
24403 (pg_add_dependence_edges): New parameter. Handle alias data
24404 dependence specially and record it in the parameter if asked.
24405 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24406 (init_partition_graph_vertices, add_partition_graph_edge): New.
24407 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24408 (free_partition_graph_vdata, build_partition_graph): New.
24409 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24410 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24411 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24412 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24413 (version_for_distribution_p, finalize_partitions): New.
24414 (distribute_loop): Handle alias data dependence specially. Factor
24415 out loop fusion code as functions and call these functions.
24417 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24419 * tree-loop-distribution.c (classify_partition): New parameter and
24420 better handle reduction statement.
24421 (rdg_build_partitions): Revise comment.
24422 (distribute_loop): Compute statements in all partitions and pass it
24423 to classify_partition.
24425 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24427 * tree-loop-distribution.c (enum partition_type): New.
24428 (struct partition): New field type.
24429 (partition_merge_into): Add parameter. Update partition type.
24430 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24431 (build_rdg_partition_for_vertex): Compute partition type.
24432 (rdg_build_partitions): Dump partition type.
24433 (distribute_loop): Update calls to partition_merge_into.
24435 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24437 * tree-loop-distribution.c (struct ddr_hasher): New.
24438 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24440 (classify_partition): Call get_data_dependence.
24441 (pg_add_dependence_edges): Ditto.
24442 (distribute_loop): Release data dependence hash table.
24444 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24446 * tree-loop-distribution.c (ref_base_address): Delete.
24447 (similar_memory_accesses): Rename ...
24448 (share_memory_accesses): ... to this. Check if partitions access
24449 the same memory reference.
24450 (distribute_loop): Call share_memory_accesses.
24452 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24454 * tree-loop-distribution.c (struct partition): New field recording
24455 its data reference.
24456 (partition_alloc, partition_free): Init and release data refs.
24457 (partition_merge_into): Merge data refs.
24458 (build_rdg_partition_for_vertex): Collect data refs for partition.
24459 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24461 (distribute_loop): Remve data refs from vertice data of partition
24464 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24466 * tree-loop-distribution.c (params.h): Include header file.
24467 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24468 (datarefs_vec): New global var.
24469 (create_rdg_vertices): Use datarefs_vec directly.
24470 (free_rdg): Don't free data references.
24471 (build_rdg): Update use. Don't free data references.
24472 (distribute_loop): Compute global variable for data references.
24473 Bail out if there are too many data references.
24475 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24477 * tree-loop-distribution.c (loop_nest): New global var.
24478 (build_rdg): Use loop directly, rather than loop nest.
24479 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24481 (distribute_loop): Compute global variable loop nest. Update use.
24483 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24485 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24486 (partition_merge_into): New parameter. Dump reason for fusion.
24487 (distribute_loop): Update use of partition_merge_into.
24489 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24491 * tree-loop-distribution.c (bb_top_order_index): New.
24492 (bb_top_order_index_size, bb_top_order_cmp): New.
24493 (stmts_from_loop): Use topological order.
24494 (pass_loop_distribution::execute): Compute and release topological
24495 order for basic blocks.
24497 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24499 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24502 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24504 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24505 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24506 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24507 * internal-fn.def (LOOP_DIST_ALIAS): New.
24508 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24509 (fold_loop_internal_call): ... this.
24510 (vect_loop_dist_alias_call): New function.
24511 (set_uid_loop_bbs): Call fold_loop_internal_call.
24512 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24515 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24518 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24519 Require dead FLAGS_REG at the beginning of a peephole.
24521 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24524 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24525 arguments in the call to __builtin_ia32_sbb_u32.
24526 (_subborrow_u64): Swap _X and _Y arguments in the call to
24527 __builtin_ia32_sbb_u64.
24529 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24532 * tree-vrp.c (compare_assert_loc): Turn into a function template
24533 with stable template parameter. Only test if a->e is NULL,
24534 !a->e == !b->e has been verified already. Use e == NULL or
24535 e != NULL instead of e or ! e tests. If stable is true, don't use
24536 iterative_hash_expr, on the other side allow a or b or both NULL
24537 and sort the NULLs last.
24538 (process_assert_insertions): Sort using compare_assert_loc<false>
24539 instead of compare_assert_loc, later sort using
24540 compare_assert_loc<true> before calling process_assert_insertions_for
24541 in a loop. Use break instead of continue once seen NULL pointer.
24543 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24545 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24546 Cortex-R7 and Cortex-R8 processors.
24548 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24550 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24551 uninitialized while src is not.
24553 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24555 * common/config/arm/arm-common.c: Adjust include path for
24557 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24558 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24559 (arm-cpu-data.h): Likewise.
24560 (arm-cpu-cdata.h): Likewise.
24561 * config/arm/arm-cpu.h: Delete.
24562 * config/arm/arm-cpu-cdata.h: Delete.
24563 * config/arm/arm-cpu-data.h: Delete.
24565 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24567 * config/arm/arm-cpus.in (cortex-a55): New.
24568 (cortex-a75): Likewise.
24569 (cortex-a75.cortex-a55): Likewise.
24570 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24572 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24573 * config/arm/arm-cpu-cdata.h: Regenerate.
24574 * config/arm/arm-cpu-data.h: Regenerate.
24575 * config/arm/arm-cpu.h: Regenerate.
24576 * config/arm/arm-tables.opt: Regenerate.
24577 * config/arm/arm-tune.md: Regenerate.
24579 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24581 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24583 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24585 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24588 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24590 PR tree-optimization/81292
24591 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24592 full_string_p, also call adjust_related_strinfos if the adjustment
24593 is simple, otherwise invalidate related strinfos.
24595 2017-07-04 Martin Liska <mliska@suse.cz>
24598 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24599 newly created variable as DECL_IGNORED_P.
24601 2017-07-04 Martin Liska <mliska@suse.cz>
24604 * ipa-inline.c (inline_small_functions):
24605 Use xstrdup_for_dump.
24607 2017-07-04 Tom de Vries <tom@codesourcery.com>
24609 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24611 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24614 * config/darwin.c (darwin_function_switched_text_sections):
24617 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24619 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24621 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24623 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24625 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24627 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24628 min_profitable_iters, and th as inclusive lower bounds.
24629 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24630 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24631 for min_profitable_iters and min_profitable_estimate.
24632 (vect_transform_loop): Treat th as an inclusive lower bound.
24633 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24635 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24638 * config/darwin.c (darwin_function_switched_text_sections):
24639 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24640 in two pieces, and suppress the use of buf.
24642 2017-07-03 Nathan Sidwell <nathan@acm.org>
24644 * hash-table.h (hash_table_mod1): Fix indentation.
24646 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24648 PR middle-end/81290
24649 * predict.c (force_edge_cold): Be more careful about propagation
24651 * profile-count.h (profile_probability::guessed,
24652 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24654 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24656 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24658 * doc/invoke.texi (rcpc architecture extension): Document it.
24660 2017-07-03 Richard Biener <rguenther@suse.de>
24662 PR tree-optimization/60510
24663 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24664 the scalar reduction PHI and use it.
24665 (vectorizable_reduction): Properly guard the single_defuse_cycle
24666 path for non-SLP reduction chains where we cannot use it.
24667 Rework reduc_def/index and vector type deduction. Rework
24668 vector operand gathering during reduction op code-gen.
24669 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24670 chains dissolve the chain and leave it to non-SLP reduction
24673 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24675 * tree-data-ref.h (dr_alignment): Declare.
24676 * tree-data-ref.c (dr_alignment): New function.
24677 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24678 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24680 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24682 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24684 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24685 and base_misalignment fields.
24686 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24687 * tree-data-ref.c: Include builtins.h.
24688 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24689 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24690 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24691 * tree-vect-data-refs.c: Include tree-cfg.h.
24692 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24693 fields instead of calculating an alignment here.
24694 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24695 innermost_loop_behavior fields.
24697 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24699 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24701 (DR_STEP_ALIGNMENT): New macro.
24702 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24703 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24704 (create_data_ref): Print it.
24705 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24706 to tell whether the step preserves vector (mis)alignment.
24707 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24708 Move the check for an integer step and generalise to all INTEGER_CST.
24709 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24710 Print the outer step alignment.
24712 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24714 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24715 with offset_alignment.
24716 (DR_ALIGNED_TO): Delete.
24717 (DR_OFFSET_ALIGNMENT): New macro.
24718 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24719 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24720 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24721 (create_data_ref): Likewise.
24722 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24723 (vect_analyze_data_refs): Likewise.
24724 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24725 creating dummy innermost behavior.
24727 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24729 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24730 with a "innermost_loop_behavior *" and refeence tree.
24731 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24732 (create_data_ref): Update call accordingly.
24733 * tree-predcom.c (find_looparound_phi): Likewise.
24735 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24737 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24738 fields with dr_wrt_vec_loop.
24739 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24740 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24741 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24742 (vect_dr_behavior): New function.
24743 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24744 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24745 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24746 track whether the step preserves the misalignment.
24747 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24748 Use vect_dr_behavior.
24749 (vect_setup_realignment): Update call accordingly.
24750 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24751 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24752 call to vect_create_addr_base_for_vector_ref.
24753 (vect_create_cond_for_align_checks): Likewise.
24754 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24755 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24756 (vect_recog_mask_conversion_pattern): Likewise.
24757 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24758 (new_stmt_vec_info): Remove redundant zeroing.
24760 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24762 * common/config/arm/arm-common.c (arm_be8_option): New function.
24763 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24764 (ISA_ARMv6): Add isa_bit_be8.
24765 * config/arm/arm.h (arm_be8_option): Add prototype.
24766 (BE8_SPEC_FUNCTION): New define.
24767 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24768 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24769 (mlittle-endian): Similarly.
24770 (mbe8, mbe32): New options.
24771 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24772 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24774 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24776 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24778 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24780 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24781 (cleanup_tree_cfg_bb): Use it.
24782 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24784 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24786 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24789 * loop-doloop.c (add_test): Update profile.
24791 2017-07-03 Martin Liska <mliska@suse.cz>
24794 * sanopt.c (rewrite_usage_of_param): New function.
24795 (sanitize_rewrite_addressable_params): Likewise.
24796 (pass_sanopt::execute): Call rewrite_usage_of_param.
24798 2017-07-03 Richard Biener <rguenther@suse.de>
24800 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
24801 back to using VIEW_CONVERT_EXPR.
24803 2017-07-03 Martin Liska <mliska@suse.cz>
24806 * doc/extend.texi: Document when a resolver function is
24807 generated for target_clones.
24809 2017-07-03 Martin Liska <mliska@suse.cz>
24811 * asan.c (asan_emit_stack_protection): Unpoison just red zones
24812 and shadow memory of auto variables which are subject of
24813 use-after-scope sanitization.
24814 (asan_expand_mark_ifn): Add do set only when is_poison.
24816 2016-07-03 Richard Biener <rguenther@suse.de>
24818 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
24820 (vect_force_simple_reduction): Record reduction def -> phi mapping.
24821 (vectorizable_reduction): Perform reduction PHI creation when
24822 visiting a reduction PHI and adjust and simplify code generation
24823 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
24824 (vect_transform_loop): Visit reduction PHIs.
24825 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
24826 defs into the SLP tree.
24827 (vect_build_slp_tree): Reduction defs terminate the recursion.
24828 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
24830 (vect_get_vec_defs_for_stmt_copy): Export.
24831 (vect_get_vec_defs): Likewise.
24832 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
24834 (vect_get_vec_defs_for_stmt_copy): Declare.
24835 (vect_get_vec_defs): Likewise.
24837 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24839 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
24840 parameter with a "loop" parameter and use it instead of the
24841 loop containing DR_STMT. Don't check simple_iv when doing
24842 BB analysis. Describe the two analysis modes in the comment.
24844 2017-07-03 Tom de Vries <tom@codesourcery.com>
24846 PR tree-optimization/69468
24847 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
24848 (find_same_succ_bb): Handle ignore_edge_flags.
24850 2017-07-03 Tom de Vries <tom@codesourcery.com>
24852 PR tree-optimization/81192
24853 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
24855 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
24857 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
24859 2017-07-03 Tom de Vries <tom@codesourcery.com>
24861 PR tree-optimization/81192
24862 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
24863 BB_SAME_SUCC (bb) == NULL.
24865 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24867 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
24870 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24872 * dumpfile.c: Include profile-count.h
24873 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
24875 (insert_cond_bb): Update profile.
24876 * tree-cfg.h (insert_cond_bb): Update prototype.
24877 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
24878 * tree-dump.c: Do not include tree-cfg.
24880 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24882 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
24884 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24886 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
24889 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24891 * tree-complex.c (expand_complex_div_wide): update profile.
24893 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24894 Alan Hayward <alan.hayward@arm.com>
24895 David Sherwood <david.sherwood@arm.com>
24897 * Makefile.in (MACHMODE_H): Remove insn-modes.h
24898 (CORETYPES_H): New define.
24899 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
24900 (insn-modes-inline.h, s-modes-inline-h): New rules.
24901 (generated_files): Add insn-modes-inline.h.
24902 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
24903 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
24904 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
24905 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
24906 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
24907 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
24908 (build/gencodes.o, build/genconditions.o): Likewise.
24909 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
24910 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
24911 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
24912 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
24913 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
24914 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
24915 * coretypes.h: Include everything up to real.h for generators.
24916 Include insn-modes.h first. Include wide-int-print.h after
24917 wide-int.h. Include insn-modes-inline.h and then machmode.h.
24918 * machmode.h: Don't include insn-modes.h here.
24919 * function-tests.c: Remove includes of signop.h, machmode.h,
24920 double-int.h and wide-int.h.
24922 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
24924 * optc-save-gen.awk: Likewise.
24925 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
24926 * godump.c: Remove include of wide-int-print.h.
24927 * pretty-print.h: Likewise.
24928 * wide-int-print.cc: Likewise.
24929 * wide-int.cc: Likewise.
24930 * hash-map-tests.c: Remove include of signop.h.
24931 * hash-set-tests.c: Likewise.
24932 * rtl-tests.c: Likewise.
24933 * mkconfig.sh: Remove include of machmode.h.
24934 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
24936 (emit_insn_modes_inline_h): ...this new function. Emit the code
24937 into an insn-modes-inline.h header file, adding appropriate
24938 include guards and end comments.
24939 (emit_insn_modes_c_header): Remove include of machmode.h.
24940 (emit_min_insn_modes_c_header): Include coretypes.h rather than
24942 (main): Handle -i flag and call emit_insn_modes_inline_h when
24945 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24947 * tree-ssa-strlen.c (strinfo): Rename the length field to
24948 nonzero_chars. Add a full_string_p field.
24949 (compare_nonzero_chars, zero_length_string_p): New functions.
24950 (get_addr_stridx): Add an offset_out parameter.
24951 Use compare_nonzero_chars.
24952 (get_stridx): Update accordingly. Use compare_nonzero_chars.
24953 (new_strinfo): Update after above changes to strinfo.
24954 (set_endptr_and_length): Set full_string_p.
24955 (get_string_length): Update after above changes to strinfo.
24956 (unshare_strinfo): Update call to new_strinfo.
24957 (maybe_invalidate): Likewise.
24958 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
24959 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
24960 as a uhwi instead of an shwi. Update after above changes to
24961 strinfo and new_strinfo.
24962 (zero_length_string): Assert that chainsi contains full strings.
24963 Use zero_length_string_p. Update call to new_strinfo.
24964 (adjust_related_strinfos): Update after above changes to strinfo.
24965 Copy full_string_p from origsi.
24966 (adjust_last_stmt): Use zero_length_string_p.
24967 (handle_builtin_strlen): Update after above changes to strinfo and
24968 new_strinfo. Install the lhs as the string length if the previous
24969 entry didn't describe a full string.
24970 (handle_builtin_strchr): Update after above changes to strinfo
24972 (handle_builtin_strcpy): Likewise.
24973 (handle_builtin_strcat): Likewise.
24974 (handle_builtin_malloc): Likewise.
24975 (handle_pointer_plus): Likewise.
24976 (handle_builtin_memcpy): Likewise. Track nonzero characters
24977 that aren't necessarily followed by a nul terminator.
24978 (handle_char_store): Likewise.
24980 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24982 PR tree-optimization/80769
24983 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
24984 for malloc and calloc. Document the new invariant that all related
24985 strinfos have delayed lengths or none do.
24986 (verify_related_strinfos): Move earlier in file.
24987 (set_endptr_and_length): New function, split out from...
24988 (get_string_length): ...here. Also set the lengths of related
24990 (zero_length_string): Assert that chainsi has known (rather than
24992 (adjust_related_strinfos): Likewise.
24994 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24996 PR tree-optimization/81136
24997 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
24998 assert that two references with the same misalignment have the same
24999 compile-time misalignment if those compile-time misalignments
25002 2017-07-01 Andi Kleen <ak@linux.intel.com>
25004 * print-tree.c (print_node): Print all attributes.
25006 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25008 * cfg.c (scale_bbs_frequencies): New function.
25009 * cfg.h (scale_bbs_frequencies): Declare it.
25010 * cfgloopanal.c (single_likely_exit): Cleanup.
25011 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25013 (scale_loop_profile): Likewise.
25014 (loop_version): Likewise.
25015 (create_empty_loop_on_edge): Update.
25016 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25017 scale_loop_frequencies, scale_loop_profile, loopify,
25018 loop_version): Update prototypes.
25019 * modulo-sched.c (sms_schedule): Update.
25020 * predict.c (unlikely_executed_edge_p): Also check probability.
25021 (probably_never_executed_edge_p): Fix typo.
25022 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25023 * tree-parloops.c (gen_parallel_loop): Update.
25024 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25025 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25026 * tree-ssa-loop-split.c (split_loop): Update.
25027 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25028 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25029 (vect_loop_versioning): Update.
25030 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25032 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25034 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25036 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25038 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25039 to keep profile consistent.
25041 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25043 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25044 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25045 * profile-count.h (max_safe_multiplier): Make unsigned.
25046 (profile_count::guessed_zero): New.
25048 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25050 * bb-reorder.c (fix_up_crossing_landing_pad,
25051 fix_crossing_conditional_branches): Use make_single_succ_edge
25052 to keep profile consistent.
25054 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25056 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25059 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25062 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25063 the right scopes, make sure cond_jump isn't preserved between multiple
25064 iterations. Search for fallthru edge whenever there are 3+ edges and
25065 use find_fallthru_edge for it.
25067 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25069 Patch by Alexander Monakov <amonakov@ispras.ru>
25070 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25071 probabilities consistently.
25073 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25075 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25076 emit_cmp_and_jump_insns.
25078 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25081 * tree-inline.c (expand_call_inline): Combine profile statuses.
25083 2017-06-30 Andrew Pinski <apinski@cavium.com>
25085 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25086 fold_stmt returned true.
25088 2017-06-30 Nathan Sidwell <nathan@acm.org>
25090 * ggc.h (empty_string): Delete.
25091 * cfgexpand.c (expand_asm_stmt): Use plain "".
25092 * optabs.c (expand_asm_memory_barrier): Likewise.
25093 * stringpool.c (empty_string): Delete.
25094 (digit_vector, digit_string): Delete.
25095 (ggc_alloc_string): Use plain "", don't optimize single digit
25096 strings. Use ggc_alloc_atomic.
25098 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25100 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25101 comparison set and one other set, use the cost of the non-comparison
25104 2017-06-30 Nathan Sidwell <nathan@acm.org>
25106 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25109 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25111 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25112 loops. Remove now unneeded calls to gimple_switch_set_label() that
25113 just set removed labels to NULL_TREE.
25115 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25117 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25118 (set_range_info): ...here. Only call set_range_info_raw if domain
25120 (set_nonzero_bits): Call set_range_info_raw.
25121 * tree-ssanames.h (set_range_info_raw): New.
25123 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25126 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25127 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25128 of nonimmediate_operand and <store_mask_constraint> instead of m
25129 for the input operand. For V8FI iterator, always split if input
25130 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25131 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25132 <store_mask_predicate> instead of register_operand and
25133 <store_mask_constraint> instead of v for the input operand. Make
25134 sure both operands aren't MEMs for if not <mask_applied>.
25136 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25138 * lto-wrapper.c (copy_file) Close both file descriptors before
25141 2017-06-30 Martin Liska <mliska@suse.cz>
25144 * multiple_target.c (create_dispatcher_calls): Make ifunc
25145 also for function that don't have calls or are not referenced.
25147 2017-06-30 Richard Biener <rguenther@suse.de>
25149 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25150 analyze the first scalar stmt. Move vector type computation
25151 for the BB case here from ...
25152 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25153 live operation processing in the SLP case properly.
25155 2017-06-30 Richard Biener <rguenther@suse.de>
25157 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25159 2017-06-30 Martin Liska <mliska@suse.cz>
25162 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25163 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25165 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25167 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25170 2017-06-30 Richard Biener <rguenther@suse.de>
25172 PR tree-optimization/81249
25173 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25174 condition reduction result to original scalar type.
25176 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25178 * profile-count.h (enum profile_quality): Fix typos and whitespace
25181 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25183 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25184 type for branch probabilities.
25186 2017-06-29 Julian Brown <julian@codesourcery.com>
25187 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25189 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25190 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25191 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25192 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25194 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25196 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25197 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25198 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25199 CC usage from generic code to here.
25200 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25201 CC usage into the target macros.
25203 2017-06-29 Maya Rashish <coypu@sdf.org>
25205 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25208 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25210 * arm/arm-builtins.c: Include profile-count.h
25211 * except.c (sjlj_emit_function_enter): Use
25212 profile_probability::unlikely.
25214 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25216 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25217 and tocrel_offset be pointer args rather than implicitly using
25219 (legitimate_constant_pool_address_p, rs6000_emit_move,
25220 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25221 tocrel_offset and use in toc_relative_expr_p call.
25222 (print_operand, print_operand_address): Use static tocrel_base_oac
25223 and tocrel_offset_oac.
25224 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25227 2017-06-29 Maya Rashish <coypu@sdf.org>
25229 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25231 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25233 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25234 objects, take into account only the alignment of 'op0' and 'mode1' if
25237 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25239 * ccmp.c (ccmp_tree_comparison_p): New function.
25240 (ccmp_candidate_p): Update to use above function.
25241 (get_compare_parts): New function.
25242 (expand_ccmp_next): Update to use new functions.
25243 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25245 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25246 take mode as argument.
25247 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25248 * expr.c (expand_expr_real_1): Pass mode as argument.
25250 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25252 * combine.c (combine_instructions): Print insns to dump_file, together
25255 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25257 * asan.c (asan_emit_stack_protection): Update.
25258 (create_cond_insert_point): Update.
25259 * auto-profile.c (afdo_propagate_circuit): Update.
25260 * basic-block.h (struct edge_def): Turn probability to
25261 profile_probability.
25262 (EDGE_FREQUENCY): Update.
25263 * bb-reorder.c (find_traces_1_round): Update.
25264 (better_edge_p): Update.
25265 (sanitize_hot_paths): Update.
25266 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25267 (make_single_succ_edge): Update.
25268 (check_bb_profile): Update.
25269 (dump_edge_info): Update.
25270 (update_bb_profile_for_threading): Update.
25271 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25272 probabilitycount to 0.
25273 * cfgbuild.c (compute_outgoing_frequencies): Update.
25274 * cfgcleanup.c (try_forward_edges): Update.
25275 (outgoing_edges_match): Update.
25276 (try_crossjump_to_edge): Update.
25277 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25278 (expand_gimple_tailcall): Update.
25279 (construct_init_block): Use make_single_succ_edge.
25280 (construct_exit_block): Use make_single_succ_edge.
25281 * cfghooks.c (verify_flow_info): Update.
25282 (redirect_edge_succ_nodup): Update.
25283 (split_edge): Update.
25284 (account_profile_record): Update.
25285 * cfgloopanal.c (single_likely_exit): Update.
25286 * cfgloopmanip.c (scale_loop_profile): Update.
25287 (set_zero_probability): Remove.
25288 (duplicate_loop_to_header_edge): Update.
25289 * cfgloopmanip.h (loop_version): Update prototype.
25290 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25291 (force_nonfallthru_and_redirect): Update.
25292 (update_br_prob_note): Update.
25293 (rtl_verify_edges): Update.
25294 (purge_dead_edges): Update.
25295 (rtl_lv_add_condition_to_bb): Update.
25296 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25297 * cgraphunit.c (init_lowered_empty_function): Update.
25298 (cgraph_node::expand_thunk): Update.
25299 * cilk-common.c: Include profile-count.h
25300 * dojump.c (inv): Remove.
25301 (jumpifnot): Update.
25302 (jumpifnot_1): Update.
25303 (do_jump_1): Update.
25305 (do_jump_by_parts_greater_rtx): Update.
25306 (do_compare_rtx_and_jump): Update.
25307 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25308 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25309 * dwarf2cfi.c: Include profile-count.h
25310 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25311 (sjlj_emit_dispatch_table): Likewise.
25312 * explow.c: Include profile-count.h
25313 * expmed.c (emit_store_flag_force): Update.
25314 (do_cmp_and_jump): Update.
25315 * expr.c (compare_by_pieces_d::generate): Update.
25316 (compare_by_pieces_d::finish_mode): Update.
25317 (emit_block_move_via_loop): Update.
25318 (store_expr_with_bounds): Update.
25319 (store_constructor): Update.
25320 (expand_expr_real_2): Update.
25321 (expand_expr_real_1): Update.
25322 * expr.h (try_casesi, try_tablejump): Update prototypes.
25323 * gimple-pretty-print.c (dump_probability): Update.
25324 (dump_profile): New.
25325 (dump_gimple_label): Update.
25326 (dump_gimple_bb_header): Update.
25327 * graph.c (draw_cfg_node_succ_edges): Update.
25328 * hsa-gen.c (convert_switch_statements): Update.
25329 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25330 (find_if_case_1): Update.
25331 (find_if_case_2): Update.
25332 * internal-fn.c (expand_arith_overflow_result_store): Update.
25333 (expand_addsub_overflow): Update.
25334 (expand_neg_overflow): Update.
25335 (expand_mul_overflow): Update.
25336 (expand_vector_ubsan_overflow): Update.
25337 * ipa-cp.c (good_cloning_opportunity_p): Update.
25338 * ipa-split.c (split_function): Use make_single_succ_edge.
25339 * ipa-utils.c (ipa_merge_profiles): Update.
25340 * loop-doloop.c (add_test): Update.
25341 (doloop_modify): Update.
25342 * loop-unroll.c (compare_and_jump_seq): Update.
25343 (unroll_loop_runtime_iterations): Update.
25344 * lra-constraints.c (lra_inheritance): Update.
25345 * lto-streamer-in.c (input_cfg): Update.
25346 * lto-streamer-out.c (output_cfg): Update.
25347 * mcf.c (adjust_cfg_counts): Update.
25348 * modulo-sched.c (sms_schedule): Update.
25349 * omp-expand.c (expand_omp_for_init_counts): Update.
25350 (extract_omp_for_update_vars): Update.
25351 (expand_omp_ordered_sink): Update.
25352 (expand_omp_for_ordered_loops): Update.
25353 (expand_omp_for_generic): Update.
25354 (expand_omp_for_static_nochunk): Update.
25355 (expand_omp_for_static_chunk): Update.
25356 (expand_cilk_for): Update.
25357 (expand_omp_simd): Update.
25358 (expand_omp_taskloop_for_outer): Update.
25359 (expand_omp_taskloop_for_inner): Update.
25360 * omp-simd-clone.c (simd_clone_adjust): Update.
25361 * optabs.c (expand_doubleword_shift): Update.
25362 (expand_abs): Update.
25363 (emit_cmp_and_jump_insn_1): Update.
25364 (expand_compare_and_swap_loop): Update.
25365 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25366 * predict.c (predictable_edge_p): Update.
25367 (edge_probability_reliable_p): Update.
25368 (set_even_probabilities): Update.
25369 (combine_predictions_for_insn): Update.
25370 (combine_predictions_for_bb): Update.
25371 (propagate_freq): Update.
25372 (estimate_bb_frequencies): Update.
25373 (force_edge_cold): Update.
25374 * profile-count.c (profile_count::dump): Add missing space into dump.
25375 (profile_count::debug): Add newline.
25376 (profile_count::differs_from_p): Explicitly convert to unsigned.
25377 (profile_count::stream_in): Update.
25378 (profile_probability::dump): New member function.
25379 (profile_probability::debug): New member function.
25380 (profile_probability::differs_from_p): New member function.
25381 (profile_probability::differs_lot_from_p): New member function.
25382 (profile_probability::stream_in): New member function.
25383 (profile_probability::stream_out): New member function.
25384 * profile-count.h (profile_count_quality): Rename to ...
25385 (profile_quality): ... this one.
25386 (profile_probability): New.
25387 (profile_count): Update.
25388 * profile.c (compute_branch_probabilities): Update.
25389 * recog.c (peep2_attempt): Update.
25390 * sched-ebb.c (schedule_ebbs): Update.
25391 * sched-rgn.c (find_single_block_region): Update.
25392 (compute_dom_prob_ps): Update.
25393 (schedule_region): Update.
25394 * sel-sched-ir.c (compute_succs_info): Update.
25395 * stmt.c (struct case_node): Update.
25396 (do_jump_if_equal): Update.
25397 (get_outgoing_edge_probs): Update.
25398 (conditional_probability): Update.
25399 (emit_case_dispatch_table): Update.
25400 (expand_case): Update.
25401 (expand_sjlj_dispatch_table): Update.
25402 (emit_case_nodes): Update.
25403 * targhooks.c: Update.
25404 * tracer.c (better_p): Update.
25405 (find_best_successor): Update.
25406 * trans-mem.c (expand_transaction): Update.
25407 * tree-call-cdce.c: Update.
25408 * tree-cfg.c (gimple_split_edge): Upate.
25409 (move_sese_region_to_fn): Upate.
25410 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25411 * tree-eh.c (lower_resx): Upate.
25412 (cleanup_empty_eh_move_lp): Upate.
25413 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25414 * tree-inline.c (copy_edges_for_bb): Update.
25415 (copy_cfg_body): Update.
25416 * tree-parloops.c (gen_parallel_loop): Update.
25417 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25418 (gimple_gen_time_profiler): Update.
25419 * tree-ssa-dce.c (remove_dead_stmt): Update.
25420 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25421 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25422 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25423 (unloop_loops): Update.
25424 (try_peel_loop): Update.
25425 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25426 * tree-ssa-loop-split.c (connect_loops): Update.
25427 (split_loop): Update.
25428 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25429 (hoist_guard): Update.
25430 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25431 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25432 (value_replacement): Update.
25433 * tree-ssa-reassoc.c (branch_fixup): Update.
25434 * tree-ssa-tail-merge.c (replace_block_by): Update.
25435 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25436 (create_edge_and_update_destination_phis): Update.
25437 (compute_path_counts): Update.
25438 (recompute_probabilities): Update.
25439 (update_joiner_offpath_counts): Update.
25440 (freqs_to_counts_path): Update.
25441 (duplicate_thread_path): Update.
25442 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25443 (struct switch_conv_info): Update.
25444 (gen_inbound_check): Update.
25445 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25446 (vect_do_peeling): Update.
25447 (vect_loop_versioning): Update.
25448 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25449 (optimize_mask_stores): Update.
25450 * ubsan.c (ubsan_expand_null_ifn): Update.
25451 * value-prof.c (gimple_divmod_fixed_value): Update.
25452 (gimple_divmod_fixed_value_transform): Update.
25453 (gimple_mod_pow2): Update.
25454 (gimple_mod_pow2_value_transform): Update.
25455 (gimple_mod_subtract): Update.
25456 (gimple_mod_subtract_transform): Update.
25457 (gimple_ic): Update.
25458 (gimple_stringop_fixed_value): Update.
25459 (gimple_stringops_transform): Update.
25460 * value-prof.h: Update.
25462 2017-06-29 Carl Love <cel@us.ibm.com>
25464 * config/rs6000/rs6000-c.c: Add support for built-in functions
25465 vector signed int vec_signed (vector float);
25466 vector signed long long vec_signed (vector double);
25467 vector signed int vec_signed2 (vector double, vector double);
25468 vector signed int vec_signede (vector double);
25469 vector signed int vec_signedo (vector double);
25470 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25471 instruction generator.
25472 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25473 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25474 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25476 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25477 vunsignede_v2df): Add define_expands.
25478 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25479 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25480 VEC_UNSIGNEDO): Add definitions.
25481 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25482 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25483 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25484 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25485 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25486 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25487 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25488 vec_unsignede, vec_unsignedo): Add builtin defines.
25489 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25491 * doc/extend.texi: Update the built-in documentation file for the
25492 new built-in functions.
25494 2017-06-29 Richard Biener <rguenther@suse.de>
25496 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25497 reduction chains to LOOP_VINFO_REDUCTIONS.
25498 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25499 SLP reductions after processing reduction chains.
25501 2017-06-29 Nathan Sidwell <nathan@acm.org>
25503 * builtins.c (fold_builtin_FUNCTION): Use
25504 lang_hooks.decl_printable_name.
25506 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25508 PR middle-end/81194
25509 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25510 with only one label.
25511 * stmt.c (expand_case): Assert NCASES is greater than one.
25513 2017-06-29 Richard Biener <rguenther@suse.de>
25515 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25517 (group_case_labels): Likewise.
25518 (find_taken_edge): Push sanity checking on val to workers...
25519 (find_taken_edge_cond_expr): ... here
25520 (find_taken_edge_switch_expr): ... and here, handle cases
25521 with just a default label.
25522 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25523 (group_case_labels): Likewise.
25524 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25525 group_case_labels does anything cleanup the CFG again.
25527 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25529 PR tree-optimization/81196
25530 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25531 exit condition comparing two IVs.
25533 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25535 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25536 profile to the dummy entry at the end of the list of architectures.
25537 * config/arm/arm-cpu-cdata.h: Regenerated.
25539 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25540 Michael Collison <michael.collison@arm.com>
25543 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25545 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25546 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25547 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25548 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25549 of shift when the shift amount is masked with constant equal to
25550 the size of the mode.
25551 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25554 2017-06-29 Martin Liska <mliska@suse.cz>
25556 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25558 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25561 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25564 (edge_growth_cache_entry::edge_growth_cache_entry): New
25566 (reset_edge_growth_cache): Update to use constructor.
25568 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25570 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25571 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25572 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25574 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25576 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25577 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25579 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25581 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25582 (*-linux-uclibc*): Add t-uclibc tmake_file.
25583 * config/t-musl: New.
25584 * config/t-uclibc: New.
25586 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25588 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25590 (gen_comm_data): Emit architectural setting of arch_prof.
25591 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25593 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25594 (armv8-m.base, armv8-m.main): Likewise.
25595 * arm-protos.h (arm_build_target): Add profile field.
25596 (arch_option): Likewise.
25597 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25599 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25600 arm_active_target.profile.
25602 2017-06-28 Richard Biener <rguenther@suse.de>
25604 PR middle-end/81227
25605 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25606 TYPE_OVERFLOW_WRAPS.
25607 * match.pd (negate_expr_p): Likewise.
25608 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25609 fold_build2, not fold_binary.
25611 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25613 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25614 Convert memory address to Pmode.
25615 (aarch64_print_operand): Assert MEM operands are always Pmode.
25617 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25620 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25621 Remove redundant if.
25622 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25623 * config/arm/aarch-common-protos.h
25624 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25625 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25627 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25630 * multiple_target.c (create_dispatcher_calls): Set the default
25631 clone to be static, not public.
25633 2017-06-28 Richard Biener <rguenther@suse.de>
25635 * tree-vect-loop.c (vectorizable_reduction): Move special
25636 cond reduction IV var creation ...
25637 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25638 parameter. Use STMT_VINFO_VECTYPE.
25639 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25642 2017-06-28 Martin Liska <mliska@suse.cz>
25645 * ipa-visibility.c (non_local_p): Handle visibility.
25647 2017-06-28 Martin Liska <mliska@suse.cz>
25650 * common.opt: Add IntegerRange to various options.
25651 * opt-functions.awk (integer_range_info): New function.
25652 * optc-gen.awk: Add integer_range_info to cl_options struct.
25653 * opts-common.c (decode_cmdline_option): Handle
25654 CL_ERR_INT_RANGE_ARG.
25655 (cmdline_handle_error): Likewise.
25656 * opts.c (print_filtered_help): Show valid interval in
25657 when --help is provided.
25658 * opts.h (struct cl_option): Add range_min and range_max fields.
25659 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25661 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25663 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25664 (x * C EQ/NE y * C): New transformation.
25666 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25668 * genmultilib (combination_space): Accept '+' in option names.
25670 2017-06-28 Martin Liska <mliska@suse.cz>
25673 * asan.c (instrument_derefs): Bail out inner references
25674 that are hard register variables.
25676 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25679 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25680 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25682 2017-06-28 Richard Biener <rguenther@suse.de>
25684 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25685 (vect_get_slp_defs): Adjust.
25686 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25688 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25690 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25691 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25692 (vectorizable_reduction): Adjust.
25693 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25695 (vect_get_slp_defs): Likewise.
25696 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25697 (vectorizable_bswap): Adjust.
25698 (vectorizable_call): Likewise.
25699 (vectorizable_conversion): Likewise.
25700 (vectorizable_assignment): Likewise.
25701 (vectorizable_shift): Likewise.
25702 (vectorizable_operation): Likewise.
25703 (vectorizable_store): Likewise.
25704 (vectorizable_condition): Likewise.
25705 (vectorizable_comparison): Likewise.
25707 2017-06-28 Michael Collison <michael.collison@arm.com>
25710 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25712 (arm_gen_movmemqi): Removed unused variable 'i'.
25713 Convert 'for' loop into 'while' loop.
25714 (arm_expand_prologue): Remove last unnecessary set of insn.
25715 (thumb_pop): Remove unused variable 'pushed_words'.
25716 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25718 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25720 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25721 * config/s390/s390-protos.h: Add prototype of
25722 s390_rel_address_ok_p.
25723 * config/s390/s390.c (s390_got_symbol): New function.
25724 (s390_rel_address_ok_p): New function.
25725 (legitimize_pic_address): Use s390_rel_address_ok_p.
25726 (s390_load_got): Use s390_got_symbol.
25727 (s390_option_override): Issue error if
25728 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25729 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25731 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25733 2017-06-27 Andrew Pinski <apinski@cavium.com>
25735 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25736 (X * copysign (1.0, X)): New pattern.
25737 (X * copysign (1.0, -X)): New pattern.
25738 (copysign (-1.0, CST)): New pattern.
25740 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25742 * genmultilib (combination_space): Remove variable.
25743 Validate reuse rules against regular expression for any sequence
25744 of multilib options in any order.
25746 2017-06-27 Michael Collison <michael.collison@arm.com>
25748 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25749 call aarch64_split_simd_combine.
25750 * (aarch64_combine_internal<mode>): Delete pattern.
25751 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25752 Allow register and subreg operands.
25754 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25756 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25757 specific need, just fallback on defaults.
25758 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25760 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25761 Olivier Hainque <hainque@adacore.com>
25763 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25765 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25766 targets. Pick a default if no particular attempt applied.
25767 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25770 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25772 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25773 (x86_64-wrs-vxworks7): Likewise.
25775 2017-06-27 Marek Polacek <polacek@redhat.com>
25778 * ubsan.c (instrument_null): Check get_base_address's result for null.
25780 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25782 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25784 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25786 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25787 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25788 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25789 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25790 New function types.
25791 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25792 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25793 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
25794 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
25795 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
25796 BUILT_IN_FEUPDATEENV): New builtins.
25797 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
25798 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
25799 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
25800 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
25802 (builtin_structptr_types): Adjust size.
25803 * tree.c (builtin_structptr_types): Add four entries.
25805 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25806 Olivier Hainque <hainque@adacore.com>
25808 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
25809 (TLS_SYM): New local macro, forcing reference to __tls__ on
25810 link command lines for VxWorks 7 RTPs, triggering initialization
25812 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
25813 OS features TLS support, true for RTPs on VxWorks 7.
25814 * config/vxworks.c (vxworks_override_options): Setup emutls
25817 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25819 * predict.c (test_prediction_value_range): Use -1U instead of -1
25820 to avoid narrowing conversion warning.
25821 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
25822 to avoid narrowing conversion warning.
25823 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
25825 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
25827 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25829 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
25830 64bit configurations.
25831 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
25832 (SIZE_TYPE): Likewise.
25833 * config/vxworks.c (vxworks_emutls_var_fields): Use
25834 long_unsigned_type_node instead of unsigned_type_node as the offset
25835 field type, which is "pointer" mode in emutls.c.
25837 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25840 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
25842 PR middle-end/81207
25843 * gimple-fold.c (replace_call_with_call_and_fold): Handle
25844 gimple_vuse copying separately from gimple_vdef copying.
25846 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25848 * value-prof.c (free_hist): Remove call to memset and the enclosing if
25851 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25852 Olivier Hainque <hainque@adacore.com>
25854 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
25855 for all vxworks7 targets.
25856 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
25857 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
25858 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
25859 variations for VX6/VX7 and 32/64bits later on in ...
25860 (VXWORKS_LIB_SPEC): Leverage new macros.
25861 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
25862 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
25864 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25866 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
25867 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
25869 2017-06-26 Carl Love <cel@us.ibm.com>
25871 * config/rs6000/rs6000-c.c: Add support for built-in functions
25872 vector bool char vec_reve (vector bool char);
25873 vector signed char vec_reve (vector signed char);
25874 vector unsigned char vec_reve (vector unsigned char);
25875 vector bool int vec_reve (vector bool int);
25876 vector signed int vec_reve (vector signed int);
25877 vector unsigned int vec_reve (vector unsigned int);
25878 vector bool long long vec_reve (vector bool long long);
25879 vector signed long long vec_reve (vector signed long long);
25880 vector unsigned long long vec_reve (vector unsigned long long);
25881 vector bool short vec_reve (vector bool short);
25882 vector signed short vec_reve (vector signed short);
25883 vector double vec_reve (vector double);
25884 vector float vec_reve (vector float);
25885 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
25886 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
25887 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
25888 (altivec_vreve): New pattern.
25889 * config/rs6000/altivec.h (vec_reve): New define.
25890 * doc/extend.texi (vec_rev): Update the built-in documentation file
25891 for the new built-in functions.
25893 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25895 PR tree-optimization/71815
25896 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
25898 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
25900 (slsr_process_phi): Likewise.
25901 (replace_uncond_cands_and_profitable_phis): Don't replace a
25902 multiply candidate with a stride of 1 (copy or cast).
25903 (phi_incr_cost): Call uses_consumed_by_stmt rather than
25905 (lowest_cost_path): Likewise.
25906 (total_savings): Likewise.
25908 2017-06-26 Richard Biener <rguenther@suse.de>
25911 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
25912 Use def_builtin_pure for all gather builtins.
25914 2017-06-26 Richard Biener <rguenther@suse.de>
25916 PR tree-optimization/81203
25917 * tree-tailcall.c (find_tail_calls): Do not move stmts into
25918 non-dominating BBs.
25920 2017-06-26 Marek Polacek <polacek@redhat.com>
25923 * doc/invoke.texi: Document -Wmultistatement-macros.
25925 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
25927 * doc/sourcebuild.texi (ARM-specific attributes): Document new
25928 arm_neon_ok_no_float_abi effective target.
25930 2017-06-26 Richard Biener <rguenther@suse.de>
25932 PR tree-optimization/80928
25933 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
25934 (copy_bbs): Set BB_DUPLICATED flag early.
25935 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
25937 (execute_on_shrinking_pred): Likewise.
25938 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
25939 BB_DUPLICATED blocks.
25940 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
25941 iterate over all PHIs considering removal of *gsi.
25943 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
25945 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
25948 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25950 * config/rs6000/rs6000-string.c: (expand_block_clear,
25951 do_load_for_compare, select_block_compare_mode,
25952 compute_current_alignment, expand_block_compare,
25953 expand_strncmp_align_check, expand_strn_compare,
25954 expand_block_move, rs6000_output_load_multiple)
25955 Move functions related to string/block move/compare
25956 to a separate file.
25957 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
25958 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
25959 for this function which is now used in two files.
25960 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
25961 * config.gcc: Add rs6000-string.o to extra_objs for
25962 targets powerpc*-*-* and rs6000*-*-*.
25964 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
25967 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
25968 32-bit, since indexed is not valid for DImode.
25969 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
25970 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
25971 (define_peephole2 for Altivec d-form load): Add 32-bit support.
25972 (define_peephole2 for Altivec d-form store): Likewise.
25975 * multiple_target.c (create_dispatcher_calls): Only create the
25976 dispatcher call if the function is the default clone of a
25977 versioned function.
25979 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
25981 PR middle-end/80902
25982 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
25983 a call, force the call to not be a tail call.
25985 2017-06-23 Jeff Law <law@redhat.com>
25987 * doc/contrib.texi: Add entry for Steven Pemberton's work on
25990 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
25992 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
25993 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
25994 handling for early expansion of vector shifts (sl,sr,sra,rl).
25995 (builtin_function_type): Add vector shift right instructions
25996 to the unsigned argument list.
25998 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26000 rtl-optimizatoin/79286
26001 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26002 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26003 trap. PIC register plus a const unspec without offset can never trap.
26005 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26007 * tree.h (builtin_structptr_type): New type.
26008 (builtin_structptr_types): Declare new array.
26009 * tree.c (builtin_structptr_types): New array.
26010 (free_lang_data, build_common_tree_nodes): Use it.
26012 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26015 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26018 2017-06-22 Matt Turner <mattst88@gmail.com>
26020 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26021 Lake models to skylake case. Assume skylake for unknown
26022 models with clflushopt.
26024 2017-06-22 Jeff Law <law@redhat.com>
26026 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26027 frame sizes that do not satisfy aarch64_uimm12_shift.
26029 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26031 * profile-count.h (apply_probability,
26032 apply_scale, probability_in): Fix checks for zero.
26034 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26036 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26037 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26039 2016-06-22 Richard Biener <rguenther@suse.de>
26041 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26042 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26043 REDUC_MAX_EXPR support.
26044 (vectorizable_reduction): Likewise.
26045 (vect_create_epilog_for_reduction): Likewise.
26047 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26049 * match.pd (A / (1 << B) -> A >> B): New.
26050 * generic-match-head.c: Include optabs-tree.h.
26051 * gimple-match-head.c: Likewise.
26052 * optabs-tree.h (target_supports_op_p): New.
26053 * optabs-tree.c (target_supports_op_p): New.
26055 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26057 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26058 $gcc_cv_ld --help output.
26059 (gcc_cv_ld_demangle): Likewise.
26060 (gcc_cv_ld_eh_frame_hdr): Likewise.
26061 (gcc_cv_ld_pie): Likewise.
26062 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26063 (gcc_cv_ld_buildid): Likewise.
26064 (gcc_cv_ld_sysroot): Likewise.
26065 (ld_bndplt_support): Likewise.
26066 (ld_pushpopstate_support): Likewise.
26067 * configure: Regenerate.
26068 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26070 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26073 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26074 operands indexes to avoid gap between operands and match_dups.
26076 2017-06-21 Andrew Pinski <apinski@cavium.com>
26078 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26079 Increment Arith_shift and Arith_shift_reg by 1.
26080 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26082 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26083 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26084 (aarch64_strip_extend): Add new argument and test for it.
26085 (aarch64_cheap_mult_shift_p): New function.
26086 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26087 add a cost if it is true.
26088 Update calls to aarch64_strip_extend.
26089 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26091 2017-06-21 Andrew Pinski <apinski@cavium.com>
26093 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26095 (thunderxt88): Likewise.
26096 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26097 (thunderx_prefetch_tune): New variable.
26098 (thunderx2t99_prefetch_tune): Update for the correct values.
26099 (thunderxt88_tunings): New variable.
26100 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26101 generic_prefetch_tune.
26102 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26104 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26106 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26107 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26108 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26109 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26110 (aarch64_atomic_cas<mode>, GPI): Likewise.
26112 2017-06-21 Martin Liska <mliska@suse.cz>
26114 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26115 statements on cold and hot labels.
26116 * predict.c (tree_estimate_probability_bb): Remove the
26117 prediction from this place.
26119 2017-06-21 Martin Liska <mliska@suse.cz>
26121 PR tree-optimization/79489
26122 * gimplify.c (maybe_add_early_return_predict_stmt): New
26124 (gimplify_return_expr): Call the function.
26125 * predict.c (tree_estimate_probability_bb): Remove handling
26127 * predict.def: Update comment about early return predictor.
26128 * gimple-predict.h (is_gimple_predict): New function.
26129 * predict.def: Change default value of early return to 66.
26130 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26132 * passes.def: Put pass_strip_predict_hints to the beginning of
26135 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26137 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26138 FUNCTION_DECL declarations.
26139 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26141 (dwaf2out_decl): Likewise.
26142 * godump.c (go_early_global_decl): Skip call to the real debug hook
26143 for FUNCTION_DECL declarations.
26144 * passes.c (rest_of_decl_compilation): Skip call to the
26145 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26146 -fdump-go-spec is passed.
26148 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26150 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26151 Reorder for compactness.
26152 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26153 (def_builtin_pure, def_builtin_pure2): New functions.
26154 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26156 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26158 * match.pd (nop_convert): New predicate.
26159 ((A +- CST1) +- CST2): Allow some NOP conversions.
26161 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26164 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26165 with ctors/dtors if GOVD_SHARED is set.
26167 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26169 * config/aarch64/aarch64.md (movti_aarch64):
26170 Emit mov rather than orr.
26171 (movtf_aarch64): Likewise.
26172 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26173 Emit mov rather than orr.
26175 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26177 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26178 Swap alternatives, make integer dup more expensive.
26180 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26182 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26183 Return true for non-tls symbols.
26185 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26187 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26188 (cortex-a75): Likewise.
26189 (cortex-a75.cortex-a55): Likewise.
26190 * config/aarch64/aarch64-tune.md: Regenerate.
26191 * doc/invoke.texi (-mtune): Document new values for -mtune.
26193 2017-06-21 Tom de Vries <tom@codesourcery.com>
26195 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26196 stack_size feature.
26197 (Effective-Target Keywords, Other attributes): Suggest using
26198 dg-add-options stack_size feature to get stack limit in stack_size
26199 effective target documentation.
26201 2017-06-21 Julian Brown <julian@codesourcery.com>
26202 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26204 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26205 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26206 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26208 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26209 attribute type list for neon_multiply.
26210 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26211 attribute type list for neon_multiply.
26212 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26213 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26214 attribute type list for neon_multiply.
26215 * config/arm/types.md (crypto_pmull): Add.
26216 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26217 attribute type list.
26219 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26221 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26224 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26226 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26227 to make sure not to dereference a NULL cost_classes_ptr pointer.
26229 2017-06-20 Carl Love <cel@us.ibm.com>
26231 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26232 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26233 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26234 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26235 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26236 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26237 VMULOSW): New enum "unspec" values.
26238 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26239 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26240 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26241 altivec_vmulosw): New patterns.
26242 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26243 VMULOSW): Add definitions.
26245 2017-06-20 Julia Koval <julia.koval@intel.com>
26247 * config/i386/i386.c: Fix rounding expand for new pattern.
26248 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26250 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26252 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26253 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26255 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26257 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26260 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26262 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26263 architecture, then by implementer ID.
26264 * config/aarch64/aarch64-tune.md: Regenerate.
26266 2017-06-20 Richard Biener <rguenther@suse.de>
26268 PR middle-end/81097
26269 * fold-const.c (split_tree): Fold to type before negating.
26271 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26273 * diagnostic-show-locus.c
26274 (selftest::test_fixit_deletion_affecting_newline): New function.
26275 (selftest::diagnostic_show_locus_c_tests): Call it.
26277 2017-06-20 Andreas Schwab <schwab@suse.de>
26280 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26283 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26285 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26286 __ARM_FEATURE_COPROC according to support.
26288 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26290 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26291 Rewritten to avoid overflow for > 32-bit pointers.
26294 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26295 by removing enum keyword.
26296 (ubsan_type_descriptor): Likewise. Formatting fix.
26299 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26300 splitter): Require TARGET_SSE2 in the condition.
26302 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26305 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26306 for doing vector set of SFmode on ISA 3.0.
26307 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26308 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26310 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26311 SFmode value into a V4SF variable that was extracted from another
26312 V4SF variable without converting the element to double precision
26313 and back to single precision vector format.
26314 (vsx_insert_extract_v4sf_p9_2): Likewise.
26316 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26318 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26319 in UWHI to avoid undefined overflow.
26322 * ubsan.h (enum ubsan_encode_value_phase): New.
26323 (ubsan_encode_value): Change second argument to
26324 enum ubsan_encode_value_phase with default value of
26325 UBSAN_ENCODE_VALUE_GENERIC.
26326 * ubsan.c (ubsan_encode_value): Change second argument to
26327 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26328 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26329 create_tmp_var_raw instead of create_tmp_var and use a
26331 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26332 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26333 ubsan_encode_value callers.
26336 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26337 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26338 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26340 2017-06-19 Richard Biener <rguenther@suse.de>
26342 PR middle-end/81118
26343 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26344 estimates if we changed anything.
26346 2017-06-19 Richard Biener <rguenther@suse.de>
26348 PR tree-optimization/80887
26349 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26350 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26351 simplified lookups, then reset mprts_hook.
26352 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26354 (try_to_simplify): Likewise.
26356 2017-06-19 Martin Liska <mliska@suse.cz>
26359 * gimplify.c (gimplify_switch_expr):
26360 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26362 2017-06-19 Martin Liska <mliska@suse.cz>
26364 * doc/install.texi: Document that PGO runs in 4 stages.
26366 2017-06-19 Martin Liska <mliska@suse.cz>
26369 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26370 to dispatcher function name.
26371 * multiple_target.c (replace_function_decl): New function.
26372 (create_dispatcher_calls): Redirect both edges and references.
26374 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26376 * profile-count.c (profile_count::dump): Dump quality.
26377 (profile_count::differs_from_p): Update for unsigned val.
26378 * profile-count.h (profile_count_quality): New enum.
26379 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26381 2017-06-19 Richard Biener <rguenther@suse.de>
26383 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26384 struct function as arg.
26385 (estimate_numbers_of_iterations): Export overload with loop arg.
26386 (free_numbers_of_iterations_estimates_loop): Use an overload of
26387 free_numbers_of_iterations_estimates instead.
26388 * tree-cfg.c (remove_bb): Adjust.
26389 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26390 * tree-parloops.c (gen_parallel_loop): Likewise.
26391 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26393 (tree_unroll_loops_completely): Likewise.
26394 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26395 Use an overload instead and export.
26396 (estimated_loop_iterations): Adjust.
26397 (max_loop_iterations): Likewise.
26398 (likely_max_loop_iterations): Likewise.
26399 (estimate_numbers_of_iterations): Take struct function as arg
26401 (loop_exits_before_overflow): Adjust.
26402 (free_numbers_of_iterations_estimates_loop): Use an overload.
26403 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26404 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26406 2017-06-19 Richard Biener <rguenther@suse.de>
26409 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26410 RANGE_EXPR conservatively.
26412 2017-06-16 Carl Love <cel@us.ibm.com>
26414 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26415 definitions for vec_float, vec_float2, vec_floato,
26416 vec_floate built-ins.
26417 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26418 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26420 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26421 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26422 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26423 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26424 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26425 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26426 vec_floato): Add builtin defines.
26427 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26428 Update the built-in documentation file for the new built-in
26431 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26433 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26434 (mthumb): Mark as the negative of -marm.
26436 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26438 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26440 (ARM Options, -mtune): Document that this accepts the same
26441 extension options as -mcpu.
26442 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26444 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26446 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26447 permitted extensions.
26449 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26451 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26452 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26453 (armv8-m.main): Add option +nodsp.
26454 * config/arm/arm-cpu-cdata.h: Regenerated.
26456 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26458 * config/arm/t-fuchsia: New file.
26459 * config.gcc (arm*-*-fuchsia*): Use it.
26461 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26463 * config/arm/t-symbian: Rewrite for new option infrastructure.
26465 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26467 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26468 (MULTILIB_REQUIRED): Likewise.
26470 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26472 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26473 (MULTILIB_RESUE): Likewise.
26474 (MULTILIB_MATCHES): Likewise.
26475 (MULTLIB_REQUIRED): Likewise.
26477 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26479 * config/arm/t-rtems: Rewrite for new option framework.
26481 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26483 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26484 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26485 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26486 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26487 * config/arm/t-multilib: ... here.
26488 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26489 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26490 armv7-a and armv8*-a when A-profile libraries have not been built.
26491 * config/arm/t-rmprofile: Rewrite.
26493 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26495 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26496 with a backslash. Remove the backslash after substituting unescaped
26498 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26500 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26502 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26503 the driver to pass through the default setting of -mfloat-abi.
26504 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26506 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26507 (all_feat_combs): New rule.
26508 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26510 * config/arm/t-aprofile: Rewrite.
26512 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26514 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26515 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26516 fpu is not specified by the user/command-line.
26517 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26518 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26519 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26520 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26521 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26522 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26524 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26526 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26527 * config/arm/t-arm-elf: Rewritten.
26529 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26531 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26532 have some floating-point instructions.
26533 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26534 (TARGET_MAYBE_HARD_FLOAT): New macro.
26535 * config/arm/arm-builtins.c (arm_init_builtins): Use
26536 TARGET_MAYBE_HARD_FLOAT.
26537 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26539 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26541 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26542 (configargs.h): Include it.
26543 (arm_print_hint_for_fpu_option): New function.
26544 (arm_parse_fpu_option): New function.
26545 (candidate_extension): New class.
26546 (arm_canon_for_multilib): New function.
26547 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26548 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26549 (ARCH_CANONICAL_SPECS): New macro.
26550 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26552 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26554 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26555 are set after handling multilib fragments. Set target_cpu_default2
26558 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26560 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26562 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26563 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26565 (check_arch): Likewise.
26566 * config/arm/arm.c (arm_configure_build_target): Handle
26567 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26568 options in the default.
26570 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26572 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26573 when an option is an alias of another.
26574 * config/arm/parsecpu.awk (optalias): New parser token.
26575 (gen_comm_data): Mark non-alias options as such. Emit entries
26576 for extension aliases.
26577 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26578 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26579 (armv6kz, armv6zk, armv6t2): Likewise.
26580 (armv7): Make vfpv3-d16 an alias.
26581 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26583 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26584 Sort in canonical order.
26585 (armv8-a): Sort in canonical order.
26586 (armv8.1-a, armv8.2-a): Likewise.
26587 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26589 (cortex-a9): Sort in canonical order.
26590 * config/arm/arm.c (selftests.h): Include it.
26591 (arm_test_cpu_arch_data): New function.
26592 (arm_run_self_tests): New function.
26593 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26594 (targetm): Move declaration to the end of the file.
26595 * arm-cpu-cdata.h: Regenerated.
26597 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26599 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26600 call to target_mode_check describing the type of option passed.
26601 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26602 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26603 arm_parse_cpu_option_name to match parameters against list of
26605 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26606 arm_arch_core_flags data structure.
26607 * config/arm/arm-cpu_cdata.h: Regenerated.
26609 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26611 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26613 (arm_print_hint_for_cpu_option): Likewise.
26614 (arm_print_hint_for_arch_option): Likewise.
26615 (arm_parse_cpu_option_name): Likewise.
26616 (arm_parse_arch_option_name): Likewise.
26617 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26618 of entries in the all_fpus list.
26619 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26620 (arm_parse_cpu_option_name): Declare.
26621 (arm_parse_arch_option_name): Declare.
26622 (arm_parse_option_features): Declare.
26623 (arm_intialize_isa): Declare.
26624 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26626 (gen_comm_data): ... here. Make definitions non-static.
26627 * config/arm/arm-cpu-data.h: Regenerated.
26628 * config/arm/arm-cpu-cdata.h: Regenerated.
26630 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26632 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26633 (cpu_arch_extension): New structure.
26634 (cpu_arch_option, arch_option, cpu_option): New structures.
26635 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26636 architecture types.
26637 (gen_data): Generate new format data tables.
26638 * config/arm/arm.c (cpu_tune): New structure.
26639 (cpu_option, processors): Delete.
26640 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26641 (arm_print_hint_for_cpu_option): ... this and ...
26642 (arm_print_hint_for_arch_option): ... this.
26643 (arm_parse_arch_cpu_name): Delete. Replace with ...
26644 (arm_parse_cpu_option_name): ... this and ...
26645 (arm_parse_arch_option_name): ... this.
26646 (arm_unrecognized_feature): Change type of target parameter to
26648 (arm_parse_arch_cpu_features): Delete. Replace with ...
26649 (arm_parse_option_features): ... this.
26650 (arm_configure_build_target): Rework to use new configuration data
26652 (arm_print_tune_info): Rework for new configuration data tables.
26653 * config/arm/arm-cpu-data.h: Regenerated.
26654 * config/arm/arm-cpu.h: Regenerated.
26656 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26658 * Makefile.in (OBJS): Move sbitmap.o from here ...
26659 (OBJS-libcommon): ... to here.
26661 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26663 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26664 (ISA_ALL_CRYPTO): New macro.
26665 (ISA_ALL_SIMD): New macro
26666 (ISA_ALL_FP): New macro.
26667 * config/arm/arm.c (fpu_bitlist): Update initializer.
26668 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26670 (arm9e): Add fpu. Add option for nofp
26671 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26672 (arm926ej-s, arm1026ej-s): Likewise.
26673 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26674 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26675 neon-fp16, neon-vfpv4, nofp and nosimd.
26676 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26677 (cortex-a8): Add fpu. Add option for nofp.
26678 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26679 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26680 (cortex-r4f): Add fpu.
26681 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26682 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26684 (cortex-r8): Likewise.
26685 (cortex-m4): Add fpu. Add option for nofp.
26686 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26687 (cortex-a17.cortex-a7): Likewise.
26688 (cortex-a32): Add fpu. Add options for crypto and nofp.
26689 (cortex-a35, cortex-a53): Likewise.
26690 (cortex-a57): Add fpu. Add option for crypto.
26691 (cortex-a72, cortex-a73): Likewise.
26692 (exynos-m1): Likewise.
26693 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26694 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26695 (cortex-m33): Add fpu. Add option for nofp.
26696 * config/arm/arm-cpu-cdata.h: Regenerated
26697 * config/arm/arm-cpu-data.h: Regenerated.
26699 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26701 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26702 (armv5te, armv5tej): Likewise.
26703 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26704 (armv7): Add options fp and vfpv3-d16.
26705 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26706 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26708 (armv7ve): Likewise.
26709 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26710 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26711 (armv8-a): Add nocrypto option.
26712 (armv8.1-a, armv8.2-a): Likewise.
26713 (armv8-m.main): add options fp, fp.dp and nofp.
26715 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26717 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26719 (armv8-a+crc): Delete.
26720 (armv8.1-a): Add options simd, crypto and nofp.
26721 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26722 (armv8.2-a+fp16): Delete.
26723 (armv8-m.main): Add option dsp.
26724 (armv8-m.main+dsp): Delete.
26725 (cortex-a8): Add fpu. Add nofp option.
26726 (cortex-a9): Add fpu. Add nofp and nosimd options.
26727 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26728 link to main cpu and architecture data structures.
26729 (gen_comm_data): Only put isa attributes from the main architecture
26731 (option): New statement for architecture and CPU entries.
26732 * arm.c (struct cpu_option): New structure.
26733 (struct processors): Add entry for options.
26734 (arm_unrecognized_feature): New function.
26735 (arm_parse_arch_cpu_name): Ignore any characters after the first
26737 (arm_parse_arch_cpu_feature): New function.
26738 (arm_configure_build_target): Separate out any CPU and architecture
26739 features and parse separately. Don't error out if -mfpu=auto is
26740 used with only an architecture string.
26741 (arm_print_asm_arch_directives): New function.
26742 (arm_file_start): Call it.
26743 * config/arm/arm-cpu-cdata.h: Regenerated.
26744 * config/arm/arm-cpu-data.h: Likewise.
26745 * config/arm/arm-tables.opt: Likewise.
26747 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26749 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26750 assembler when it is not -mfpu=auto.
26752 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26754 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26755 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26756 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26757 (ASM_CPU_SPEC): Rewrite.
26758 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26759 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26760 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26761 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26762 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26763 copied string is NUL-terminated. Also strip any characters prefixed
26765 (arm_rewrite_selected_arch): New function.
26766 (arm_rewrite_march): New function.
26768 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26770 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26771 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26772 (march, mcpu, mtune): Convert to string-based options.
26773 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26774 (arm_parse_arch_cpu_name): New function.
26775 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26776 identify selected architecture or CPU.
26777 (arm_option_save): New function.
26778 (TARGET_OPTION_SAVE): Redefine.
26779 (arm_option_restore): Restore string options.
26780 (arm_option_print): Print string options.
26782 2017-06-16 Martin Sebor <msebor@redhat.com>
26784 PR tree-optimization/80933
26785 PR tree-optimization/80934
26786 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26787 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26788 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26789 (gimple_fold_builtin): Call them.
26791 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26793 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
26794 as unlikely; update profile.
26796 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26798 * predict.c (force_edge_cold): Handle declaring edges impossible
26801 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26803 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
26805 (try_unroll_loop_completely): Fix reporting.
26807 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26809 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
26811 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
26814 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
26815 if given a non-constant argument for an intrinsic which requires a
26818 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26820 * profile.c (compare_freqs): New function.
26821 (branch_prob): Sort edge list.
26822 (find_spanning_tree): Assume that the list is priority sorted.
26824 2017-06-16 Richard Biener <rguenther@suse.de>
26826 PR tree-optimization/81090
26827 * passes.def (pass_record_bounds): Remove.
26828 * tree-pass.h (make_pass_record_bounds): Likewise.
26829 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
26830 make_pass_record_bounds): Likewise.
26831 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
26832 not free niter estimates at the beginning but at the end.
26833 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
26835 2017-06-16 Richard Biener <rguenther@suse.de>
26837 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
26838 initializer to workaround ICE in host GCC 4.8.
26840 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26842 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
26844 (clone_inlined_nodes): Update.
26846 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26848 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
26849 prefetch settings, and enable prefetching by default at -O3.
26851 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26853 * config/aarch64/aarch64.c (aarch64_override_options_internal):
26854 Set flag_prefetch_loop_arrays according to tuning data.
26856 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26858 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
26859 New tune structure.
26860 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
26861 [Unrelated to main purpose of the patch] Place the pointer field last
26862 to enable type checking errors when tune structure are wrongly merged.
26863 * config/aarch64/aarch64.c (generic_prefetch_tune,)
26864 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
26865 (thunderx2t99_prefetch_tune): New tune constants.
26866 (tune_params *_tunings): Update all tunings (no functional change).
26867 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
26868 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
26869 from tunings structures.
26871 2017-06-16 Jakub Jelinek <jakub@redhat.com>
26874 * ubsan.c (instrument_null): Add T argument, use it instead
26875 of computing it based on IS_LHS.
26876 (instrument_object_size): Likewise.
26877 (pass_ubsan::execute): Adjust instrument_null and
26878 instrument_object_size callers to pass gimple_get_lhs or
26879 gimple_assign_rhs1 result to it. Use instrument_null instead of
26880 calling get_base_address and instrument_mem_ref. Handle
26881 aggregate call arguments for object-size sanitization.
26883 2017-06-16 Yury Gribov <tetra2005@gmail.com>
26885 PR tree-optimization/81089
26886 * tree-vrp.c (is_masked_range_test): Validate operands of
26889 2017-06-15 Martin Sebor <msebor@redhat.com>
26892 * dumpfile.c (dump_register): Avoid calling memset to initialize
26893 a class with a default ctor.
26894 * gcc.c (struct compiler): Remove const qualification.
26895 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
26896 * hash-table.h: Ditto.
26897 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
26899 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
26900 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
26902 * params.h (struct param_info): Make struct members non-const.
26903 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
26904 with default initialization.
26905 * vec.h (vec_copy_construct, vec_default_construct): New helper
26907 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
26908 with vec_copy_construct.
26909 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
26910 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
26911 * doc/invoke.texi (-Wclass-memaccess): Document.
26913 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26915 * emit-rtl.h (is_leaf): Update comment about local
26916 register allocator.
26918 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26921 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
26922 for a variable to have a section before checking if the section has a
26924 Set section to.persistent if persistent attribute is set.
26925 Warn if .persistent attribute is used on an automatic variable.
26927 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
26929 PR rtl-optimization/80474
26930 * reorg.c (update_block): Do not ignore instructions in a delay slot.
26932 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
26934 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
26937 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
26939 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
26940 (casesi): Emit bounds checking as RTL.
26941 (casesi_internal_mips16_<mode>): Remove bounds checking.
26943 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
26945 * config/xtensa/xtensa.c (xtensa_option_override): Append
26946 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
26947 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
26948 xtensa_doloop_hooks): Define unconditionally.
26949 (xtensa_reorg_loops): Only call reorg_loops in the presence of
26951 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
26952 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
26953 for it in xtensa_option_override.
26954 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
26955 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
26957 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
26959 * doc/cppopts.texi: Document '-' special value to -MF.
26961 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
26963 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
26964 (cortex_a53_fconst): Likewise.
26965 (cortex_a53_fpmul): Likewise.
26966 (cortex_a53_f_load_64): Likewise.
26967 (cortex_a53_f_load_many): Likewise.
26968 (cortex_a53_advsimd_alu): Likewise.
26969 (cortex_a53_advsimd_alu_q): Likewise.
26970 (cortex_a53_advsimd_mul): Likewise.
26971 (cortex_a53_advsimd_mul_q): Likewise.
26972 (fpmac bypass): Add new bypass for fpmac-fpmac case.
26973 Add missing fmul, r2f_cvt and fconst cases.
26975 2017-06-14 Richard Biener <rguenther@suse.de>
26977 PR middle-end/81088
26978 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
26980 (fold_binary_loc): When associating do not treat pre-existing
26981 TREE_OVERFLOW on literal constants as a reason to allow
26982 TREE_OVERFLOW on associated literal constants.
26984 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26986 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
26987 (MASK_FEATURES): New macro.
26988 * config/sparc/sparc.c (sparc_option_override): Remove the special
26989 handling of -mfpu and generalize it to all MASK_FEATURES switches.
26991 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26993 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
26994 a division of 0 if non-call exceptions are enabled.
26996 2017-06-14 Andrew Pinski <apinski@cavium.com>
26997 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27000 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27001 Improve vector initialization code gen for only variable case.
27003 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27005 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27007 2017-06-14 Richard Biener <rguenther@suse.de>
27009 PR tree-optimization/81083
27010 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27013 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27015 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27016 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27017 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27018 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27019 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27020 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27022 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27024 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27025 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27027 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27029 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27031 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27033 * config/rs6000/t-rtems: Don't handle SPE.
27035 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27037 * config/rs6000/t-linux: Don't handle SPE.
27039 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27041 * config/rs6000/eabispe.h: Delete file.
27043 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27045 * config/rs6000/t-spe: Delete file.
27047 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27049 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27050 (rs6000_legitimate_offset_address_p): Return false for anything in
27051 V2SImode or V2SFmode.
27053 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27055 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27056 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27057 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27059 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27060 (rs6000_legitimate_offset_address_p): Ditto.
27061 (rs6000_emit_move): Ditto.
27062 (rs6000_init_builtins): Remove V4HI_type_node.
27064 2017-06-13 Martin Liska <mliska@suse.cz>
27067 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27068 (gate_asan): Likewise.
27069 * asan.h (asan_no_sanitize_address_p): Remove the function.
27070 (sanitize_flags_p): New function.
27071 * builtins.def: Fix coding style.
27072 * common.opt: Use renamed enum value.
27073 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27074 * doc/extend.texi: Document no_sanitize attribute.
27075 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27076 to SANITIZE_UNDEFINED_NONDEFAULT.
27077 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27078 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27079 Use sanitize_flags_p.
27080 * gimplify.c (gimplify_function_tree): Likewise.
27081 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27082 * opts.c (parse_no_sanitize_attribute): New function.
27083 (common_handle_option): Use renamed enum value.
27084 * opts.h (parse_no_sanitize_attribute): Declare.
27085 * tree.c (sanitize_flags_p): New function.
27086 * tree.h: Declared here.
27087 * tsan.c: Use sanitize_flags_p.
27088 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27089 (instrument_mem_ref): Likewise.
27090 (instrument_bool_enum_load): Likewise.
27091 (do_ubsan_in_current_function): Remove the function.
27092 (pass_ubsan::execute): Use sanitize_flags_p.
27093 * ubsan.h: Remove do_ubsan_in_current_function
27094 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27095 (dump_function_to_file): Use it here.
27097 2017-06-13 Martin Jambor <mjambor@suse.cz>
27099 PR tree-optimization/80803
27100 PR tree-optimization/81063
27101 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27102 (propagate_subaccesses_across_link): Enqueue subtree whenever
27103 necessary instead of relying on the caller.
27105 2017-06-13 Martin Jambor <mjambor@suse.cz>
27107 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27108 that have a first_link.
27109 (sort_and_splice_var_accesses): Do not check first_link before
27111 (subtree_mark_written_and_enqueue): Likewise.
27112 (propagate_all_subaccesses): Likewise and do not stop at first
27113 parent with a first_link.
27115 2017-06-13 Martin Jambor <mjambor@suse.cz>
27117 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27120 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27122 * match.pd: New pattern.
27124 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27126 * tree-vrp.c (is_masked_range_test): New function.
27127 (register_edge_assert_for): Determine ranges for
27130 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27132 PR tree-optimization/67328
27133 * fold-const.c (maskable_range_p): New function.
27134 (build_range_check): Generate bittests if possible.
27136 2017-06-13 Martin Liska <mliska@suse.cz>
27138 * gimple-pretty-print.c (dump_probability): Add new argument.
27139 (dump_edge_probability): Dump both probability and count.
27140 (dump_gimple_label): Likewise.
27141 (dump_gimple_bb_header): Likewise.
27143 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27146 * config/avr/avr-devices.c: Fix indentation.
27147 * config/avr/gen-avr-mmcu-specs.c: Dito.
27149 2017-06-13 Richard Biener <rguenther@suse.de>
27151 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27152 instead get vector type from stmt_info.
27153 (vectorizable_reduction): Adjust. Remove dead code.
27155 2017-06-13 Richard Biener <rguenther@suse.de>
27157 PR middle-end/81065
27158 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27159 case of C * (x * C2 + C3).
27160 (fold_addr_of_array_ref_difference): Properly fold index difference.
27162 2017-06-12 David S. Miller <davem@davemloft.net>
27165 * config/sparc/sparc.md (return expander): Emit frame blockage if
27166 function uses alloca.
27168 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27170 * combine.c (make_field_assignment): Check len rather than the mode
27171 precision when calling force_to_mode.
27173 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27175 Support multilibs and devices that see flash in RAM address range.
27178 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27179 (avr_mcu_t) <flash_pm_offset>: New field.
27180 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27181 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27182 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27183 (AVR_TINY_PM_OFFSET): Remove macro.
27184 * config/avr/avr.opt (-mshort-calls): New option.
27185 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27186 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27187 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27188 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27189 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27190 instead of avr_arch->have_jmp_call.
27191 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27192 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27193 avr_arch->flash_pm_offset to define.
27194 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27195 new field flash_pm_offset. Add entry for avrxmega3.
27196 (avr_texinfo): Add entry for avrxmega3.
27197 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27198 attiny212, attiny214,
27199 attiny412, attiny414, attiny416, attiny417,
27200 attiny814, attiny816, attiny817,
27201 attiny1614, attiny1616, attiny1617,
27202 attiny3214, attiny3216, attiny3217.
27203 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27204 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27205 (avr_print_operand_address) [AVR_TINY]: Same.
27206 (avr_asm_init_sections) <readonly_data_section>: Only patch
27207 callback if avr_arch->flash_pm_offset = 0.
27208 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27209 for rodata if avr_arch->flash_pm_offset != 0.
27210 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27211 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27212 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27213 (m_options): Append opt_rcall.
27214 (m_dirnames): Append dir_rcall.
27215 * config/avr/t-multilib: Regenerate.
27217 * configure.ac [target=avr]: Check whether avrxmega3 default
27218 linker description file works as needed.
27219 * configure: Regenerate.
27220 * doc/avr-mmcu.texi: Regenerate.
27221 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27222 <__AVR_ARCH__>: Document avrxmega3 and 103.
27223 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27224 <__AVR_SHORT_CALLS__>: Document it.
27225 <__AVR_PM_BASE_ADDRESS__>: Document it.
27226 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27227 (AVR Variable Attributes) <progmem>: Document this is
27228 not needed for avrxmega3.
27229 (AVR Named Address Spaces) <__flash>: Dito.
27231 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27233 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27235 2017-06-12 Doug Rupp <rupp@adacore.com>
27237 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27238 Append vxworks-stdint.h to the tm_file list.
27239 * config/vxworks-stdint.h: New file.
27241 2017-06-12 Martin Liska <mliska@suse.cz>
27243 PR tree-optimization/81041
27244 * tree-profile.c (gimple_gen_ic_func_profiler):
27245 Create an extra BB in profile-generate
27246 (gimple_gen_time_profiler): Likewise.
27248 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27250 PR tree-optimization/81003
27251 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27252 (update_range_test): Use it instead of force_gimple_operand_gsi.
27254 2017-06-12 Richard Biener <rguenther@suse.de>
27256 PR tree-optimization/81053
27257 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27258 with backedge value not defined in loop. Simplify def stmt
27261 2017-06-11 Tom de Vries <tom@codesourcery.com>
27264 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27266 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27267 nvptx_cannot_force_const_mem.
27269 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27271 * opts.c (finish_options): Move test for flag_split_stack after
27272 it has been initialized.
27274 2017-06-11 Jason Merrill <jason@redhat.com>
27276 * tree.h (id_equal): New.
27277 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27278 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27279 instead of strcmp of IDENTIFIER_POINTER.
27281 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27283 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27284 (mark_all_inlined_calls_cdtor): Fix formating.
27285 (inline_transform): Rescale profile before inlining.
27287 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27289 * cgraph.h (cgraph_edge::clone): Update prototype.
27290 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27291 (cgraph_node::create_clone): Update.
27292 (cgraph_node::create_version_clone): Update.
27293 * tree-inline.c (copy_bb): Update.
27294 (expand_call_inline): Update.
27296 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27298 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27299 factored out from ...
27300 (rs6000_emit_prologue): ... here.
27302 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27304 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27305 factored out from ...
27306 (rs6000_emit_prologue): ... here.
27308 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27310 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27312 (handle_missing_profiles): Fix computation of tp_first_run.
27313 (counts_to_freqs): Do not touch freqs when count is 0.
27315 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27317 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27320 2017-06-10 Tom de Vries <tom@codesourcery.com>
27322 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27323 attributes): Document signal effective target.
27325 2017-06-10 Tom de Vries <tom@codesourcery.com>
27327 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27328 Document effective target stack_size.
27330 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27332 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27333 to the edit_context if they can be auto-applied.
27335 2017-06-9 Ian Lance Taylor <iant@golang.org>
27337 * opts.c (finish_options): If -fsplit-stack, disable implicit
27338 -forder-blocks-and-partition.
27339 * doc/invoke.texi (Optimize Options): Document that when using
27340 -fsplit-stack -forder-blocks-and-partition is not implicitly
27343 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27345 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27346 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27347 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27348 * builtins.def (abort, trap, unreachable): Declare cold.
27349 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27350 * tree-core.h (ECF_COLD): New.
27351 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27352 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27354 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27356 * predict.c (unlikely_executed_stmt_p): Cleanup.
27358 2017-06-09 Richard Biener <rguenther@suse.de>
27360 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27361 model if the ref is always written to.
27363 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27365 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27367 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27369 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27372 2017-06-09 Tom de Vries <tom@codesourcery.com>
27375 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27376 "target cannot support label values" when encountering LABEL_REF.
27378 2017-06-09 Martin Liska <mliska@suse.cz>
27380 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27381 (gimple_gen_ic_func_profiler): Emit direct comparison
27382 of __gcov_indirect_call_callee with NULL.
27383 (gimple_gen_time_profiler): Change probability from
27384 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27386 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27388 * profile.c (edge_gcov_counts): Turn to pointer.
27389 (compute_branch_probabilities, compute_branch_probabilities): Update.
27390 (branch_prob): Do not clear edge_gcov_count.
27391 * profile.h (edge_gcov_counts): Turn to pointer.
27392 (edge_gcov_count): Update.
27394 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27396 * gimple.h (gimple_check_failed): Mark cold.
27398 2017-06-09 Richard Biener <rguenther@suse.de>
27400 PR tree-optimization/66623
27401 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27402 refactor check_reduction into two parts, properly computing
27403 whether we have to check reduction validity for outer loop
27406 2017-06-09 Richard Biener <rguenther@suse.de>
27408 PR tree-optimization/79483
27409 * graphite-scop-detection.c (order): New global.
27410 (get_order): Compute bb to order mapping that satisfies code
27411 generation constraints.
27412 (cmp_pbbs): New helper.
27413 (build_scops): Start domwalk at entry block, sort generated
27416 2017-06-09 Richard Biener <rguenther@suse.de>
27418 PR middle-end/81007
27419 * ipa-polymorphic-call.c
27420 (ipa_polymorphic_call_context::restrict_to_inner_class):
27421 Skip FIELD_DECLs with error_mark_node type.
27422 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27425 2017-06-09 Martin Liska <mliska@suse.cz>
27427 * predict.c (struct branch_predictor): New struct.
27428 (test_prediction_value_range): New test.
27429 (predict_c_tests): New function.
27430 * selftest-run-tests.c (selftest::run_tests): Run the function.
27431 * selftest.h: Declare new tests.
27433 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27436 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27437 gen_add3_insn did not fail.
27438 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27439 r0, construct that number in a temporary reg and add that reg to r0.
27440 If asked to put the result in r0 as well, fail.
27442 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27444 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27445 for early expansion of vec_eqv.
27447 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27449 PR middle-end/81005
27450 * ubsan.c (instrument_null): Avoid pointless code temporary.
27451 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27453 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27457 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27460 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27463 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27465 * predict.c (unlikely_executed_edge_p): Move ahead.
27466 (probably_never_executed_edge_p): Use it.
27468 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27470 PR middle-end/79988
27471 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27472 gimple_call_builtin_p call.
27474 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27476 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27477 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27478 rtl_check_failed_type2, rtl_check_failed_code1,
27479 rtl_check_failed_code2, rtl_check_failed_code_mode,
27480 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27481 rtvec_check_failed_bounds, rtl_check_failed_flag,
27482 _fatal_insn_not_found, _fatal_insn): Likewise.
27483 * tree.h (tree_contains_struct_check_failed,
27484 tree_check_failed, tree_not_check_failed,
27485 tree_class_check_failed, tree_range_check_failed,
27486 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27487 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27488 tree_operand_check_failed, omp_clause_check_failed,
27489 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27492 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27494 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27495 flag_branch_probabilities.
27496 * ipa-inline.c (edge_badness): Likewise.
27497 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27498 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27499 * predict.c (maybe_hot_frequency_p): Likewise.
27500 (probably_never_executed): Likewise.
27501 * sched-ebb.c (schedule_ebbs): Likewise.
27502 * sched-rgn.c (find_single_block_region): Likewise.
27503 * tracer.c (tail_duplicate): Likewise.
27505 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27507 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27508 longer requires x_flag_profile_use.
27510 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27512 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27513 instead of flag_reorder_blocks_and_partition.
27514 * dbxout.c (dbxout_function_end): Likewise.
27515 * dwarf2out.c (gen_subprogram_die): Likewise.
27516 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27517 * hw-doloop.c (reorg_loops): Likewise.
27518 * varasm.c (assemble_start_function,
27519 assemble_end_function): Likewise.
27520 (decide_function_section): Do not check for
27521 flag_reorder_blocks_and_partition.
27523 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27525 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27527 (chkp_get_hard_register_fake_addr_expr): Ditto.
27528 (chkp_build_addr_expr): Add check for hard reg case.
27529 (chkp_parse_array_and_component_ref): Ditto.
27530 (chkp_find_bounds_1): Ditto.
27531 (chkp_process_stmt): Don't generate bounds store for
27534 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27536 * predict.c (maybe_hot_bb_p): Do not check profile status.
27537 (maybe_hot_edge_p): Likewise.
27538 (probably_never_executed): Check for zero counts even if profile
27540 (unlikely_executed_edge_p): New function.
27541 (unlikely_executed_stmt_p): New function.
27542 (unlikely_executed_bb_p): New function.
27543 (set_even_probabilities): Use unlikely predicates.
27544 (combine_predictions_for_bb): Likewise.
27545 (predict_paths_for_bb): Likewise.
27546 (predict_paths_leading_to_edge): Likewise.
27547 (determine_unlikely_bbs): New function.
27548 (estimate_bb_frequencies): Use it.
27549 (compute_function_frequency): Use zero counts even if profile is
27551 * profile-count.h: Fix typo.
27553 2017-08-08 Julia Koval <julia.koval@intel.com>
27555 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27556 _mm512_mask_cvtsepi16_storeu_epi8,
27557 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27558 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27559 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27560 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27561 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27562 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27563 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27564 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27565 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27566 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27567 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27568 __builtin_ia32_pmovuswb256mem_mask,
27569 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27570 __builtin_ia32_pmovwb512mem_mask): New builtins.
27572 2017-08-08 Julia Koval <julia.koval@intel.com>
27574 PR target/73350,80862
27575 * config/i386/subst.md (round): Fix round pattern.
27576 * config/i386/i386.c (ix86_erase_embedded_rounding):
27577 Fix erasing rounding for the fixed pattern.
27579 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27581 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27583 2017-06-08 Martin Liska <mliska@suse.cz>
27585 PR gcov-profile/80911
27586 * gcov.c (block_info::block_info): New constructor.
27588 2017-06-07 Carl Love <cel@us.ibm.com>
27590 * config/rs6000/rs6000-c: The return type of the following
27591 built-in functions was implemented as int not long long. Fix sign
27592 of return value for the unsigned version of vec_mulo and vec_mule.
27593 vector unsigned long long vec_bperm (vector unsigned long long,
27594 vector unsigned char)
27595 vector signed long long vec_mule (vector signed int,
27597 vector unsigned long long vec_mule (vector unsigned int,
27598 vector unsigned int)
27599 vector signed long long vec_mulo (vector signed int,
27601 vector unsigned long long vec_mulo (vector unsigned int,
27602 vector unsigned int)
27603 * doc/extend.texi: Fix the documentation for the built-in
27606 2017-06-07 Carl Love <cel@us.ibm.com>
27609 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27612 2017-06-07 Carl Love <cel@us.ibm.com>
27614 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27615 support, Generate doublehv for signed int/float for BE case only.
27617 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27619 * doc/invoke.texi (mcx16): Rewrite.
27621 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27623 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27624 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27625 *mov<mode>_softfloat, and an anonymous splitter): Use
27626 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27628 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27630 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27632 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27633 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27634 (rs6000_debug_reg_global): Adjust.
27635 (rs6000_init_hard_regno_mode_ok): Adjust.
27636 (rs6000_dbx_register_number): Adjust.
27637 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27638 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27639 Remove SPE_ACC and SPEFSCR.
27640 (REG_ALLOC_ORDER): Ditto.
27641 (FRAME_POINTER_REGNUM): Change to 111.
27642 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27643 (REG_CLASS_NAMES): Ditto.
27644 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27645 (REGISTER_NAMES): Ditto.
27646 (ADDITIONAL_REG_NAMES): Ditto.
27647 (rs6000_reg_names): Ditto.
27648 * config/rs6000/rs6000.md: Renumber some register number
27651 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27653 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27655 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27656 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27658 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27659 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27660 Delete the SPE high registers.
27661 (REG_ALLOC_ORDER): Ditto.
27662 (enum reg_class): Remove SPE_HIGH_REGS.
27663 (REG_CLASS_NAMES): Ditto.
27664 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27665 (REGISTER_NAMES): Ditto.
27666 (rs6000_reg_names): Ditto.
27667 * doc/tm.texi.in: Remove SPE as example.
27668 * doc/tm.texi: Regenerate.
27670 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27672 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27673 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27674 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27675 * config/rs6000/rs6000.md (type): Remove "brinc".
27677 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27679 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27680 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27681 * config/rs6000/linuxspe.h: Delete file.
27682 * config/rs6000/rs6000.md: Don't include spe.md.
27683 * config/rs6000/spe.h: Delete file.
27684 * config/rs6000/spe.md: Delete file.
27685 * config/rs6000/t-rs6000: Remove spe.md.
27687 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27689 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27690 (reg_or_none500mem_operand): Delete.
27691 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27692 instead of reg_or_none500mem_operand.
27694 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27696 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27697 handling of SPE flags.
27698 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27700 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27702 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27704 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27705 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27706 paired_divv2sf3): Similar.
27707 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27708 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27709 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27711 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27712 Rename the paired_* instruction patterns.
27713 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27715 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27716 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27717 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27718 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27719 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27720 PAIRED_VECTOR_MODE.
27721 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27722 (spe_func_has_64bit_regs_p): Delete.
27723 (spe_expand_predicate_builtin): Delete.
27724 (spe_expand_evsel_builtin): Delete.
27725 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27726 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27727 (invalid_e500_subreg): Delete.
27728 (rs6000_legitimize_address): Always force_reg op2 as well, for
27729 paired single memory accesses.
27730 (rs6000_member_type_forces_blk): Delete.
27731 (rs6000_spe_function_arg): Delete.
27732 (rs6000_expand_unop_builtin): Delete SPE handling.
27733 (rs6000_expand_binop_builtin): Ditto.
27734 (spe_expand_stv_builtin): Delete.
27735 (bdesc_2arg_spe): Delete.
27736 (spe_expand_builtin): Delete.
27737 (spe_expand_predicate_builtin): Delete.
27738 (spe_expand_evsel_builtin): Delete.
27739 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27740 (spe_init_builtins): Delete.
27741 (spe_func_has_64bit_regs_p): Delete.
27742 (savres_routine_name): Delete "info" parameter. Adjust callers.
27743 (rs6000_emit_stack_reset): Ditto.
27744 (rs6000_dwarf_register_span): Delete.
27745 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27746 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27747 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27749 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27751 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27752 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27753 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27754 mulv2sf3, divv2sf3): Delete expanders.
27756 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27758 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27760 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27762 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27763 * config/rs6000/rs6000.c: Ditto.
27765 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27767 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27768 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27769 comparison_operator.
27771 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27773 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27774 * config/rs6000/rs6000.opt: Ditto.
27775 * config/rs6000/t-rtems: Ditto.
27777 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27779 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27780 TARGET_E500_SINGLE by 0, simplify.
27781 * config/rs6000/rs6000.c: Ditto.
27782 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27783 (spe_build_register_parallel): Delete.
27784 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27785 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27786 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27787 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27788 (E500_CONVERT): Delete.
27789 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27791 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27793 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
27794 * config/rs6000/dfp.md: Ditto.
27795 (negdd2, *negdd2_fpr): Merge.
27796 (absdd2, *absdd2_fpr): Merge.
27797 (negtd2, *negtd2_fpr): Merge.
27798 (abstd2, *abstd2_fpr): Merge.
27799 * config/rs6000/e500.h: Delete file.
27800 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
27801 TARGET_FPRS by 1 and simplify.
27802 * config/rs6000/rs6000-c.c: Ditto.
27803 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
27804 TARGET_DF_SPE by 0.
27805 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
27807 * config/rs6000/rs6000.md: Ditto.
27808 (floatdidf2, *floatdidf2_fpr): Merge.
27809 (move_from_CR_gt_bit): Delete.
27810 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
27811 (E500_CR_IOR_COMPARE): Delete.
27812 (All patterns that require !TARGET_FPRS): Delete.
27813 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
27815 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27817 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
27819 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27821 * graphds.c (add_edge): Intitialize edge's attached data.
27822 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
27823 pointer parameter. Call pointed function on each edge during
27824 graph traversing. Skip traversing the edge when the function
27826 (graphds_dfs, graphds_scc): Ditto.
27827 (for_each_edge): New parameter. Pass the new parameter to callback
27829 * graphds.h (skip_edge_callback): New function pointer type.
27830 (graphds_dfs, graphds_scc): New function pointer parameter.
27831 (graphds_edge_callback, for_each_edge): New parameter.
27833 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27835 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
27836 out code checking if runtime alias check is possible to below ...
27837 Call the new function.
27838 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
27839 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
27841 2017-06-07 Marek Polacek <polacek@redhat.com>
27844 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
27845 TYPE_OVERFLOW_WRAPS check.
27847 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27849 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
27850 if versioning is required.
27851 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
27852 peeling with the check for versioning.
27854 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27856 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
27857 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
27858 Set true to new parameter if new ssa variable is defined.
27859 (vect_gen_vector_loop_niters): Refactor. Set range information
27860 for the new vector loop bound variable.
27861 (vect_do_peeling): Ditto.
27863 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27865 * tree-affine.c (ssa.h): Include header file.
27866 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
27867 has wrapping overflow behavior.
27869 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27871 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
27873 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27875 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
27876 (tree_to_aff_combination): ... here.
27878 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27880 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
27881 reg_pressure model function.
27882 (ivopts_global_cost_for_size): Delete.
27883 (determine_set_costs, iv_ca_recount_cost): Call new model function
27884 ivopts_estimate_reg_pressure.
27886 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27888 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
27889 expensive than udiv. Remove floating point cases from mod.
27891 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27893 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
27894 Increase idiv cost.
27896 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27898 * config/aarch64/aarch64.md
27899 (copysignsf3): Fix mask generation.
27901 2017-06-07 Jakub Jelinek <jakub@redhat.com>
27903 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
27905 (class dump_manager): Add register_dumps method.
27906 * dumpfile.c: Include langhooks.h.
27907 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
27908 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
27909 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
27910 (dump_manager::dump_register): Start with 512 entries instead of 32.
27911 (dump_manager::register_dumps): New method.
27912 * toplev.c (general_init): Instead of invoking register_dumps
27913 langhook, invoke register_dumps method on the dump manager.
27914 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
27917 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
27919 * doc/md.texi: Clarify the restrictions on a define_insn condition.
27920 Say that # requires an associated define_split to exist, and that
27921 the define_split must be suitable for use after register allocation.
27923 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27925 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
27926 (compute_outgoing_frequencies): Also initialize zero counts.
27927 (find_many_sub_basic_blocks): Do not produce uninitialized profile
27928 around loops; preserve more of profile when nothing changes.
27930 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
27932 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
27934 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
27935 * config/arm/arm-cpu-cdata.h: Regenerate.
27936 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
27937 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
27938 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
27939 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
27940 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
27942 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
27944 * config/arm/t-rmprofile: Likewise.
27945 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
27947 2017-06-06 David S. Miller <davem@davemloft.net>
27950 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
27951 blockage if function uses alloca.
27953 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27955 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
27956 New "uid" fields to hold pretty-print IDs of group and ref.
27957 Memory references are now identified as <group_id>:<ref_id>
27958 instead of using [random] addresses.
27959 (dump_mem_details): Simplify, no functional change.
27960 (dump_mem_ref): Simplify and make output more concise.
27961 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
27962 (find_or_create_group): Initialize group uid.
27963 (record_ref): Initialize ref uid. Improve debug output.
27964 (prune_group_by_reuse, should_issue_prefetch_p,)
27965 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
27966 (mark_nontemporal_store, determine_loop_nest_reuse):
27967 Improve debug output.
27969 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27971 * dbgcnt.def (prefetch): New debug counter.
27972 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
27973 (schedule_prefetches): Stop issueing prefetches if debug counter
27976 2017-06-06 Tom de Vries <tom@codesourcery.com>
27978 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
27979 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
27981 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27983 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
27984 Use aarch64_reg_or_zero predicate for operand 4.
27985 (aarch64_compare_and_swap<mode> define_insn_and_split):
27986 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
27987 (aarch64_store_exclusive<mode>): Likewise for operand 2.
27989 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
27991 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
27992 (arm_compute_save_core_reg_mask): This.
27993 (thumb1_compute_save_reg_mask): Rename into ...
27994 (thumb1_compute_save_core_reg_mask): This.
27995 (arm_compute_save_reg0_reg12_mask): Adapt comment.
27996 (arm_compute_frame_layout): Likewise.
27998 2017-06-06 Richard Biener <rguenther@suse.de>
28000 PR tree-optimization/80974
28001 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28002 keep or clear leaders SSA info.
28004 2017-06-06 Tom de Vries <tom@codesourcery.com>
28006 * config/nvptx/nvptx.c (split_mode_p): New function.
28007 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28009 2017-06-06 Tom de Vries <tom@codesourcery.com>
28011 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28013 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28016 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28019 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28021 * shrink-wrap.c (handle_simple_exit): Update profile.
28022 (try_shrink_wrapping): Upate profile.
28024 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28026 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28027 (tree_guess_outgoing_edge_probabilities): New.
28028 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28029 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28031 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28033 * ipa-split.c (split_function): Initialize return bb profile.
28035 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28037 * profile.c (compute_branch_probabilities): Also initialize
28038 EXIT_BLOCK profile.
28040 2017-06-06 Richard Biener <rguenther@suse.de>
28042 PR tree-optimization/80928
28043 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28044 (vect_analyze_loop_operations): Properly guard analysis for
28046 (vect_transform_loop): Likewise.
28047 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28048 (vect_model_induction_cost): Do not cost for pure SLP.
28049 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28050 of induction in inner loop vectorization.
28051 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28052 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28053 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28055 (vect_analyze_slp_cost_1): Cost induction.
28056 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28057 (vect_get_slp_vect_defs): Likewise.
28058 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28059 (vect_transform_stmt): Handle SLP reductions.
28060 * tree-vectorizer.h (vectorizable_induction): Adjust.
28062 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28064 * config/rs6000/rs6000.c (make_resolver_func): Update
28065 init_lowered_empty_function call.
28067 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28069 * doc/invoke.texi: Document the -fprofile-abs-path option.
28070 * common.opt (fprofile-abs-path): New option.
28071 * gcov-io.h (gcov_write_filename): Declare.
28072 * gcov-io.c (gcov_write_filename): New function.
28073 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28074 * profile.c (output_location): Likewise.
28076 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28078 * shring-wrap.c: Revert accidental commit.
28080 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28082 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28084 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28086 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28088 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28089 profile in callgraph edge.
28090 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28091 (apply_scale): Likewise.
28092 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28093 Also scale profile when inlining function with zero profile.
28094 (initialize_cfun): Update exit block profile even when it is zero.
28095 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28096 when profile is read.
28098 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28100 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28101 (CLONE_*): New constants to define the processors we can generate
28102 code for with the target_clone attribute.
28103 (rs6000_clone_map): New array to identify which clone processors
28104 the current program is running on.
28105 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28106 target_clone attribute.
28107 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28108 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28109 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28110 (cpu_expand_builtin): Add support for target_clone attribute.
28111 (rs6000_valid_attribute_p): Allow "default" attribute.
28112 (get_decl_name): New debug function to simplify printing the
28113 current function name in debugging statements.
28114 (rs6000_clone_priority): New functions to support the target_clone
28115 attribute, and be able to generate code to switch between ISA 2.05
28116 through ISA 3.0 (power6 through power9).
28117 (rs6000_compare_version_priority): Likewise.
28118 (rs6000_get_function_versions_dispatcher): Likewise.
28119 (make_resolver_func): Likewise.
28120 (add_condition_to_bb): Likewise.
28121 (dispatch_function_versions): Likewise.
28122 (rs6000_generate_version_dispatcher_body): Likewise.
28123 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28124 (fusion_gpr_load_p): Fix a spacing issue.
28125 * doc/extend.texi (Common Function Attributes): Document that the
28126 PowerPC supports the target_clone attribute.
28128 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28130 * config/arm/arm.h: explain F symbol found in description of ARM
28131 register allocation in its legend.
28133 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28135 * config/mips/frame-header-opt.c: Include profile-count.h.
28136 * config/riscv/riscv.c: Include profile-count.h
28138 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28140 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28142 (sm_set_flag_if_changed): Add bbs field.
28143 (execute_sm_if_changed_flag_set): Pass BBS.
28144 (execute_sm): Update.
28146 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28148 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28151 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28153 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28154 (peephole2): New peephole2 to emit the above.
28155 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28157 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28159 * config/aarch64/aarch64.c (define_peephole2 above
28160 *sub_<shift>_<mode>): New peephole.
28162 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28164 * config/i386/i386.c (make_resolver_func): Update.
28165 * Makefile.in: Add profile-count.h and profile-count.o
28166 * auto-profile.c (afdo_indirect_call): Update to new API.
28167 (afdo_set_bb_count): Update.
28168 (afdo_propagate_edge): Update.
28169 (afdo_propagate_circuit): Update.
28170 (afdo_calculate_branch_prob): Update.
28171 (afdo_annotate_cfg): Update.
28172 * basic-block.h: Include profile-count.h
28173 (struct edge_def): Turn count to profile_count.
28174 (struct basic_block_def): Likewie.
28175 (REG_BR_PROB_BASE): Move to profile-count.h
28176 (RDIV): Move to profile-count.h
28177 * bb-reorder.c (max_entry_count): Turn to profile_count.
28178 (find_traces): Update.
28179 (rotate_loop):Update.
28180 (connect_traces):Update.
28181 (sanitize_hot_paths):Update.
28182 * bt-load.c (migrate_btr_defs): Update.
28183 * cfg.c (RDIV): Remove.
28184 (init_flow): Use alloc_block.
28185 (alloc_block): Uninitialize count.
28186 (unchecked_make_edge): Uninitialize count.
28187 (check_bb_profile): Update.
28188 (dump_edge_info): Update.
28189 (dump_bb_info): Update.
28190 (update_bb_profile_for_threading): Update.
28191 (scale_bbs_frequencies_int): Update.
28192 (scale_bbs_frequencies_gcov_type): Update.
28193 (scale_bbs_frequencies_profile_count): New.
28194 * cfg.h (update_bb_profile_for_threading): Update.
28195 (scale_bbs_frequencies_profile_count): Declare.
28196 * cfgbuild.c (compute_outgoing_frequencies): Update.
28197 (find_many_sub_basic_blocks): Update.
28198 * cfgcleanup.c (try_forward_edges): Update.
28199 (try_crossjump_to_edge): Update.
28200 * cfgexpand.c (expand_gimple_tailcall): Update.
28201 (construct_exit_block): Update.
28202 * cfghooks.c (verify_flow_info): Update.
28203 (dump_bb_for_graph): Update.
28204 (split_edge): Update.
28205 (make_forwarder_block): Update.
28206 (duplicate_block): Update.
28207 (account_profile_record): Update.
28208 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28209 (get_estimated_loop_iterations): Update.
28210 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28211 (single_likely_exit): Update.
28212 * cfgloopmanip.c (scale_loop_profile): Update.
28214 (set_zero_probability): Update.
28215 (lv_adjust_loop_entry_edge): Update.
28216 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28217 (purge_dead_edges): Update.
28218 (rtl_account_profile_record): Update.
28219 * cgraph.c (cgraph_node::create): Uninitialize count.
28220 (symbol_table::create_edge): Uninitialize count.
28221 (cgraph_update_edges_for_call_stmt_node): Update.
28222 (cgraph_edge::dump_edge_flags): Update.
28223 (cgraph_node::dump): Update.
28224 (cgraph_edge::maybe_hot_p): Update.
28225 * cgraph.h: Include profile-count.h
28226 (create_clone), create_edge, create_indirect_edge): Update.
28227 (cgraph_node): Turn count to profile_count.
28228 (cgraph_edge0: Likewise.
28229 (make_speculative, clone): Update.
28230 (create_edge): Update.
28231 (init_lowered_empty_function): Update.
28232 * cgraphclones.c (cgraph_edge::clone): Update.
28233 (duplicate_thunk_for_node): Update.
28234 (cgraph_node::create_clone): Update.
28235 * cgraphunit.c (cgraph_node::analyze): Update.
28236 (cgraph_node::expand_thunk): Update.
28237 * final.c (dump_basic_block_info): Update.
28238 * gimple-streamer-in.c (input_bb): Update.
28239 * gimple-streamer-out.c (output_bb): Update.
28240 * graphite.c (print_global_statistics): Update.
28241 (print_graphite_scop_statistics): Update.
28242 * hsa-brig.c: Include basic-block.h.
28243 * hsa-dump.c: Include basic-block.h.
28244 * hsa-gen.c (T sum_slice): Update.
28245 (convert_switch_statements):Update.
28246 * hsa-regalloc.c: Include basic-block.h.
28247 * ipa-chkp.c (chkp_produce_thunks): Update.
28248 * ipa-cp.c (struct caller_statistics): Update.
28249 (init_caller_stats): Update.
28250 (gather_caller_stats): Update.
28251 (ipcp_cloning_candidate_p): Update.
28252 (good_cloning_opportunity_p): Update.
28253 (get_info_about_necessary_edges): Update.
28254 (dump_profile_updates): Update.
28255 (update_profiling_info): Update.
28256 (update_specialized_profile): Update.
28257 (perhaps_add_new_callers): Update.
28258 (decide_about_value): Update.
28259 (ipa_cp_c_finalize): Update.
28260 * ipa-devirt.c (struct odr_type_warn_count): Update.
28261 (struct decl_warn_count): Update.
28262 (struct final_warning_record): Update.
28263 (possible_polymorphic_call_targets): Update.
28264 (ipa_devirt): Update.
28265 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28266 * ipa-icf.c (sem_function::merge): Update.
28267 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28268 * ipa-inline.c (compute_uninlined_call_time): Update.
28269 (compute_inlined_call_time): Update.
28270 (want_inline_small_function_p): Update.
28271 (want_inline_self_recursive_call_p): Update.
28272 (edge_badness): Update.
28273 (lookup_recursive_calls): Update.
28274 (recursive_inlining): Update.
28275 (inline_small_functions): Update.
28276 (dump_overall_stats): Update.
28277 (dump_inline_stats): Update.
28278 * ipa-profile.c (ipa_profile_generate_summary): Update.
28279 (ipa_propagate_frequency): Update.
28280 (ipa_profile): Update.
28281 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28282 * ipa-utils.c (ipa_merge_profiles): Update.
28283 * loop-doloop.c (doloop_modify): Update.
28284 * loop-unroll.c (report_unroll): Update.
28285 (unroll_loop_runtime_iterations): Update.
28286 * lto-cgraph.c (lto_output_edge): Update.
28287 (lto_output_node): Update.
28288 (input_node): Update.
28289 (input_edge): Update.
28290 (merge_profile_summaries): Update.
28291 * lto-streamer-in.c (input_cfg): Update.
28292 * lto-streamer-out.c (output_cfg): Update.
28293 * mcf.c (create_fixup_graph): Update.
28294 (adjust_cfg_counts): Update.
28295 (sum_edge_counts): Update.
28296 * modulo-sched.c (sms_schedule): Update.
28297 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28298 * predict.c (maybe_hot_count_p): Update.
28299 (probably_never_executed): Update.
28300 (dump_prediction): Update.
28301 (combine_predictions_for_bb): Update.
28302 (propagate_freq): Update.
28303 (handle_missing_profiles): Update.
28304 (counts_to_freqs): Update.
28305 (rebuild_frequencies): Update.
28306 (force_edge_cold): Update.
28307 * predict.h: Include profile-count.h
28308 (maybe_hot_count_p, counts_to_freqs): UPdate.
28309 * print-rtl-function.c: Do not include cfg.h
28310 * print-rtl.c: Include basic-block.h
28311 * profile-count.c: New file.
28312 * profile-count.h: New file.
28313 * profile.c (is_edge_inconsistent): Update.
28314 (correct_negative_edge_counts): Update.
28315 (is_inconsistent): Update.
28316 (set_bb_counts): Update.
28317 (read_profile_edge_counts): Update.
28318 (compute_frequency_overlap): Update.
28319 (compute_branch_probabilities): Update; Initialize and deinitialize
28321 (branch_prob): Update.
28322 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28323 (edge_gcov_count): New.
28324 (bb_gcov_count): New.
28325 * shrink-wrap.c (try_shrink_wrapping): Update.
28326 * tracer.c (better_p): Update.
28327 * trans-mem.c (expand_transaction): Update.
28328 (ipa_tm_insert_irr_call): Update.
28329 (ipa_tm_insert_gettmclone_call): Update.
28330 * tree-call-cdce.c: Update.
28331 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28332 (gimple_duplicate_sese_tail): Update.
28333 (gimple_account_profile_record): Update.
28334 (execute_fixup_cfg): Update.
28335 * tree-inline.c (copy_bb): Update.
28336 (copy_edges_for_bb): Update.
28337 (initialize_cfun): Update.
28338 (freqs_to_counts): Update.
28339 (copy_cfg_body): Update.
28340 (expand_call_inline): Update.
28341 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28342 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28343 (try_unroll_loop_completely): Update.
28344 (try_peel_loop): Update.
28345 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28346 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28347 * tree-ssa-loop-split.c (connect_loops): Update.
28348 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28349 * tree-ssa-reassoc.c (branch_fixup): Update.
28350 * tree-ssa-tail-merge.c (replace_block_by): Update.
28351 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28352 (compute_path_counts): Update.
28353 (update_profile): Update.
28354 (recompute_probabilities): Update.
28355 (update_joiner_offpath_counts): Update.
28356 (estimated_freqs_path): Update.
28357 (freqs_to_counts_path): Update.
28358 (clear_counts_path): Update.
28359 (ssa_fix_duplicate_block_edges): Update.
28360 (duplicate_thread_path): Update.
28361 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28362 (struct switch_conv_info): Update.
28363 * tree-tailcall.c (decrease_profile): Update.
28364 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28365 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28366 * value-prof.c (check_counter): Update.
28367 (gimple_divmod_fixed_value): Update.
28368 (gimple_mod_pow2): Update.
28369 (gimple_mod_subtract): Update.
28370 (gimple_ic_transform): Update.
28371 (gimple_stringop_fixed_value): Update.
28372 * value-prof.h (gimple_ic): Update.
28374 2017-06-02 Carl Love <cel@us.ibm.com>
28376 * config/rs6000/rs6000-c: Add support for built-in functions
28377 vector double vec_doublee (vector signed int);
28378 vector double vec_doublee (vector unsigned int);
28379 vector double vec_doublee (vector float);
28380 vector double vec_doubleh (vector signed int);
28381 vector double vec_doubleh (vector unsigned int);
28382 vector double vec_doubleh (vector float);
28383 vector double vec_doublel (vector signed int);
28384 vector double vec_doublel (vector unsigned int);
28385 vector double vec_doublel (vector float);
28386 vector double vec_doubleo (vector signed int);
28387 vector double vec_doubleo (vector unsigned int);
28388 vector double vec_doubleo (vector float);.
28389 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28390 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28392 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28393 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28394 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28396 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28397 vec_doublel, vec_doubleh.
28398 * doc/extend.texi: Update the built-in documentation file for the
28399 new built-in functions.
28401 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28404 * ipa-inline-analysis.c (free_growth_caches): Set
28405 edge_removal_hook_holder to NULL after removing it.
28407 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28409 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28410 comparision with zero.
28412 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28413 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28414 for early expansion of vec_min and vec_max builtins.
28415 (builtin_function_type): Add min/max unsigned variants to those
28416 identified as having unsigned arguments.
28418 2017-06-02 Olivier Hainque <hainque@adacore.com>
28420 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28422 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28424 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28425 Use VALL_F16 iterator rather than VALL.
28427 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28429 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28430 Emit CBNZ inside loop when doing a strong exchange and comparing
28431 against zero. Generate the CC flags after the loop.
28433 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28435 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28436 (dl_section_ref): New.
28437 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28438 On AIX, append an expression to subtract the size of the
28439 section length to dl_section_ref.
28441 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28443 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28444 for early expansion of vector absolute builtins.
28446 2017-06-02 Richard Biener <rguenther@suse.de>
28448 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28449 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28451 2017-06-02 Richard Biener <rguenther@suse.de>
28453 PR tree-optimization/80948
28454 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28455 stmt order as well.
28457 2017-06-02 Richard Biener <rguenther@suse.de>
28459 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28461 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28462 for inductions as relevant.
28464 2017-06-02 Richard Biener <rguenther@suse.de>
28466 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28467 (vectorizable_induction): ... this. Remove dead code.
28469 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28471 * builtins. (expand_builtin_alloca): Remove second parameter and
28472 infer its value from the first parameter instead.
28473 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28475 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28477 PR rtl-optimization/80903
28478 * loop-doloop.c (add_test): Unshare sequence.
28480 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28482 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28484 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28486 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28488 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28489 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28490 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28491 (xlogue_layout::get_stub_rtx): Make static.
28492 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28493 (xlogue_layout::compute_stub_managed_regs): Rename to...
28494 (xlogue_layout::count_stub_managed_regs): ...this.
28495 (xlogue_layout::is_stub_managed_reg): New function.
28496 (xlogue_layout::m_stub_names): Rename to...
28497 (xlogue_layout::s_stub_names): ...this, make static.
28498 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28499 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28500 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28501 xlogue_layout::s_stub_names): Instantiate statics.
28502 (stub_managed_regs): Remove.
28503 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28504 (disable_call_ms2sysv_xlogues): Rename to...
28505 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28506 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28508 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28509 change after reload_completed.
28510 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28512 (ix86_expand_prologue): Likewise.
28513 (ix86_expand_epilogue): Likewise.
28514 (ix86_expand_split_stack_prologue): Likewise.
28515 (ix86_compute_frame_layout): Remove frame parameter ...
28516 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28517 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28519 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28520 (ix86_frame): Move from here ...
28521 * config/i386/i386.h (ix86_frame): ... to here.
28522 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28523 complete ix86_frame data structure instead. Remove some_ld_name.
28525 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28527 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28528 symbols that hold a DECL_VALUE_EXPR.
28530 2017-06-01 Martin Jambor <mjambor@suse.cz>
28532 PR tree-optimization/80898
28533 * tree-sra.c (process_subtree_disqualification): Removed.
28534 (disqualify_candidate): Do not acll
28535 process_subtree_disqualification.
28536 (subtree_mark_written_and_enqueue): New function.
28537 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28538 RHS has been disqualified and re-queue LHS if necessary. Apart
28539 from that, ignore disqualified RHS.
28541 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28543 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28544 address fetch for z10 or later.
28546 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28548 * config/arc/arc.md (tst_movb): Add guard when splitting.
28550 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28552 * config/arc/arc.c (arc_can_eliminate): Test against
28553 arc_frame_pointer_needed.
28555 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28557 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28558 to prevent store reordering.
28559 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28560 (type): Add block type.
28561 (stack_tie): Define special instruction to be used in
28564 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28566 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28567 constraint. It is not valid for the pattern.
28568 (noncommutative_binary_comparison): Likewise.
28570 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28572 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28575 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28577 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28578 be used by the reg-alloc.
28580 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28582 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28583 reg-alloc when having mul64 or mul32x16 instructions.
28584 (mulsidi3): Likewise.
28585 (umulsidi3): Likewise.
28586 (mulsi32x16): New pattern.
28587 (mulsi64): Likewise.
28588 (mulsidi64): Likewise.
28589 (umulsidi64): Likewise.
28590 (MUL32x16_REG): Define.
28591 (mul64_600): Use MUL32x16_REG.
28592 (mac64_600): Likewise.
28593 (umul64_600): Likewise.
28594 (umac64_600): Likewise.
28596 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28598 * config/arc/arc.md (mulsi3_700): Make it commutative.
28600 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28602 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28604 (*sign_extendsidi2_insn): Likewise for movstosw.
28606 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28608 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28609 the type of the input discriminant value. Convert the
28610 discriminant value of signedness vary.
28612 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28614 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28615 Add to -Wall section.
28617 2017-06-01 Richard Biener <rguenther@suse.de>
28619 PR middle-end/66313
28620 * fold-const.c (fold_plusminus_mult_expr): If the factored
28621 factor may be zero use a wrapping type for the inner operation.
28622 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28623 and handle moved defs.
28624 (process_assignment): Properly guard the unary op case. Return a
28625 tri-state indicating that moving the stmt before the call may allow
28626 to continue. Pass through to_move.
28627 (find_tail_calls): Handle moving unrelated defs before
28630 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28633 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28634 splitter result in the canonical way.
28636 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28638 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28639 also for 32bit target. Update insn attributes.
28640 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28642 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28644 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28645 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28646 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28647 (_mm_maskz_min_ss): New intrinsics.
28649 2017-05-31 Martin Liska <mliska@suse.cz>
28651 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28652 Change comment style to one we normally use.
28653 (vectorizable_reduction): Likewise.
28654 (vectorizable_induction): Likewise.
28655 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28656 (vectorizable_call): Likewise.
28657 (vectorizable_simd_clone_call): Likewise.
28658 (vectorizable_conversion): Likewise.
28659 (vectorizable_assignment): Likewise.
28660 (vectorizable_shift): Likewise.
28661 (vectorizable_operation): Likewise.
28662 (vectorizable_store): Likewise.
28663 (vectorizable_load): Likewise.
28664 * tree-vectorizer.h: Likewise.
28666 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28668 * passes.c (emergency_dump_function): New.
28669 * tree-pass.h (emergency_dump_function): Declare.
28670 * plugin.c (plugins_internal_error_function): Remove.
28671 * plugin.h (plugins_internal_error_function): Remove declaration.
28672 * toplev.c (internal_error_function): New static function. Use it...
28673 (general_init): ...here.
28675 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28677 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28678 (arc_rtx_costs): Add costs for new patterns.
28679 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28680 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28682 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28684 * tree-ssa-strlen.c (get_next_strinfo): New function.
28685 (get_stridx_plus_constant): Use it.
28686 (zero_length_string): Likewise.
28687 (adjust_related_strinfos): Likewise.
28688 (adjust_last_stmt): Likewise.
28690 2017-05-31 Richard Biener <rguenther@suse.de>
28693 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28694 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28696 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28698 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28699 loop_vinfo argument and use of dependence distance vectors.
28700 Check instead whether the two references differ only in their
28701 initial value and assume that they have the same alignment if the
28702 difference is a multiple of the vector alignment.
28703 (vect_analyze_data_refs_alignment): Update call accordingly.
28705 2017-05-31 Martin Liska <mliska@suse.cz>
28708 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28710 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28712 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28713 (create_intersect_range_checks): Move from ...
28714 * tree-data-ref.c (create_intersect_range_checks_index)
28715 (create_intersect_range_checks): ... to here.
28716 (create_runtime_alias_checks): New function factored from ...
28717 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28718 here. Call above function.
28719 * tree-data-ref.h (create_runtime_alias_checks): New function.
28721 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28723 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28724 segment length for dr_b and compute it in wide_int.
28726 2017-05-31 Richard Biener <rguenther@suse.de>
28728 PR tree-optimization/80906
28729 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28730 and pass through iv_map.
28731 (copy_bb_and_scalar_dependences): Adjust.
28732 (translate_pending_phi_nodes): Likewise.
28733 (copy_loop_close_phi_args): Handle code-generating IVs instead
28736 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28738 * diagnostic-color.c (color_dict): Add "type-diff".
28739 (parse_gcc_colors): Update comment.
28740 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28741 -fdiagnostics-show-template-tree and -fno-elide-type.
28742 (GCC_COLORS): Add type-diff to example.
28744 (-fdiagnostics-show-template-tree): New.
28745 (-fno-elide-type): New.
28746 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28747 the pp_format_decoder callback. Call any m_format_postprocessor's
28749 (pretty_printer::pretty_printer): Initialize
28750 m_format_postprocessor.
28751 (pretty_printer::~pretty_printer): Delete any
28752 m_format_postprocessor.
28753 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28754 (class format_postprocessor): New class.
28755 (struct pretty_printer::format_decoder): Document the new parameters.
28756 (struct pretty_printer::m_format_postprocessor): New field.
28757 * tree-diagnostic.c (default_tree_printer): Update for new
28758 bool and const char ** params.
28759 * tree-diagnostic.h (default_tree_printer): Likewise.
28761 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28763 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28764 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28765 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28766 rs6000_gen_cell_microcode code.
28767 (rs6000_final_prescan_insn): Delete.
28768 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28769 "warn-cell-microcode" entries.
28770 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28771 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28772 throughout. Change cc_reg_not_micro_cr0_operand to
28773 cc_reg_not_cr0_operand throughout.
28774 (*extendhi<mode>2_noload): Delete.
28775 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28776 (mwarn-cell-microcode): Delete.
28777 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28778 -mgen-cell-microcode and -mwarn-cell-microcode.
28780 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28783 * config/i386/constraints.md (Yd): New constraint.
28785 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28786 and (?Yd, r) alternatives. Update insn attributes.
28787 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28788 and (?*Yd, r) alternatives. Update insn attributes.
28789 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28791 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28793 * gimplify.c (gimplify_modify_expr): Don't create a
28794 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
28797 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
28799 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
28801 2017-05-30 Richard Biener <rguenther@suse.de>
28803 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
28804 and reduc_def fields.
28805 (STMT_VINFO_REDUC_TYPE): New define.
28806 (STMT_VINFO_REDUC_DEF): Likewise.
28807 (vect_force_simple_reduction): Adjust prototype.
28808 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
28809 (vect_is_simple_reduction): Remove check_reduction argument.
28810 (vect_force_simple_reduction): Adjust and set
28811 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28812 (vectorizable_reduction): Do not re-do reduction analysis
28813 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28814 * tree-parloops.c (gather_scalar_reductions): Adjust.
28816 2017-05-30 Richard Biener <rguenther@suse.de>
28818 PR middle-end/80901
28819 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
28822 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
28824 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
28825 Introduce unknown_misalignment parameter and remove vf.
28826 (vect_peeling_hash_get_lowest_cost):
28827 Pass unknown_misalignment parameter.
28828 (vect_enhance_data_refs_alignment):
28829 Fix unsupportable data ref treatment.
28831 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28833 * tree-vect-data-refs.c (vect_get_data_access_cost):
28834 Workaround for SLP handling.
28835 (vect_enhance_data_refs_alignment):
28836 Compute costs for doing no peeling at all, compare to the best
28837 peeling costs so far and avoid peeling if cheaper.
28839 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28841 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
28842 Return peeling info and set costs to zero for unlimited cost
28844 (vect_enhance_data_refs_alignment): Also inspect all datarefs
28845 with unknown misalignment. Compute and costs for unknown
28846 misalignment, compare them to the costs for known misalignment
28847 and choose the cheapest for peeling.
28849 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28851 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
28852 (vect_get_peeling_costs_all_drs): Create function.
28853 (vect_peeling_hash_get_lowest_cost):
28854 Use vect_get_peeling_costs_all_drs.
28855 (vect_peeling_supportable): Create function.
28856 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
28858 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28860 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
28861 DR_HAS_NEGATIVE_STEP.
28862 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
28863 (vect_enhance_data_refs_alignment): Use.
28864 (vect_duplicate_ssa_name_ptr_info): Use.
28865 * tree-vectorizer.h (dr_misalignment): Use.
28866 (known_alignment_for_access_p): Use.
28868 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28871 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
28873 (has_section_name): New function.
28875 2017-05-30 Martin Liska <mliska@suse.cz>
28878 * auto-profile.c (get_function_decl_from_block): Fix
28881 2017-05-30 Richard Biener <rguenther@suse.de>
28883 PR middle-end/80876
28884 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
28886 2017-05-30 Martin Liska <mliska@suse.cz>
28888 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
28889 * dumpfile.h (struct dump_file_info): Remove ctors.
28891 2017-05-30 Martin Liska <mliska@suse.cz>
28893 * predict.def: Fix GNU coding style.
28895 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28897 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28898 Mark 'to' argument with ATTRIBUTE_UNUSED.
28900 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28902 * config/xtensa/xtensa.c (xtensa_emit_call): Use
28903 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
28904 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
28907 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
28909 * doc/install.texi (Options specification): Restore entry of
28910 --enable-sjlj-exceptions.
28912 2017-05-27 Michael Eager <eager@eagercon.com>
28915 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
28917 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
28919 * config/microblaze/microblaze.h
28920 (FIXED_REGISTERS): Update in macro.
28921 (CALL_USED_REGISTERS): Update in macro.
28923 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
28925 * doc/install.texi: Add links to macOS binary distributions.
28927 2017-05-27 Jakub Jelinek <jakub@redhat.com>
28931 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28933 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28935 2017-05-26 Martin Liska <mliska@suse.cz>
28937 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
28939 2017-05-26 Martin Liska <mliska@suse.cz>
28941 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
28942 always leading ';; '.
28943 (dump_bb_info): Likewise.
28944 (brief_dump_cfg): Likewise.
28945 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
28946 * dumpfile.c: Remove usage of TDF_VERBOSE.
28947 * dumpfile.h (enum dump_kind): Likewise.
28948 (dump_gimple_bb_header): Do not use TDF_COMMENT.
28949 * print-tree.c (debug_verbose): Remove.
28950 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
28951 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
28952 * tree-diagnostic.c (default_tree_printer): Replace
28953 TDF_DIAGNOSTIC with TDF_SLIM.
28955 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28957 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
28958 in parameter loop, rather than loop_vinfo.
28959 (create_intersect_range_checks): Ditto.
28960 (vect_create_cond_for_alias_checks): Update call to above functions.
28962 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28964 PR tree-optimization/80815
28965 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
28966 for merging runtime alias checks. Handle negative DR_STEPs.
28968 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28970 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
28972 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
28973 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
28974 out code pruning runtime alias checks.
28975 * tree-data-ref.c (prune_runtime_alias_test_list): New function
28976 factored out from above.
28977 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28979 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28981 (prune_runtime_alias_test_list): New decalaration.
28983 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28985 * tree-vect-data-refs.c (compare_tree): Rename and move ...
28986 * tree-data-ref.c (data_ref_compare_tree): ... to here.
28987 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
28988 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
28989 (operator==, comp_dr_with_seg_len_pair): Ditto.
28990 (vect_prune_runtime_alias_test_list): Ditto.
28992 2017-05-26 Martin Liska <mliska@suse.cz>
28995 * params.def: Bound partial-inlining-entry-probability param.
28997 2017-05-26 Marek Polacek <polacek@redhat.com>
29000 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29003 2017-05-26 Richard Biener <rguenther@suse.de>
29005 PR tree-optimization/80842
29006 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29009 2017-05-26 Richard Biener <rguenther@suse.de>
29011 PR tree-optimization/80844
29012 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29014 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29016 * doc/md.texi (Machine Constraints): Update x86 family
29017 machine constraints section to match 'config/i386/constraints.md'.
29019 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29021 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29023 2017-05-25 Nathan Sidwell <nathan@acm.org>
29025 * doc/invoke.texi (--enable-languages): Update documentation.
29027 2017-05-25 Martin Liska <mliska@suse.cz>
29029 * dumpfile.c: Add TDF_FOLDING.
29030 * dumpfile.h (enum dump_kind): Likewise.
29031 * genmatch.c (dt_simplify::gen_1): Use it.
29033 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29035 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29037 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29039 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29040 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29042 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29044 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29045 * match.pd (X == C): Rewrite it here.
29046 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29047 with_certain_nonzero_bits2): New predicates.
29048 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29050 2017-05-24 Nathan Sidwell <nathan@acm.org>
29052 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29055 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29058 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29060 * config/powerpcspe: New port. Files are copied from the rs6000
29061 port, with "rs6000" in filenames replaced by "powerpcspe".
29063 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29065 PR rtl-optimization/80754
29066 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29068 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29070 * config/sparc/sparc.md (length): Return the correct value for -mflat
29071 sibcalls to match output_sibcall.
29073 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29077 * config/rs6000/rs6000.c (struct machine_function): Add new field
29079 (rs6000_get_separate_components): Init that field, use it.
29080 (rs6000_components_for_bb): Use the field.
29082 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29084 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29086 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29088 PR middle-end/80823
29089 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29091 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29094 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29095 address against address_operand predicate.
29096 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29098 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29100 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29101 parameters passed indirectly.
29103 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29105 * config/i386/i386.md (*movdi_internal): Remove SSE4
29106 alternative 18 (?r, *v). Update insn attributes.
29107 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29108 Update insn attributes.
29109 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29110 Update insn attributes.
29111 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29112 alternative 1 (r, v). Remove isa attribute.
29113 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29114 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29115 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29117 2017-05-23 Tom de Vries <tom@codesourcery.com>
29119 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29122 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29124 * cgraphunit.c (symbol_table::process_new_functions): Update.
29125 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29126 (inline_generate_summary): Rename to ...
29127 (ipa_fn_summary_generate): ... this one.
29128 (inline_read_summary): Rename to ...
29129 (ipa_fn_summary_read): ... this one.
29130 (inline_write_summary): Rename to ...
29131 (ipa_fn_summary_write): ... this one.
29132 (inline_free_summary): Rename to ...
29133 (ipa_free_fn_summary): ... this one.
29134 (pass_data_local_fn_summary, pass_local_fn_summary,
29135 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29136 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29137 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29138 make_pass_ipa_fn_summary): New.
29139 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29140 inline_write_summary, inline_free_summary): Remove.
29141 (ipa_free_fn_summary) : New.
29142 * ipa-inline.c (ipa_inline): Update.
29143 (pass_ipa_inline): Do not generate summaries.
29144 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29146 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29147 and add pass_ipa_fn_summary.
29148 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29150 (make_pass_inline_parameters): Remove.
29152 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29154 * omp-low.c (struct omp_context): Remove "default_kind" member.
29157 * omp-offload.c (execute_oacc_device_lower): Remove the
29158 parallelism dimensions function attributes for unparallelized
29159 OpenACC kernels constructs.
29161 2017-05-23 Martin Liska <mliska@suse.cz>
29163 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29165 (cgraph_edge::make_speculative): Likewise.
29166 (cgraph_edge::resolve_speculation): Likewise.
29167 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29168 (cgraph_node::dump): Likewise.
29169 * cgraph.h: Likewise.
29170 * cgraphunit.c (analyze_functions): Likewise.
29171 (symbol_table::compile): Likewise.
29172 * ipa-cp.c (print_all_lattices): Likewise.
29173 (determine_versionability): Likewise.
29174 (initialize_node_lattices): Likewise.
29175 (ipcp_verify_propagated_values): Likewise.
29176 (estimate_local_effects): Likewise.
29177 (update_profiling_info): Likewise.
29178 (create_specialized_node): Likewise.
29179 (perhaps_add_new_callers): Likewise.
29180 (decide_about_value): Likewise.
29181 (decide_whether_version_node): Likewise.
29182 (identify_dead_nodes): Likewise.
29183 (ipcp_store_bits_results): Likewise.
29184 * ipa-devirt.c (dump_targets): Likewise.
29185 (ipa_devirt): Likewise.
29186 * ipa-icf.c (sem_item::dump): Likewise.
29187 (sem_function::equals): Likewise.
29188 (sem_variable::equals): Likewise.
29189 (sem_item_optimizer::read_section): Likewise.
29190 (sem_item_optimizer::execute): Likewise.
29191 (congruence_class::dump): Likewise.
29192 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29193 (dump_inline_summary): Likewise.
29194 (estimate_node_size_and_time): Likewise.
29195 (inline_analyze_function): Likewise.
29196 * ipa-inline-transform.c (inline_call): Likewise.
29197 * ipa-inline.c (report_inline_failed_reason): Likewise.
29198 (want_early_inline_function_p): Likewise.
29199 (edge_badness): Likewise.
29200 (update_edge_key): Likewise.
29201 (inline_small_functions): Likewise.
29202 * ipa-profile.c (ipa_profile): Likewise.
29203 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29204 (ipa_make_edge_direct_to_target): Likewise.
29205 (remove_described_reference): Likewise.
29206 (ipa_impossible_devirt_target): Likewise.
29207 (propagate_controlled_uses): Likewise.
29208 (ipa_print_node_params): Likewise.
29209 (ipcp_transform_function): Likewise.
29210 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29211 (propagate_pure_const): Likewise.
29212 * ipa-reference.c (generate_summary): Likewise.
29213 (read_write_all_from_decl): Likewise.
29214 (propagate): Likewise.
29215 (ipa_reference_read_optimization_summary): Likewise.
29216 * ipa-utils.c (ipa_merge_profiles): Likewise.
29217 * ipa.c (walk_polymorphic_call_targets): Likewise.
29218 (symbol_table::remove_unreachable_nodes): Likewise.
29219 (ipa_single_use): Likewise.
29220 * passes.c (execute_todo): Likewise.
29221 * predict.c (drop_profile): Likewise.
29222 * symtab.c (symtab_node::get_dump_name): New function.
29223 (symtab_node::dump_name): Likewise.
29224 (symtab_node::dump_asm_name): Likewise.
29225 (symtab_node::dump_references): Likewise.
29226 (symtab_node::dump_referring): Likewise.
29227 (symtab_node::dump_base): Likewise.
29228 (symtab_node::debug_symtab): Likewise.
29229 * tree-sra.c (convert_callers_for_node): Likewise.
29230 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29231 * value-prof.c (init_node_map): Likewise.
29233 2017-05-23 Martin Liska <mliska@suse.cz>
29235 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29236 and symtab_node::debug_symtab to symbol_table::debug.
29237 * cgraphunit.c (analyze_functions): Use the renamed function.
29238 (symbol_table::compile): Likewise.
29239 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29240 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29241 * passes.c (execute_todo): Likewise.
29242 * symtab.c (symbol_table::dump): New function.
29243 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29245 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29247 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29248 that nonconst implies exec.
29250 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29252 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29253 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29254 (inline_edge_summary_vec): Turn into ...
29255 (ipa_call_summaries): ... this one.
29256 (redirect_to_unreachable, edge_set_predicate,
29257 evaluate_properties_for_edge, inline_summary_alloc,
29258 reset_ipa_call_summary, reset_inline_summary,
29259 inline_summary_t::duplicate): Update.
29260 (inline_edge_duplication_hook): Turn to ...
29261 (ipa_call_summary_t::duplicate): ... this one.
29262 (inline_edge_removal_hook): Turn to ...
29263 (ipa_call_summary_t::remove): ... this one.
29264 (dump_inline_edge_summary): Turn to ...
29265 (dump_ipa_call_summary): ... this one.
29266 (estimate_function_body_sizes): Update.
29267 (inline_update_callee_summaries): Update.
29268 (remap_edge_change_prob): Update.
29269 (remap_edge_summaries): Update.
29270 (inline_merge_summary): Update.
29271 (do_estimate_edge_time): Update.
29272 (inline_generate_summary): Update.
29273 (inline_read_section): Update.
29274 (inline_read_summary): Update.
29275 (inline_free_summary): Update.
29276 * ipa-inline.c (can_inline_edge_p): Update.
29277 (compute_inlined_call_time): Update.
29278 (want_inline_small_function_p): Update.
29279 (edge_badness): Update.
29280 (early_inliner): Update.
29281 * ipa-inline.h (inline_edge_summary): Turn to ...
29282 (ipa_call_summary): ... this one.
29283 (ipa_call_summary_t): New class.
29284 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29285 (ipa_call_summaries): New.
29286 (inline_edge_summary): Remove.
29287 (estimate_edge_growth): Update.
29288 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29289 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29290 * ipa-split.c (execute_split_functions): Update.
29291 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29293 2017-05-23 Tom de Vries <tom@codesourcery.com>
29295 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29296 attributes): Document rdrand effective target.
29298 2017-05-23 Tom de Vries <tom@codesourcery.com>
29300 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29301 attributes): Sort alphabetically.
29303 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29305 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29307 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29310 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29311 V2DF/V2DI splat into two separate patterns, one that handles
29312 registers, and the other that only handles memory. Drop support
29313 for splatting from a GPR on ISA 2.07 and then splitting the
29314 splat into direct move and splat.
29315 (vsx_splat_<mode>_reg): Likewise.
29316 (vsx_splat_<mode>_mem): Likewise.
29318 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29320 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29322 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29324 PR middle-end/80809
29325 * omp-low.c (finish_taskreg_remap): New function.
29326 (finish_taskreg_scan): If unit size of ctx->record_type
29327 is non-constant, unshare the size expression and replace
29328 decls in it with possible outer var refs.
29330 PR middle-end/80809
29331 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29332 GOVD_SHARED rather than GOVD_PRIVATE with it.
29333 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29334 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29336 PR middle-end/80853
29337 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29338 as last argument to build_outer_var_ref for pointer bases of array
29339 section reductions.
29341 2017-05-19 Martin Sebor <msebor@redhat.com>
29343 * print-tree.c (print_node): Print DECL_READ_P flag.
29345 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29347 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29348 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29349 * cgraph.c: Likewise.
29350 * cgraphunit.c: Likewise.
29351 * gengtype.c: Likewise.
29352 * ipa-cp.c: Likewise.
29353 * ipa-devirt.c: Likewise.
29354 * ipa-icf.c: Likewise.
29355 * ipa-predicate.c: Likewise.
29356 * ipa-profile.c: Likewise.
29357 * ipa-prop.c: Likewise.
29358 * ipa-split.c: Likewise.
29360 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29361 edge_predicate_pool, dump_inline_hints,
29362 inline_summary::account_size_time, redirect_to_unreachable,
29363 edge_set_predicate, set_hint_predicate,
29364 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29365 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29366 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29367 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29368 ipa_call_summary_t::remove, initialize_growth_caches,
29369 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29370 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29371 mark_modified, unmodified_parm_1, unmodified_parm,
29372 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29373 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29374 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29375 will_be_nonconstant_predicate, record_modified_bb_info,
29376 get_minimal_bb, record_modified, param_change_prob,
29377 phi_result_unknown_predicate, predicate_for_phi_result,
29378 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29379 estimate_function_body_sizes, compute_inline_parameters,
29380 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29381 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29382 inline_update_callee_summaries, remap_edge_change_prob,
29383 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29384 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29385 inline_analyze_function, inline_summary_t::insert,
29386 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29387 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29388 inline_free_summary): Move to ipa-fnsummary.h
29389 (predicate_t): Remove.
29390 * ipa-fnsummary.c: New file.
29391 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29392 (enum inline_hints_vals, inline_hints, agg_position_info,
29393 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29394 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29395 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29396 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29397 inline_read_summary, inline_write_summary, inline_free_summary,
29398 inline_analyze_function, initialize_inline_failed,
29399 inline_merge_summary, inline_update_overall_summary,
29400 compute_inline_parameters): Move to ipa-fnsummary.h
29401 * ipa-fnsummary.h: New file.
29402 * ipa-inline-transform.h: Include ipa-inline.h.
29403 * ipa-inline.c: LIkewise.
29405 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29407 * ipa-inline.c (edge_badness): Use inlined_time instead of
29408 inline_summaries->get.
29410 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29412 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29414 2017-05-22 Nathan Sidwell <nathan@acm.org>
29416 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29417 (fdump-lang): Document 'raw' option.
29418 * dumpfile.h (TDI_tu): Delete.
29419 * dumpfile.c (dump_files): Remove translation-unit.
29420 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29422 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29424 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29425 command option from $(AWK) call.
29426 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29427 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29428 [FORMAT]: Remove handling of variable.
29429 * config/avr/t-multilib: Regenerate.
29431 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29433 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29435 (dump_inline_summary): Do not print self_time.
29436 (estimate_function_body_sizes): Do not set self_time.
29437 (compute_inline_parameters): Likewise.
29438 (inline_read_section, inline_write_summary): Do not stream self_time.
29439 * ipa-inline.h (inline_summary): Drop self_time.
29441 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29443 * ipa-inline-analysis.c (account_size_time): Rename to ...
29444 (inline_summary::account_size_time): ... this one.
29445 (reset_ipa_call_summary): Turn to ...
29446 (ipa_call_summary::reset): ... this one.
29447 (reset_inline_summary): Turn to ...
29448 (inline_summary::reset): ... this one.
29449 (inline_summary_t::remove): Update.
29450 (inline_summary_t::duplicate): Update.
29451 (ipa_call_summary_t::remove): Update.
29452 (dump_inline_summary): Update.
29453 (estimate_function_body_sizes): Update.
29454 (compute_inline_parameters): Update.
29455 (estimate_node_size_and_time): Update.
29456 (inline_merge_summary): Update.
29457 (inline_update_overall_summary): Update.
29458 (inline_read_section): Update.
29459 (inline_write_summary): Update.
29460 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29461 add account_size_time and reset member functions.
29462 (ipa_call_summary): Add reset function.
29463 * ipa-predicate.h (predicate::operator &): Constify.
29465 2017-05-22 Richard Biener <rguenther@suse.de>
29467 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29469 2017-05-19 Jason Merrill <jason@redhat.com>
29471 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29473 2017-05-19 Marek Polacek <polacek@redhat.com>
29476 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29477 TYPE_OVERFLOW_WRAPS checks.
29479 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29481 * tree-core.h (enum omp_clause_default_kind): Add
29482 "OMP_CLAUSE_DEFAULT_PRESENT".
29483 * tree-pretty-print.c (dump_omp_clause): Handle it.
29484 * gimplify.c (enum gimplify_omp_var_data): Add
29485 "GOVD_MAP_FORCE_PRESENT".
29486 (gimplify_adjust_omp_clauses_1): Map it to
29487 "GOMP_MAP_FORCE_PRESENT".
29488 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29490 * gimplify.c (oacc_default_clause): Clarify.
29492 2017-05-19 Nathan Sidwell <nathan@acm.org>
29494 LANG_HOOK_REGISTER_DUMPS
29495 * toplev.c (general_init): Call register dump lang hook.
29496 * doc/invoke.texi: Document -fdump-lang option family.
29497 * dumpfile.c (dump_files): Remove class dump here.
29498 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29499 * dumpfile.h (tree_dump_index): Remove TDI_class.
29500 * langhooks-def.h (lhd_register_dumps): Declare.
29501 (LANG_HOOKS_REGISTER_DUMPS): Define.
29502 (LANG_HOOKS_INITIALIZER): Add it.
29503 * langhooks.c (lhd_register_dumps): Define.
29504 * langhooks.h (struct lang_hooks): Add register_dumps.
29506 2017-05-19 Nathan Sidwell <nathan@acm.org>
29508 * context.h (context::set_passes): New.
29509 * context.c (context::context): Do not create pass manager.
29510 * toplev.c (general_init): Create pass manager here.
29512 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29514 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29515 use this splitter if two add or or instructions would also work for
29516 the constant we want to generate.
29518 2017-05-19 Richard Biener <rguenther@suse.de>
29521 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29522 predicate evaluation.
29524 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29526 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29528 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29531 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29533 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29534 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29535 (inline_edge_summary_vec): Turn into ...
29536 (ipa_call_summaries): ... this one.
29537 (redirect_to_unreachable, edge_set_predicate,
29538 evaluate_properties_for_edge, inline_summary_alloc,
29539 reset_ipa_call_summary, reset_inline_summary,
29540 inline_summary_t::duplicate): Update.
29541 (inline_edge_duplication_hook): Turn to ...
29542 (ipa_call_summary_t::duplicate): ... this one.
29543 (inline_edge_removal_hook): Turn to ...
29544 (ipa_call_summary_t::remove): ... this one.
29545 (dump_inline_edge_summary): Turn to ...
29546 (dump_ipa_call_summary): ... this one.
29547 (estimate_function_body_sizes): Update.
29548 (inline_update_callee_summaries): Update.
29549 (remap_edge_change_prob): Update.
29550 (remap_edge_summaries): Update.
29551 (inline_merge_summary): Update.
29552 (do_estimate_edge_time): Update.
29553 (inline_generate_summary): Update.
29554 (inline_read_section): Update.
29555 (inline_read_summary): Update.
29556 (inline_free_summary): Update.
29557 * ipa-inline.c (can_inline_edge_p): Update.
29558 (compute_inlined_call_time): Update.
29559 (want_inline_small_function_p): Update.
29560 (edge_badness): Update.
29561 (early_inliner): Update.
29562 * ipa-inline.h (inline_edge_summary): Turn to ...
29563 (ipa_call_summary): ... this one.
29564 (ipa_call_summary_t): New class.
29565 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29566 (ipa_call_summaries): New.
29567 (inline_edge_summary): Remove.
29568 (estimate_edge_growth): Update.
29569 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29570 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29571 * ipa-split.c (execute_split_functions): Update.
29572 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29574 2017-05-19 Richard Biener <rguenther@suse.de>
29576 PR middle-end/80764
29577 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29579 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29581 * config/rs6000/rs6000.c (struct machine_function): Add field
29582 fpr_is_wrapped_separately.
29583 (rs6000_get_separate_components): Use 64 components. Handle the
29584 new FPR components.
29585 (rs6000_components_for_bb): Handle the FPR components.
29586 (rs6000_emit_prologue_components): Handle the FPR components.
29587 (rs6000_emit_epilogue_components): Handle the FPR components.
29588 (rs6000_set_handled_components): Handle the FPR components.
29589 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29590 that are already separately shrink-wrapped.
29591 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29592 that are already separately shrink-wrapped.
29594 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29597 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29600 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29601 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29602 cases where the register allocator uses a move and an offsettable
29603 memory operation to/from a FPR register on ISA 2.06/2.07.
29604 (define_peephole2 for Altivec d-form store): Likewise.
29606 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29609 * config/i386/mmx.md (*mov<mode>_internal): Enable
29610 alternatives 11, 12, 13 and 14 also for 32bit targets.
29611 Remove alternatives 15, 16, 17 and 18.
29612 * config/i386/sse.md (vec_concatv2di): Change
29613 alternative (!x, *y) to (x, ?!*Yn).
29615 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29617 * dumpfile.h (enum dump_kind): Remove stray comma.
29619 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29621 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29622 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29623 predicate::num_conditions
29624 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29625 (CHANGED): turn into predicate::changed.
29626 (agg_position_info): Move to ipa-predicate.h
29627 (add_condition, predicate::add_clause, predicate::operator &=,
29628 predicate::or_with, predicate::evaluate, predicate::probability,
29629 dump_condition, dump_clause, predicate::dump,
29630 predicate::remap_after_duplication, predicate::remap_after_inlining,
29631 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29632 (evaluate_conditions_for_known_args): Update.
29633 (set_cond_stmt_execution_predicate): Update.
29634 * ipa-inline.h: Include ipa-predicate.h
29635 (condition, inline_param_summary, conditions, agg_position_info,
29636 predicate): Move to ipa-predicate.h
29637 * ipa-predicate.c: New file.
29638 * ipa-predicate.h: New file.
29640 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29642 * final.c (leaf_function_p): Check we are not in a sequence.
29644 2017-05-18 Martin Liska <mliska@suse.cz>
29646 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29647 * dumpfile.c (dump_register): Use new enum dump_kind.
29648 (get_dump_file_name): Likewise.
29649 (dump_enable_all): Likewise.
29650 (dump_switch_p_1): Likewise.
29651 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29652 * dumpfile.h (enum dump_kind): New enum type.
29653 (struct dump_file_info): Create constructor and
29654 format fields and comments.
29655 * passes.c (pass_manager::register_one_dump_file):
29657 * statistics.c (statistics_early_init): Likewise.
29658 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29659 TDF_TREE with TDF_SLIM.
29660 (gather_memory_references_ref): Likewise.
29662 2017-05-18 Martin Liska <mliska@suse.cz>
29664 * vec.h (struct vnull): Use it.
29666 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29668 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29669 (true_predicate, false_predicate, true_predicate_p,
29670 false_predicate_p): Remove.
29671 (single_cond_predicate, not_inlined_predicate): Turn to member function
29673 (add_condition): Update.
29674 (add_clause): Turn to...
29675 (predicate::add_clause): ... this one; update; allow passing NULL
29677 (and_predicates): Turn to ...
29678 (predicate::operator &=): ... this one.
29679 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29680 (or_predicates): Turn to ...
29681 (predicate::or_with): ... this one.
29682 (evaluate_predicate): Turn to ...
29683 (predicate::evaluate): ... this one.
29684 (predicate_probability): Turn to ...
29685 (predicate::probability): ... this one.
29686 (dump_condition): Update.
29687 (dump_predicate): Turn to ...
29688 (predicate::dump): ... this one.
29689 (account_size_time): Update.
29690 (edge_set_predicate): Update.
29691 (set_hint_predicate): UPdate.
29692 (evaluate_conditions_for_known_args): Update.
29693 (evaluate_properties_for_edge): Update.
29694 (remap_predicate_after_duplication): Turn to...
29695 (predicate::remap_after_duplication): ... this one.
29696 (remap_hint_predicate_after_duplication): Update.
29697 (inline_summary_t::duplicate): UPdate.
29698 (dump_inline_edge_summary): Update.
29699 (dump_inline_summary): Update.
29700 (set_cond_stmt_execution_predicate): Update.
29701 (set_switch_stmt_execution_predicate): Update.
29702 (compute_bb_predicates): Update.
29703 (will_be_nonconstant_expr_predicate): Update.
29704 (will_be_nonconstant_predicate): Update.
29705 (phi_result_unknown_predicate): Update.
29706 (predicate_for_phi_result): Update.
29707 (array_index_predicate): Update.
29708 (estimate_function_body_sizes): Update.
29709 (estimate_node_size_and_time): Update.
29710 (estimate_ipcp_clone_size_and_time): Update.
29711 (remap_predicate): Rename to ...
29712 (predicate::remap_after_inlining): ... this one.
29713 (remap_hint_predicate): Update.
29714 (inline_merge_summary): Update.
29715 (inline_update_overall_summary): Update.
29716 (estimate_size_after_inlining): Update.
29717 (read_predicate): Rename to ...
29718 (predicate::stream_in): ... this one.
29719 (read_inline_edge_summary): Update.
29720 (write_predicate): Rename to ...
29721 (predicate::stream_out): ... this one.
29722 (write_inline_edge_summary): Update.
29723 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29724 (clause_t): Turn to uint32_t
29725 (predicate): Turn to class; implement constructor and operators
29727 (size_time_entry): Update.
29728 (inline_summary): Update.
29729 (inline_edge_summary): Update.
29731 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29733 * fold-const.c (fold_binary_loc): Move transformation...
29734 * match.pd (C - X CMP X): ... here.
29736 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29738 * config/sparc/sparc.c (sparc_option_override): Set function
29739 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29740 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29742 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29744 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29746 2017-05-18 Marek Polacek <polacek@redhat.com>
29749 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29750 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29751 gimple_assign_load_p.
29753 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29755 PR middle-end/80692
29756 * real.c (do_compare): Give decimal_do_compare preference over
29757 comparing just the signs.
29759 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29761 * doc/md.texi (Canonicalization of Instructions): Describe the
29762 canonical form of instructions that inherently set a condition
29765 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29767 PR middle-end/80775
29768 * tree-cfg.c: Move deletion of unreachable case statements to after
29769 the merging of consecutive case labels.
29771 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29773 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29774 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29775 restoring of callee-saved registers.
29777 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29779 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29780 * config/visium/visium.c (single_set_and_flags): Likewise.
29781 * config/visium/visium.md (Substitutions): Likewise.
29783 2017-05-17 Martin Liska <mliska@suse.cz>
29785 * cfg.c: Introduce dump_flags_t type and
29786 use it instead of int type.
29788 * cfghooks.c: Likewise.
29789 * cfghooks.h (struct cfg_hooks): Likewise.
29790 * cfgrtl.c: Likewise.
29791 * cfgrtl.h: Likewise.
29792 * cgraph.c (cgraph_node::get_body): Likewise.
29793 * coretypes.h: Likewise.
29794 * domwalk.c: Likewise.
29795 * domwalk.h: Likewise.
29796 * dumpfile.c (struct dump_option_value_info): Likewise.
29797 (dump_enable_all): Likewise.
29798 (dump_switch_p_1): Likewise.
29799 (opt_info_switch_p): Likewise.
29800 * dumpfile.h (enum tree_dump_index): Likewise.
29801 (struct dump_file_info): Likewise.
29802 * genemit.c: Likewise.
29803 * generic-match-head.c: Likewise.
29804 * gengtype.c (open_base_files): Likewise.
29805 * gimple-pretty-print.c: Likewise.
29806 * gimple-pretty-print.h: Likewise.
29807 * graph.c (print_graph_cfg): Likewise.
29808 * graphite-scop-detection.c (dot_all_sese): Likewise.
29809 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
29810 * loop-unroll.c (report_unroll): Likewise.
29811 * passes.c (pass_manager::register_one_dump_file): Likewise.
29812 * print-tree.c: Likewise.
29813 * statistics.c: Likewise.
29814 * tree-cfg.c: Likewise.
29815 * tree-cfg.h: Likewise.
29816 * tree-dfa.c: Likewise.
29817 * tree-dfa.h: Likewise.
29818 * tree-dump.c (dump_function): Likewise.
29819 * tree-dump.h (struct dump_info): Likewise.
29820 * tree-pretty-print.c: Likewise.
29821 * tree-pretty-print.h: Likewise.
29822 * tree-ssa-live.c: Likewise.
29823 * tree-ssa-live.h: Likewise.
29824 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
29825 * tree-vect-loop.c: Likewise.
29826 * tree-vect-slp.c: Likewise.
29828 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
29829 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29831 PR tree-optimization/80457
29832 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
29833 of all arguments to a statement as scalar_to_vec operations.
29834 (vectorizable_call): Adjust call to vect_model_simple_cost for
29836 (vectorizable_conversion): Likewise.
29837 (vectorizable_assignment): Likewise.
29838 (vectorizable_shift): Likewise.
29839 (vectorizable_operation): Likewise.
29840 (vectorizable_comparison): Likewise.
29841 (vect_is_simple_cond): Record the def types for operands.
29842 (vectorizable_condition): Likewise, call vect_model_simple_cost.
29843 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
29844 for statement argument count.
29846 2017-05-16 Carl Love <cel@us.ibm.com>
29848 * config/rs6000/rs6000-c: Add support for built-in functions
29849 vector unsigned long long vec_bperm (vector unsigned long long,
29850 vector unsigned char)
29851 vector signed long long vec_mule (vector signed int,
29853 vector unsigned long long vec_mule (vector unsigned int,
29854 vector unsigned int)
29855 vector signed long long vec_mulo (vector signed int,
29857 vector unsigned long long vec_mulo (vector unsigned int,
29858 vector unsigned int)
29859 vector signed char vec_sldw (vector signed char,
29860 vector signed char,
29862 vector unsigned char vec_sldw (vector unsigned char,
29863 vector unsigned char,
29865 vector signed short vec_sldw (vector signed short,
29866 vector signed short,
29868 vector unsigned short vec_sldw (vector unsigned short,
29869 vector unsigned short,
29871 vector signed int vec_sldw (vector signed int,
29874 vector unsigned int vec_sldw (vector unsigned int,
29875 vector unsigned int,
29877 vector signed long long vec_sldw (vector signed long long,
29878 vector signed long long,
29880 vector unsigned long long vec_sldw (vector unsigned long long,
29881 vector unsigned long long,
29883 * config/rs6000/rs6000-c: Add support for built-in functions
29884 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
29885 * config/rs6000/altivec.h: Add defintion for vec_sldw.
29886 * doc/extend.texi: Update the built-in documentation for the
29887 new built-in functions.
29889 2017-05-16 Marek Polacek <polacek@redhat.com>
29893 * tree.c (save_expr): Don't fold the expression.
29895 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
29897 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
29898 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
29899 and (?*y,m). Update insn attributes.
29901 2017-05-16 Martin Liska <mliska@suse.cz>
29903 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
29904 flags argument of print_gimple_stmt, print_gimple_expr,
29905 print_generic_stmt and print_generic_expr.
29906 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
29907 * coretypes.h: Likewise.
29908 * except.c (dump_eh_tree): Likewise.
29909 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
29910 * gimple-pretty-print.h: Likewise.
29911 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
29912 (backprop::push_to_worklist): Likewise.
29913 (backprop::pop_from_worklist): Likewise.
29914 (backprop::process_use): Likewise.
29915 (backprop::intersect_uses): Likewise.
29916 (note_replacement): Likewise.
29917 * gimple-ssa-store-merging.c
29918 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
29919 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
29920 (pass_store_merging::execute): Likewise.
29921 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
29922 (ssa_base_cand_dump_callback): Likewise.
29923 (dump_incr_vec): Likewise.
29924 (replace_refs): Likewise.
29925 (replace_mult_candidate): Likewise.
29926 (create_add_on_incoming_edge): Likewise.
29927 (create_phi_basis): Likewise.
29928 (insert_initializers): Likewise.
29929 (all_phi_incrs_profitable): Likewise.
29930 (introduce_cast_before_cand): Likewise.
29931 (replace_one_candidate): Likewise.
29932 * gimplify.c (gimplify_expr): Likewise.
29933 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
29934 (set_rename): Likewise.
29935 (rename_uses): Likewise.
29936 (copy_loop_phi_nodes): Likewise.
29937 (add_close_phis_to_merge_points): Likewise.
29938 (copy_loop_close_phi_args): Likewise.
29939 (copy_cond_phi_args): Likewise.
29940 (graphite_copy_stmts_from_block): Likewise.
29941 (translate_pending_phi_nodes): Likewise.
29942 * graphite-poly.c (print_pdr): Likewise.
29943 (dump_gbb_cases): Likewise.
29944 (dump_gbb_conditions): Likewise.
29945 (print_scop_params): Likewise.
29946 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
29947 (build_cross_bb_scalars_use): Likewise.
29948 (gather_bbs::before_dom_children): Likewise.
29949 * hsa-dump.c (dump_hsa_immed): Likewise.
29950 * ipa-cp.c (print_ipcp_constant_value): Likewise.
29951 (get_replacement_map): Likewise.
29952 * ipa-inline-analysis.c (dump_condition): Likewise.
29953 (estimate_function_body_sizes): Likewise.
29954 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
29955 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
29956 * ipa-prop.c (ipa_dump_param): Likewise.
29957 (ipa_print_node_jump_functions_for_edge): Likewise.
29958 (ipa_modify_call_arguments): Likewise.
29959 (ipa_modify_expr): Likewise.
29960 (ipa_dump_param_adjustments): Likewise.
29961 (ipa_dump_agg_replacement_values): Likewise.
29962 (ipcp_modif_dom_walker::before_dom_children): Likewise.
29963 * ipa-pure-const.c (check_stmt): Likewise.
29964 (pass_nothrow::execute): Likewise.
29965 * ipa-split.c (execute_split_functions): Likewise.
29966 * omp-offload.c (dump_oacc_loop_part): Likewise.
29967 (dump_oacc_loop): Likewise.
29968 * trans-mem.c (tm_log_emit): Likewise.
29969 (tm_memopt_accumulate_memops): Likewise.
29970 (dump_tm_memopt_set): Likewise.
29971 (dump_tm_memopt_transform): Likewise.
29972 * tree-cfg.c (gimple_verify_flow_info): Likewise.
29973 (print_loop): Likewise.
29974 * tree-chkp-opt.c (chkp_print_addr): Likewise.
29975 (chkp_gather_checks_info): Likewise.
29976 (chkp_get_check_result): Likewise.
29977 (chkp_remove_check_if_pass): Likewise.
29978 (chkp_use_outer_bounds_if_possible): Likewise.
29979 (chkp_reduce_bounds_lifetime): Likewise.
29980 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
29981 (chkp_mark_completed_bounds): Likewise.
29982 (chkp_register_incomplete_bounds): Likewise.
29983 (chkp_mark_invalid_bounds): Likewise.
29984 (chkp_maybe_copy_and_register_bounds): Likewise.
29985 (chkp_build_returned_bound): Likewise.
29986 (chkp_get_bound_for_parm): Likewise.
29987 (chkp_build_bndldx): Likewise.
29988 (chkp_get_bounds_by_definition): Likewise.
29989 (chkp_generate_extern_var_bounds): Likewise.
29990 (chkp_get_bounds_for_decl_addr): Likewise.
29991 * tree-chrec.c (chrec_apply): Likewise.
29992 * tree-data-ref.c (dump_data_reference): Likewise.
29993 (dump_subscript): Likewise.
29994 (dump_data_dependence_relation): Likewise.
29995 (analyze_overlapping_iterations): Likewise.
29996 * tree-inline.c (expand_call_inline): Likewise.
29997 (tree_function_versioning): Likewise.
29998 * tree-into-ssa.c (dump_defs_stack): Likewise.
29999 (dump_currdefs): Likewise.
30000 (dump_names_replaced_by): Likewise.
30001 (dump_update_ssa): Likewise.
30002 (update_ssa): Likewise.
30003 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30004 * tree-parloops.c (build_new_reduction): Likewise.
30005 (try_create_reduction_list): Likewise.
30006 (ref_conflicts_with_region): Likewise.
30007 (oacc_entry_exit_ok_1): Likewise.
30008 (oacc_entry_exit_single_gang): Likewise.
30009 * tree-pretty-print.h: Likewise.
30010 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30011 (get_scalar_evolution): Likewise.
30012 (add_to_evolution): Likewise.
30013 (get_loop_exit_condition): Likewise.
30014 (analyze_evolution_in_loop): Likewise.
30015 (analyze_initial_condition): Likewise.
30016 (analyze_scalar_evolution): Likewise.
30017 (instantiate_scev): Likewise.
30018 (number_of_latch_executions): Likewise.
30019 (gather_chrec_stats): Likewise.
30020 (final_value_replacement_loop): Likewise.
30021 (scev_const_prop): Likewise.
30022 * tree-sra.c (dump_access): Likewise.
30023 (disqualify_candidate): Likewise.
30024 (create_access): Likewise.
30025 (reject): Likewise.
30026 (maybe_add_sra_candidate): Likewise.
30027 (create_access_replacement): Likewise.
30028 (analyze_access_subtree): Likewise.
30029 (analyze_all_variable_accesses): Likewise.
30030 (sra_modify_assign): Likewise.
30031 (initialize_constant_pool_replacements): Likewise.
30032 (find_param_candidates): Likewise.
30033 (decide_one_param_reduction): Likewise.
30034 (replace_removed_params_ssa_names): Likewise.
30035 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30036 * tree-ssa-copy.c (dump_copy_of): Likewise.
30037 (copy_prop_visit_cond_stmt): Likewise.
30038 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30039 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30040 (record_equivalences_from_stmt): Likewise.
30041 * tree-ssa-dse.c (compute_trims): Likewise.
30042 (delete_dead_call): Likewise.
30043 (delete_dead_assignment): Likewise.
30044 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30045 (forward_propagate_into_cond): Likewise.
30046 (pass_forwprop::execute): Likewise.
30047 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30048 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30050 (move_computations_worker): Likewise.
30051 (execute_sm): Likewise.
30052 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30053 (remove_exits_and_undefined_stmts): Likewise.
30054 (remove_redundant_iv_tests): Likewise.
30055 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30056 (adjust_iv_update_pos): Likewise.
30057 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30058 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30059 (value_replacement): Likewise.
30060 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30061 * tree-ssa-pre.c (print_pre_expr): Likewise.
30062 (get_representative_for): Likewise.
30063 (create_expression_by_pieces): Likewise.
30064 (insert_into_preds_of_block): Likewise.
30065 (eliminate_insert): Likewise.
30066 (eliminate_dom_walker::before_dom_children): Likewise.
30067 (eliminate): Likewise.
30068 (remove_dead_inserted_code): Likewise.
30069 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30070 * tree-ssa-reassoc.c (get_rank): Likewise.
30071 (eliminate_duplicate_pair): Likewise.
30072 (eliminate_plus_minus_pair): Likewise.
30073 (eliminate_not_pairs): Likewise.
30074 (undistribute_ops_list): Likewise.
30075 (eliminate_redundant_comparison): Likewise.
30076 (update_range_test): Likewise.
30077 (optimize_range_tests_var_bound): Likewise.
30078 (optimize_vec_cond_expr): Likewise.
30079 (rewrite_expr_tree): Likewise.
30080 (rewrite_expr_tree_parallel): Likewise.
30081 (linearize_expr): Likewise.
30082 (break_up_subtract): Likewise.
30083 (linearize_expr_tree): Likewise.
30084 (attempt_builtin_powi): Likewise.
30085 (attempt_builtin_copysign): Likewise.
30086 (transform_stmt_to_copy): Likewise.
30087 (transform_stmt_to_multiply): Likewise.
30088 (dump_ops_vector): Likewise.
30089 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30090 (print_scc): Likewise.
30091 (set_ssa_val_to): Likewise.
30092 (visit_reference_op_store): Likewise.
30093 (visit_use): Likewise.
30094 (sccvn_dom_walker::before_dom_children): Likewise.
30095 (run_scc_vn): Likewise.
30096 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30098 (expr_hash_elt::print): Likewise.
30099 (const_and_copies::pop_to_marker): Likewise.
30100 (const_and_copies::record_const_or_copy_raw): Likewise.
30101 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30102 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30103 (dump_predicates): Likewise.
30104 (find_uninit_use): Likewise.
30105 (warn_uninitialized_phi): Likewise.
30106 (pass_late_warn_uninitialized::execute): Likewise.
30107 * tree-ssa.c (verify_vssa): Likewise.
30108 (verify_ssa): Likewise.
30109 (maybe_optimize_var): Likewise.
30110 * tree-vrp.c (dump_value_range): Likewise.
30111 (dump_all_value_ranges): Likewise.
30112 (dump_asserts_for): Likewise.
30113 (register_edge_assert_for_2): Likewise.
30114 (vrp_visit_cond_stmt): Likewise.
30115 (vrp_visit_switch_stmt): Likewise.
30116 (vrp_visit_stmt): Likewise.
30117 (vrp_visit_phi_node): Likewise.
30118 (simplify_cond_using_ranges_1): Likewise.
30119 (fold_predicate_in): Likewise.
30120 (evrp_dom_walker::before_dom_children): Likewise.
30121 (evrp_dom_walker::push_value_range): Likewise.
30122 (evrp_dom_walker::pop_value_range): Likewise.
30123 (execute_early_vrp): Likewise.
30125 2017-05-16 Richard Biener <rguenther@suse.de>
30127 * dwarf2out.c (loc_list_from_tree_1): Do not create
30128 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30130 2017-05-16 Richard Biener <rguenther@suse.de>
30132 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30134 (note_variable_value_in_expr): If we resolved the decl ref
30135 do not push to the stack.
30137 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30139 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30140 operations in fast-math mode.
30141 (vaddq_f16): Likewise.
30142 (vmul_f16): Likewise.
30143 (vmulq_f16): Likewise.
30144 (vsub_f16): Likewise.
30145 (vsubq_f16): Likewise.
30146 * config/arm/neon.md (add<mode>3): New.
30148 (fma:<VH:mode>3): New. Also remove outdated comment.
30151 2017-05-16 Martin Liska <mliska@suse.cz>
30155 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30156 (odr_types_equivalent_p): Likewise.
30158 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30160 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30162 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30165 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30166 non-interunit SSE move alternatives with '?'.
30167 (zero-extendsidi peephole2): New peephole to skip intermediate
30168 general register in SSE zero-extend sequence.
30170 2017-05-15 Jeff Law <law@redhat.com>
30172 * reorg.c (relax_delay_slots): Create a new variable to hold
30173 the temporary target rather than clobbering TARGET_LABEL.
30175 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30176 missing argument to extract_bit_field call.
30177 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30179 2017-05-15 Martin Liska <mliska@suse.cz>
30182 * gcc.c (process_command): Do not allow empty argument of -o option.
30184 2017-05-15 Renlin Li <renlin.li@arm.com>
30186 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30187 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30188 * config/aarch64/constraints.md (Usf): Add long call check.
30189 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30190 (call_value): Likewise.
30191 (sibcall): Likewise.
30192 (sibcall_value): Likewise.
30194 (call_value_insn): New.
30195 (sibcall_insn): Update rtx pattern.
30196 (sibcall_value_insn): Likewise.
30197 (call_internal): Remove.
30198 (call_value_internal): Likewise.
30199 (sibcall_internal): Likewise.
30200 (sibcall_value_internal): Likewise.
30201 (call_reg): Likewise.
30202 (call_symbol): Likewise.
30203 (call_value_reg): Likewise.
30204 (call_value_symbol): Likewise.
30206 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30209 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30211 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30213 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30214 compatible with CCGOCmode and with CCZmode.
30216 2017-05-14 Martin Sebor <msebor@redhat.com>
30218 PR middle-end/77671
30219 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30220 (gimple_fold_builtin_snprintf): Same.
30221 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30222 (gimple_fold_builtin_snprintf): Same.
30223 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30224 of character types.
30225 (is_call_safe): New function.
30226 (try_substitute_return_value): Call it.
30227 (try_simplify_call): New function.
30228 (pass_sprintf_length::handle_gimple_call): Call it.
30230 2017-05-14 Martin Sebor <msebor@redhat.com>
30232 PR middle-end/80669
30233 * builtins.c (expand_builtin_stpncpy): Simplify.
30235 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30237 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30238 * config/i386/i386.h
30239 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30240 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30241 (struct machine_function): Add new members call_ms2sysv,
30242 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30243 (struct machine_frame_state): New fields sp_realigned and
30244 sp_realigned_offset.
30245 * config/i386/i386.c
30246 (enum xlogue_stub): New enum.
30247 (enum xlogue_stub_sets): New enum.
30248 (class xlogue_layout): New class.
30249 (struct ix86_frame): New fields stack_realign_allocate_offset,
30250 stack_realign_offset and outlined_save_offset. Modify comments to
30251 detail stack layout when using out-of-line stubs.
30252 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30253 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30254 -mcall-ms2sysv-xlogues.
30255 (stub_managed_regs): New static variable.
30256 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30257 registers managed by out-of-line stub.
30258 (disable_call_ms2sysv_xlogues): New function.
30259 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30260 m->call_ms2sysv when appropriate and compute frame layout for
30262 (sp_valid_at, fp_valid_at): New inline functions.
30263 (choose_basereg): New function.
30264 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30266 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30267 Use align parameter of choose_baseaddr to generated aligned SSE movs
30269 (pro_epilogue_adjust_stack): Modify to track
30270 machine_frame_state::sp_realigned.
30271 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30272 (ix86_nsaved_sseregs): Likewise.
30273 (ix86_emit_save_regs): Likewise.
30274 (ix86_emit_save_regs_using_mov): Likewise.
30275 (ix86_emit_save_sse_regs_using_mov): Likewise.
30276 (get_scratch_register_on_entry): Likewise.
30277 (gen_frame_set): New function.
30278 (gen_frame_load): Likewise.
30279 (gen_frame_store): Likewise.
30280 (emit_outlined_ms2sysv_save): Likewise.
30281 (emit_outlined_ms2sysv_restore): Likewise.
30282 (ix86_expand_prologue): Modify stack re-alignment code and call
30283 emit_outlined_ms2sysv_save when appropriate.
30284 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30285 parameter rtx_insn *insn, which allows the function to be used to only
30286 generate the notes.
30287 (ix86_expand_epilogue): Modify validity checks of frame and stack
30288 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30289 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30290 * config/i386/predicates.md
30291 (save_multiple): New predicate.
30292 (restore_multiple): Likewise.
30293 * config/i386/sse.md
30294 (save_multiple<mode>): New pattern.
30295 (save_multiple_realign<mode>): Likewise.
30296 (restore_multiple<mode>): Likewise.
30297 (restore_multiple_and_return<mode>): Likewise.
30298 (restore_multiple_leave_return<mode>): Likewise.
30299 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30301 2017-05-14 Julia Koval <julia.koval@intel.com>
30303 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30304 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30305 (__builtin_ia32_xsetbv): New builtins.
30306 * config/i386/i386.c (ix86_expand_special_args_builtin):
30308 (ix86_expand_builtin): Special expand for new intrinsics.
30309 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30310 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30311 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30313 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30315 * cfganal.c (inverted_post_order_compute): Change argument type
30317 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30318 * df-core.c (rest_of_handle_df_initialize): Adjust.
30319 (rest_of_handle_df_finish): Likewise.
30320 (df_analyze_1): Likewise.
30321 (df_analyze): Likewise.
30322 (loop_inverted_post_order_compute): Change argument to be a vec *.
30323 (df_analyze_loop): Adjust.
30324 (df_get_n_blocks): Likewise.
30325 (df_get_postorder): Likewise.
30326 * df.h (struct df_d): Change field to be a vec.
30327 * lcm.c (compute_laterin): Adjust.
30328 (compute_available): Likewise.
30329 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30330 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30331 * tree-ssa-pre.c (compute_antic): Likewise.
30333 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30335 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30336 (depth_first_search::depth_first_search): Change structure init
30337 function to this constructor.
30338 (depth_first_search::add_bb): Rename function to this member.
30339 (depth_first_search::execute): Likewise.
30340 (flow_dfs_compute_reverse_finish): Adjust.
30342 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30344 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30345 (longest_simple_path): Likewise.
30346 * shrink-wrap.c (spread_components): Likewise.
30347 (disqualify_problematic_components): Likewise.
30348 (emit_common_heads_for_components): Likewise.
30349 (emit_common_tails_for_components): Likewise.
30350 (insert_prologue_epilogue_for_components): Likewise.
30352 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30354 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30357 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30359 * df-core.c (df_set_blocks): Start using auto_bitmap.
30360 (df_compact_blocks): Likewise.
30361 * df-problems.c (df_rd_confluence_n): Likewise.
30362 * df-scan.c (df_insn_rescan_all): Likewise.
30363 (df_process_deferred_rescans): Likewise.
30364 (df_update_entry_block_defs): Likewise.
30365 (df_update_exit_block_uses): Likewise.
30366 (df_entry_block_bitmap_verify): Likewise.
30367 (df_exit_block_bitmap_verify): Likewise.
30368 (df_scan_verify): Likewise.
30369 * lra-constraints.c (lra_constraints): Likewise.
30370 (undo_optional_reloads): Likewise.
30371 (lra_undo_inheritance): Likewise.
30372 * lra-remat.c (calculate_gen_cands): Likewise.
30373 (do_remat): Likewise.
30374 * lra-spills.c (assign_spill_hard_regs): Likewise.
30375 (spill_pseudos): Likewise.
30376 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30377 (bitmap_set_subtract_values): Likewise.
30379 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30381 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30382 management with auto_bitmap.
30383 (fix_inter_tick): Likewise.
30384 (fix_recovery_deps): Likewise.
30385 * ira.c (add_store_equivs): Likewise.
30386 (find_moveable_pseudos): Likewise.
30387 (split_live_ranges_for_shrink_wrap): Likewise.
30388 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30389 (rtx_reuse_manager::seen_def_p): Likewise.
30390 (rtx_reuse_manager::set_seen_def): Likewise.
30391 * print-rtl.h (class rtx_reuse_manager): Likewise.
30393 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30395 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30397 (migrate_btr_def): Likewise.
30398 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30399 * df-core.c (loop_post_order_compute): Likewise.
30400 (loop_inverted_post_order_compute): Likewise.
30401 * hsa-common.h: Likewise.
30402 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30403 * init-regs.c (initialize_uninitialized_regs): Likewise.
30404 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30405 (inline_small_functions): Likewise.
30406 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30407 * ira.c (combine_and_move_insns): Likewise.
30408 (build_insn_chain): Likewise.
30409 * loop-invariant.c (find_invariants): Likewise.
30410 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30411 * predict.c (tree_predict_by_opcode): Likewise.
30412 (predict_paths_leading_to): Likewise.
30413 (predict_paths_leading_to_edge): Likewise.
30414 (estimate_loops_at_level): Likewise.
30415 (estimate_loops): Likewise.
30416 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30417 (spread_components): Likewise.
30418 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30419 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30420 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30421 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30422 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30423 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30424 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30425 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30426 (mark_threaded_blocks): Likewise.
30427 (thread_through_all_blocks): Likewise.
30428 * tree-ssa.c (verify_ssa): Likewise.
30429 (execute_update_addresses_taken): Likewise.
30430 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30432 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30434 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30436 (post_order_compute): Likewise.
30437 (inverted_post_order_compute): Likewise.
30438 (pre_and_rev_post_order_compute_fn): Likewise.
30440 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30442 * genrecog.c (int_set::int_set): Explicitly construct our
30443 auto_vec base class.
30444 * vec.h (auto_vec::auto_vec): New constructor.
30446 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30448 * bitmap.h (class auto_bitmap): New constructor taking
30449 bitmap_obstack * argument.
30451 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30453 * bitmap.h (class auto_bitmap): Change type of m_bits to
30454 bitmap_head, and adjust ctor / dtor and member operators.
30456 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30458 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30459 when returned register mode doesn't match original mode.
30461 2017-05-12 Jeff Law <law@redhat.com>
30462 Jakub Jelinek <jakub@redhat.com>
30464 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30465 we look for cc setter after the compare-elim changes.
30466 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30467 within the vector to match what compare-elim now expects.
30468 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30469 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30471 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30472 after the compare-elim changes.
30473 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30474 the vector to match what compare-elim now expects.
30475 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30476 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30477 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30478 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30479 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30481 * config/visium/visium.c (single_set_and_flags): Fix where
30482 we look for cc setter after the compare-elim changes.
30483 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30484 with the vector to match what compare-elim now expects.
30485 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30486 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30487 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30488 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30489 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30490 (neg<mode>2_insn_set_overflow): Likewise.
30492 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30494 PR middle-end/79794
30495 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30496 maybe_expand_insn call, set ops[0].target. If still set after call,
30497 set alt_rtl. Add extra arg to recursive calls.
30498 (extract_bit_field): Add alt_rtl argument. Pass to
30500 * expmed.h (extract_bit_field): Fix prototype.
30501 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30502 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30503 to extract_bit_field_calls.
30504 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30505 Pass alt_rtl to extract_bit_field calls.
30506 * calls.c (store_unaligned_arguments_into_psuedos)
30507 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30508 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30510 * optabs.h (struct expand_operand): Add target bitfield.
30512 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30514 * compare-elim.c (try_eliminate_compare): Canonicalize
30515 operation with embedded compare to
30516 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30517 (set (reg) (operation)].
30519 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30521 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30524 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30525 cost of adding a carry flag for ADC instruction.
30526 [case MINUS]: Ignore the cost of subtracting a carry flag
30527 for SBB instruction.
30529 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30531 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30533 * config/rs6000/bmiintrin.h: New file.
30534 * config/rs6000/bmi2intrin.h: New file.
30535 * config/rs6000/x86intrin.h: New file.
30537 2017-05-12 Jeff Law <law@redhat.com>
30539 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30542 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30544 PR middle-end/80707
30545 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30547 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30549 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30550 early expansion of vector divide builtins.
30551 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30552 builtins identified as having unsigned arguments.
30554 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30556 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30557 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30558 expansion of vector logical operations (and, andc, or, xor,
30561 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30563 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30564 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30566 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30568 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30569 early GIMPLE expansion of vector multiplies.
30571 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30573 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30574 TARGET_HAVE_MOVT conditional.
30575 (movt splitter): Likewise.
30577 2017-05-12 Richard Biener <rguenther@suse.de>
30579 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30580 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30581 Fold all stmts not inplace.
30583 2017-05-12 Richard Biener <rguenther@suse.de>
30585 PR tree-optimization/80713
30586 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30587 inserted_exprs bit for not removed stmts.
30589 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30591 PR middle-end/69921
30592 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30593 parallelized" attribute for parallelized OpenACC kernels.
30594 * omp-offload.c (execute_oacc_device_lower): Use it.
30596 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30597 Set "oacc kernels" attribute.
30598 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30599 parameter. Adjust all users.
30600 (oacc_fn_attrib_kernels_p): Remove function.
30601 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30602 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30603 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30604 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30605 assert "oacc kernels" attribute is set.
30607 2017-05-11 Carl Love <cel@us.ibm.com>
30609 * config/rs6000/rs6000-c: Add support for built-in functions
30610 vector unsigned char vec_popcnt (vector signed char)
30611 vector unsigned char vec_popcnt (vector unsigned char)
30612 vector unsigned short vec_popcnt (vector signed short)
30613 vector unsigned short vec_popcnt (vector unsigned short)
30614 vector unsigned int vec_popcnt (vector signed int)
30615 vector unsigned int vec_popcnt (vector unsigned int)
30616 vector unsigned long long vec_popcnt (vector signed long long)
30617 vector unsigned long long vec_popcnt (vector unsigned long long)
30618 vector signed long long vec_slo (vector signed long long,
30619 vector signed char)
30620 vector signed long long vec_slo (vector signed long long,
30621 vector unsigned char)
30622 vector unsigned long long vec_slo (vector unsigned long long,
30623 vector signed char)
30624 vector unsigned long long vec_slo (vector unsigned long long,
30625 vector unsigned char)
30626 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30627 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30628 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30629 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30630 * doc/extend.texi: Update the built-in documentation file for the
30631 new built-in functions.
30633 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30635 * attribs.h (sorted_attr_string): Move machine independent
30636 functions for target clone support from the i386 port to common
30637 code. Rename ix86_function_versions to common_function_versions.
30638 Rename make_name to make_unique_name.
30639 (common_function_versions): Likewise.
30640 (make_unique_name): Likewise.
30641 (make_dispatcher_decl): Likewise.
30642 (is_function_default_version): Likewise.
30643 * attribs.c (attr_strcmp): Likewise.
30644 (sorted_attr_string): Likewise.
30645 (common_function_versions): Likewise.
30646 (make_unique_name): Likewise.
30647 (make_dispatcher_decl): Likewise.
30648 (is_function_default_version): Likewise.
30649 * config/i386/i386.c (attr_strcmp): Likewise.
30650 (sorted_attr_string): Likewise.
30651 (ix86_function_versions): Likewise.
30652 (make_name): Likewise.
30653 (make_dispatcher_decl): Likewise.
30654 (is_function_default_version): Likewise.
30655 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30657 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30660 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30661 Account for direct move costs for vec_construct of integer
30664 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30667 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30668 (UNSPEC_STX_ATOMIC): Ditto.
30669 (loaddi_via_sse): New insn.
30670 (storedi_via_sse): Ditto.
30671 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30672 Update corresponding peephole2 patterns.
30673 (atomic_storedi_fpu): Ditto.
30675 2017-05-11 Julia Koval <julia.koval@intel.com>
30677 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30678 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30680 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30681 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30682 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30684 2017-05-11 Nathan Sidwell <nathan@acm.org>
30686 * graphite-poly.c: Include dumpfile.h.
30688 * dumpfle.h (dump_function): Declare here ...
30689 * tree-dump.h (dump_function): ... not here.
30690 * dumpfile.c: #include tree-cfg.h.
30691 (dump_function): Move here from ...
30692 * tree-dump.c (dump_function): ... here.
30693 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30694 * graphite-poly.c: Don't include tree-dump.h.
30695 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30696 * print-tree.c: Likewise.
30697 * stor-layout.c: Likewise.
30698 * tree-nested.c: Likewise.
30700 * dumpfile.c (dump_start): Use TDF_FLAGS.
30701 (dump_enable_all): Fix TDF_KIND check thinko.
30703 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30705 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30706 array entries to represent two legal parameterizations of the
30707 overloaded __builtin_cmpb function, as represented by the
30708 P6_OV_BUILTIN_CMPB constant.
30709 (altivec_resolve_overloaded_builtin): Add special case handling
30710 for the __builtin_cmpb function, as represented by the
30711 P6_OV_BUILTIN_CMPB constant.
30712 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30713 (BU_P6_64BIT_2): New macro.
30714 (BU_P6_OVERLOAD_2): New macro
30715 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30716 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30717 (CMPB): Add overload support to represent both 32-bit and 64-bit
30718 compare-bytes function.
30719 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30720 support for TARGET_CMPB.
30721 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30722 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30723 documentation of the __builtin_cmpb overloaded built-in function.
30725 2017-05-11 Richard Biener <rguenther@suse.de>
30727 PR tree-optimization/80705
30728 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30729 bases are not vectorizable.
30731 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30733 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30734 when counting register pressure.
30736 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30738 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30739 (iv_ca_more_deps): Renamed to ...
30740 (iv_ca_compare_deps): ... this.
30741 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30743 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30745 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30747 (determine_group_iv_costs): ... here.
30748 (find_inv_vars_cb): Record inv var if it's not recorded before.
30750 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30752 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30753 (get_shiftadd_cost): Ditto.
30755 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30757 * tree-ssa-address.c: Include header file.
30758 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30760 (add_to_parts): Refactor.
30761 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30762 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30765 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30767 PR tree-optimization/53090
30768 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30770 (extract_cond_operands): Detect condition with IV on both sides
30771 and return COMP_IV_EXPR_2.
30772 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30773 (rewrite_use_compare): Simplify by removing call to function
30774 extract_cond_operands.
30776 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30778 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30779 (extract_cond_operands): Detect condition comparing against non-
30780 invariant bound and return appropriate enum value.
30781 (find_interesting_uses_cond): Update use of extract_cond_operands.
30782 Handle its return value accordingly.
30783 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30785 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30787 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30788 nonlinear iv_use computation in loop invariant sensitive way.
30790 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30792 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30793 (find_iv_candidates): Call relate_compare_use_with_all_cands.
30795 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30797 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
30798 (dump_cand): Support iv_cand.inv_exprs.
30799 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
30801 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
30804 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30806 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
30808 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
30809 as local function. Include necessary header files.
30810 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
30812 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30814 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
30816 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30818 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
30819 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
30820 RSHIFT_EXPR and BIT_NOT_EXPR.
30822 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30824 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
30825 (adjust_setup_cost): New parameter supporting round up adjustment.
30826 (struct address_cost_data): Delete.
30827 (force_expr_to_var_cost): Don't bound cost with spill_cost.
30828 (split_address_cost, ptr_difference_cost): Delete.
30829 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
30830 (struct ainc_cost_data): New struct.
30831 (get_address_cost_ainc): New function.
30832 (get_address_cost, get_computation_cost): Reimplement.
30833 (determine_group_iv_cost_address): Record inv_expr for all uses of
30835 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
30836 (iv_ca_has_deps): Reimplemented to ...
30837 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
30839 (iv_ca_extend): Call iv_ca_more_deps.
30841 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30843 * tree-ssa-address.c (struct mem_address): Move to header file.
30844 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
30845 * tree-ssa-address.h (struct mem_address): Move from C file.
30846 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
30848 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30850 * tree-affine.h (aff_combination_type): New interface.
30851 (aff_combination_zero_p): Remove static.
30852 (aff_combination_const_p): New interface.
30853 (aff_combination_singleton_var_p): New interfaces.
30855 2017-05-11 Richard Biener <rguenther@suse.de>
30857 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30858 Skip unreachable blocks and destinations.
30859 (eliminate): Move stmt removal and fixup ...
30860 (fini_eliminate): ... here. Skip inserted exprs.
30861 (pass_pre::execute): Move fini_pre after fini_eliminate.
30862 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
30863 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
30864 PRE to get rid of dead code that has invalid SSA form and
30865 split critical edges again.
30867 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30869 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30871 2017-05-11 Richard Biener <rguenther@suse.de>
30873 * passes.c (execute_function_todo): Verify loops if they are
30874 said to be up-to-date.
30875 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
30876 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
30878 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
30881 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
30882 handle calling assemble_external ourself.
30885 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
30886 modes with zero size. Enhance comment.
30888 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30890 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
30891 built-ins for vec_xl and vec_xst with short and char pointer
30894 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30896 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
30897 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
30898 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
30899 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
30900 (_mm_maskz_min_round_ss): New intrinsics.
30901 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
30902 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
30903 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
30904 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
30905 (__builtin_ia32_minss_mask_round): New builtins.
30906 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30907 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30908 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
30910 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
30911 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
30913 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
30916 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30918 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
30919 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
30920 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
30921 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
30922 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
30923 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
30924 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
30925 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30926 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30927 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
30928 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
30929 (__builtin_ia32_mulss_mask_round): New builtins.
30930 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30931 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30932 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
30934 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
30935 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30937 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30940 2017-05-10 Julia Koval <julia.koval@intel.com>
30942 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
30943 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
30944 (_mm256_setr_m128i): New intrinsics.
30946 2017-05-10 Julia Koval <julia.koval@intel.com>
30948 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
30949 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
30950 (_mm_maskz_rcp14_ss): New intrinsics.
30951 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
30952 (__builtin_ia32_rcp14ss_mask): New builtins.
30953 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
30955 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
30957 PR tree-optimization/51513
30958 * tree-cfg.c (gimple_seq_unreachable_p): New function.
30959 (assert_unreachable_fallthru_edge_p): Use it.
30960 (group_case_labels_stmt): Likewise.
30961 * tree-cfg.h: Prototype it.
30962 * stmt.c: Include cfghooks.h and tree-cfg.h.
30963 (emit_case_dispatch_table) <gap_label>: New local variable.
30964 Use it to fill dispatch table gaps.
30965 Test for default_label before updating probabilities.
30966 (expand_case) <default_label>: Remove unneeded initialization.
30967 Test for unreachable default case statement and remove its edge.
30968 Set default_label accordingly.
30969 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
30971 2017-05-10 Carl Love <cel@us.ibm.com>
30973 * config/rs6000/rs6000-c: Add support for built-in functions
30974 vector signed char vec_neg (vector signed char)
30975 vector signed short int vec_neg (vector short int)
30976 vector signed int vec_neg (vector signed int)
30977 vector signed long long vec_neg (vector signed long long)
30978 vector float vec_neg (vector float)
30979 vector double vec_neg (vector double)
30980 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
30982 * config/rs6000/altivec.h: Add define for vec_neg
30983 * doc/extend.texi: Update the built-in documentation for the
30984 new built-in functions.
30986 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30988 PR tree-optimization/77644
30989 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
30991 2017-05-10 Nathan Sidwell <nathan@acm.org>
30993 * dumpfile.h (TDI_lang_all): New.
30994 (TDF_KIND): New. Renumber others
30995 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
30997 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
30999 (get_dump_file_name): Adjust suffix generation.
31000 (dump_enable_all): Use TDF_KIND.
31001 * doc/invoke.texi (-fdump-lang-all): Document.
31003 * dumpfile.h: Tabify.
31005 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31008 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31009 Move member access before delete.
31011 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31013 * tree-inline.c (expand_call_inline): Split block at stmt
31016 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31019 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31020 are now unused after splitting mov{sf,sd}_hardfloat.
31021 (f32_lr2): Likewise.
31022 (f32_lm): Likewise.
31023 (f32_lm2): Likewise.
31024 (f32_li): Likewise.
31025 (f32_li2): Likewise.
31026 (f32_lv): Likewise.
31027 (f32_sr): Likewise.
31028 (f32_sr2): Likewise.
31029 (f32_sm): Likewise.
31030 (f32_sm2): Likewise.
31031 (f32_si): Likewise.
31032 (f32_si2): Likewise.
31033 (f32_sv): Likewise.
31034 (f32_dm): Likewise.
31035 (f32_vsx): Likewise.
31036 (f32_av): Likewise.
31037 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31038 For movsf, order stores so the VSX stores occur before the GPR
31039 store which encourages the register allocator to use a traditional
31040 FPR instead of a GPR. For movsd, order the stores so that the GPR
31041 store comes before the VSX stores to allow the power6 to work.
31042 This is due to the power6 not having a 32-bit integer store
31043 instruction from a FPR.
31044 (movsf_hardfloat): Likewise.
31045 (movsd_hardfloat): Likewise.
31047 2017-05-09 Martin Sebor <msebor@redhat.com>
31049 PR translation/80280
31050 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31053 PR translation/80280
31054 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31055 data member added in r247778.
31056 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31058 2017-05-09 Nathan Sidwell <nathan@acm.org>
31060 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31062 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31065 2017-05-09 Marek Polacek <polacek@redhat.com>
31067 * doc/invoke.texi: Fix typo.
31069 2017-05-09 Richard Biener <rguenther@suse.de>
31071 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31072 (vrp_val_is_min): Likewise.
31073 (set_value_range_to_value): Likewise.
31074 (set_value_range_to_nonnegative): Likewise.
31075 (gimple_assign_nonzero_p): Likewise.
31076 (gimple_stmt_nonzero_p): Likewise.
31077 (vrp_int_const_binop): Likewise. Remove unreachable case.
31078 (adjust_range_with_scev): Adjust comments.
31079 (compare_range_with_value): Likewise.
31080 (extract_range_from_phi_node): Likewise.
31081 (test_for_singularity): Likewise.
31083 2017-05-09 Richard Biener <rguenther@suse.de>
31085 * tree-vrp.c (get_single_symbol): Add assert that we don't
31086 get overflowed constants as invariant part.
31087 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31088 checks. Use wi::cmp instead of recursing for integer constants.
31089 (compare_values): Just ignore whether we assumed undefined
31090 overflow instead of failing the compare.
31091 (extract_range_for_var_from_comparison_expr): Add comment before the
31092 TREE_NO_WARNING sets.
31093 (test_for_singularity): Likewise.
31094 (extract_range_from_comparison): Do not disable optimization
31095 when we assumed undefined overflow.
31096 (extract_range_basic): Remove init of unused var.
31098 2017-05-09 Richard Biener <rguenther@suse.de>
31100 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31101 (extract_range_from_multiplicative_op_1): Adjust.
31102 (extract_range_from_binary_expr_1): Use int_const_binop.
31104 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31107 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31108 rs6000_store_data_bypass_p in seven define_bypass directives and
31109 in several comments.
31110 * config/rs6000/rs6000-protos.h: Add prototype for
31111 rs6000_store_data_bypass_p function.
31112 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31113 function implements slightly different (rs6000-specific) semantics
31114 than store_data_bypass_p, returning false rather than aborting
31115 with assertion error when arguments do not satisfy the
31116 requirements of store data bypass.
31117 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31118 rs6000_store_data_bypass_p.
31120 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31122 * config/xtensa/xtensa-protos.h
31123 (xtensa_initial_elimination_offset): New declaration.
31124 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31125 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31126 macro definition, add case for FRAME_POINTER_REGNUM when
31127 FRAME_GROWS_DOWNWARD.
31128 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31129 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31130 xtensa_initial_elimination_offset.
31132 2017-05-08 Nathan Sidwell <nathan@acm.org>
31134 * doc/invoke.texi: Alphabetize -fdump options.
31136 2017-05-08 Martin Sebor <msebor@redhat.com>
31138 PR translation/80280
31139 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31141 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31143 * target.def (compute_frame_layout): New optional target hook.
31144 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31145 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31146 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31148 * reload1.c (verify_initial_elim_offsets): Likewise.
31149 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31150 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31151 (arm_get_frame_offsets): Split up into this ...
31152 (arm_compute_frame_layout): ... and this function.
31154 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31156 * config/aarch64/constraints.md (Usa): New constraint.
31157 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31159 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31161 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31162 with_multilib_list after it has been checked.
31164 2017-05-08 Richard Biener <rguenther@suse.de>
31166 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31167 (bitmap_set_subtract_values): Likewise.
31169 2017-05-08 Richard Biener <rguenther@suse.de>
31171 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31172 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31174 (gimple_stmt_nonzero_warnv_p): Rename to ...
31175 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31177 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31178 (extract_range_basic): Adjust, do not disable propagation on
31179 strict overflow sensitive simplification.
31180 (vrp_visit_cond_stmt): Likewise.
31182 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31184 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31185 body size unconditionally.
31187 2017-05-07 Jeff Law <law@redhat.com>
31190 2017-05-06 Jeff Law <law@redhat.com>
31191 PR tree-optimization/78496
31192 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31195 PR tree-optimization/78496
31196 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31197 (simplify_stmt_using_ranges): Call it.
31198 (vrp_dom_walker::before_dom_children): Extract equivalences
31199 from an ASSERT_EXPR with an equality comparison against a
31202 2017-05-06 Jeff Law <law@redhat.com>
31204 PR tree-optimization/78496
31205 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31208 PR tree-optimization/78496
31209 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31210 (simplify_stmt_using_ranges): Call it.
31211 (vrp_dom_walker::before_dom_children): Extract equivalences
31212 from an ASSERT_EXPR with an equality comparison against a
31215 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31217 * lra-constraints.c (lra_copy_reg_equiv): New function.
31218 (split_reg): Use it to copy equivalence information from the
31219 original register to the spill register.
31221 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31223 PR rtl-optimization/75964
31224 * simplify-rtx.c (simplify_const_relational_operation): Remove
31225 invalid handling of comparisons of integer ABS.
31227 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31229 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31230 initialize to zero.
31231 (init_regs): Remove declaration.
31232 (function_arg_advance_32): Initialize error_p as boolean variable.
31234 2017-05-05 Nathan Sidwell <nathan@acm.org>
31236 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31237 lines. Use for (;;).
31239 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31241 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31242 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31243 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31244 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31245 VF=2 that require versioning.
31247 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31249 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31252 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31254 * diagnostic.h (diagnostic_override_option_index): Convert from
31255 macro to inline function.
31257 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31259 * diagnostic.c (last_module_changed_p): New function.
31260 (set_last_module): New function.
31261 (diagnostic_report_current_module): Convert macro usage to
31262 the above functions.
31263 * diagnostic.h (diagnostic_context::last_module): Strengthen
31264 from const line_map * to const line_map_ordinary *.
31265 (diagnostic_last_module_changed): Delete macro.
31266 (diagnostic_set_last_module): Delete macro.
31268 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31270 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31271 with diagnostic_report_diagnostic.
31272 (diagnostic_n_impl_richloc): Likewise.
31273 * diagnostic.h (report_diagnostic): Delete macro.
31274 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31275 with diagnostic_report_diagnostic.
31276 * substring-locations.c (format_warning_va): Likewise.
31278 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31280 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31281 save/restor of format_spec. Move option-printing code to...
31282 (print_option_information): ...this new function, and
31283 reimplement by simply printing to the pretty_printer,
31284 rather than appending to the format string.
31286 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31288 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31289 handling logic into...
31290 (update_effective_level_from_pragmas): ...this new function.
31292 2017-05-04 Andrew Waterman <andrew@sifive.com>
31294 * config/riscv/riscv.opt (mstrict-align): New option.
31295 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31296 (SLOW_UNALIGNED_ACCESS): Define.
31297 (riscv_slow_unaligned_access): Declare.
31298 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31300 (riscv_slow_unaligned_access): New variable.
31301 (rocket_tune_info): Set slow_unaligned_access to true.
31302 (optimize_size_tune_info): Set slow_unaligned_access to false.
31303 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31304 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31305 (riscv_option_override): Set riscv_slow_unaligned_access.
31306 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31308 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31310 * config/riscv/riscv.md: Unify indentation.
31312 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31317 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31319 (extendsi<mode>2): Add support for doing sign extension via
31320 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31321 don't have ISA 3.0 instructions.
31322 (extendsi<mode>2 splitter): Likewise.
31323 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31324 generate the normal insns since SImode can now go in vector
31325 registers. Disallow the special UNSPECs needed for previous
31326 machines to hide SImode being used. Add new insns
31327 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31328 (fix_trunc<mode>si2_stfiwx): Likewise.
31329 (fix_trunc<mode>si2_internal): Likewise.
31330 (fixuns_trunc<mode>si2): Likewise.
31331 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31332 (fctiw<u>z_<mode>_smallint): Likewise.
31333 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31334 of floating point to 32-bit integer from doing a direct move to
31335 the GPR registers to do a store.
31336 (fctiwz_<mode>): Break long line.
31338 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31340 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31341 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31342 (addr_list, addr_offset_valid_p): New.
31343 (split_address_groups): Check offset validity with above function.
31344 (gt-tree-ssa-loop-ivopts.h): Include header file.
31346 2017-05-05 Nathan Sidwell <nathan@acm.org>
31348 * config.gcc (arm*-*-*): Add missing 'fi'.
31350 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31352 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31353 included in -fopt-info does not matter.
31354 * doc/optinfo.texi (-fopt-info): Fix description of default
31355 behavour. Explicitly say order of options included in -fopt-info
31358 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31360 * config.gcc: Allow combinations of aprofile and rmprofile values for
31361 --with-multilib-list.
31362 * config/arm/t-multilib: New file.
31363 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31364 variables. Remove setting of ISA and floating-point ABI in
31365 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31366 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31367 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31368 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31370 * config/arm/t-rmprofile: Likewise except for the matches changes.
31371 * doc/install.texi (--with-multilib-list): Document the combination of
31372 aprofile and rmprofile values and warn about pitfalls in doing that.
31374 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31376 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31377 (movdi_aarch64): Likewise.
31379 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31381 PR tree-optimization/80632
31382 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31384 (build_arrays): Initialize it for virtual phis.
31385 (fix_phi_nodes): Use it for virtual phis.
31387 PR tree-optimization/80558
31388 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31389 [x, y] op z into [x op, y op z] for op & or | if conditions
31392 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31393 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31396 * config/arm/arm.md (use_literal_pool): Remove.
31397 (64-bit immediate split): No longer takes cost into consideration
31398 if arm_disable_literal_pool is enabled.
31399 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31400 used when arm_disable_literal_pool is enabled.
31401 (arm_max_const_double_inline_cost): Remove use of
31402 arm_disable_literal_pool.
31403 (push_minipool_fix): Add assert.
31404 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31405 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31406 (no_literal_pool_sf_immediate): New.
31408 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31410 PR tree-optimization/80613
31411 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31412 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31414 2017-05-05 Richard Biener <rguenther@suse.de>
31416 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31418 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31420 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31421 of this flag from insn conditions due to removal from r247495.
31423 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31425 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31427 (arm_early_store_addr_dep_ptr): Likewise.
31428 * config/arm/aarch-common-protos.h
31429 (arm_early_load_addr_dep_ptr): Add prototype.
31430 (arm_early_store_addr_dep_ptr): Likewise.
31431 * config/arm/cortex-a53.md: Add new bypasses.
31433 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31435 * tree.c (next_type_uid): Change type to unsigned.
31436 (type_hash_canon): Decrement back next_type_uid if
31437 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31438 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31441 2017-05-04 Martin Sebor <msebor@redhat.com>
31443 * builtins.c: Fix a trivial typo in a comment.
31445 PR middle-end/79234
31446 * builtins.c (check_sizes): Adjust to handle reading past the end.
31447 Avoid printing excessive upper bound of ranges. Use %E to print
31448 tree nodes instead of converting them to %wu.
31449 (expand_builtin_memchr): New function.
31450 (compute_dest_size): Rename...
31451 (compute_objsize): ...to this.
31452 (expand_builtin_memcpy): Adjust.
31453 (expand_builtin_mempcpy): Adjust.
31454 (expand_builtin_strcat): Adjust.
31455 (expand_builtin_strcpy): Adjust.
31456 (check_strncat_sizes): Adjust.
31457 (expand_builtin_strncat): Adjust.
31458 (expand_builtin_strncpy): Adjust and simplify.
31459 (expand_builtin_memset): Adjust.
31460 (expand_builtin_bzero): Adjust.
31461 (expand_builtin_memcmp): Adjust.
31462 (expand_builtin): Handle memcmp.
31463 (maybe_emit_chk_warning): Check strncat just once.
31465 2017-05-04 Martin Sebor <msebor@redhat.com>
31467 PR preprocessor/79214
31468 PR middle-end/79222
31469 PR middle-end/79223
31470 * builtins.c (check_sizes): Add inlining context and issue
31471 warnings even when -Wno-system-headers is set.
31472 (check_strncat_sizes): Same.
31473 (expand_builtin_strncat): Same.
31474 (expand_builtin_memmove): New function.
31475 (expand_builtin_stpncpy): Same.
31476 (expand_builtin): Handle memmove and stpncpy.
31478 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31480 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31481 which is not used any more.
31483 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31485 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31487 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31489 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31490 (cortexa53_tunings): Likewise.
31491 (cortexa57_tunings): Likewise.
31492 (cortexa72_tunings): Likewise.
31493 (cortexa73_tunings): Likewise.
31495 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31497 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31498 Set loop alignment to 8.
31500 2017-05-04 Martin Sebor <msebor@redhat.com>
31502 PR translation/80280
31503 * builtins.c (expand_builtin_object_size): Add missing quoting to
31504 %D and like directives.
31505 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31506 (hsa_type_for_tree_type): Same.
31507 (verify_function_arguments): Same.
31508 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31509 * varasm.c (get_section): Same.
31512 2017-05-04 Martin Sebor <msebor@redhat.com>
31514 PR translation/80280
31515 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31517 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31519 * config/aarch64/aarch64.c (generic_addrcost_table):
31520 Change HI/TI mode setting.
31522 2017-05-04 Martin Jambor <mjambor@suse.cz>
31524 PR tree-optimization/80622
31525 * tree-sra.c (comes_initialized_p): New function.
31526 (build_accesses_from_assign): Only set write lazily when
31527 comes_initialized_p is false.
31528 (analyze_access_subtree): Use comes_initialized_p.
31529 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31530 instead of testing for PARM_DECL.
31532 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31534 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31535 constraint on operand 0 to allow more general addressing modes.
31536 Adjust output template.
31537 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31539 * config/aarch64/aarch64-protos.h
31540 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31541 * config/aarch64/constraints.md (Dp): New address constraint.
31542 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31545 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31547 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31548 update use of estimate_ipcp_clone_size_and_time.
31549 (estimate_local_effects): Update use of
31550 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31551 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31552 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31553 Return nonspecialized time.
31555 2017-05-04 Richard Biener <rguenther@suse.de>
31557 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31558 for the last VUSE which def dominates the PHI. Directly call
31560 (get_continuation_for_phi_1): Remove.
31562 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31564 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31565 to explain the use of truncating division. Cap the number of
31566 iterations to the maximum given by nb_iterations_upper_bound,
31569 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31571 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31572 * configure: Regenerate.
31573 * config.in: Regenerate.
31574 * config/i386/driver-mingw32.c: new file.
31575 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31576 * config.host: Link driver-mingw32.o on MinGW host.
31577 * doc/install.texi: Document new --enable-mingw-wildcard configure
31580 2017-05-04 Marek Polacek <polacek@redhat.com>
31582 PR tree-optimization/80612
31583 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31585 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31586 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31588 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31589 (movt splitter): Likewise.
31590 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31591 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31592 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31593 block for Thumb-1 with MOVT.
31594 (thumb2_legitimate_address_p): Move code block ...
31595 (can_avoid_literal_pool_for_label_p): ... into this new function.
31596 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31598 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31599 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31600 "M-profile targets with the MOVT instruction".
31602 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31604 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31605 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31606 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31608 2017-05-04 Martin Liska <mliska@suse.cz>
31610 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31611 variable cond_code.
31613 2017-05-04 Richard Biener <rguenther@suse.de>
31615 * tree.c (array_at_struct_end_p): Handle arrays at struct
31616 end with flexarrays more conservatively. Refactor and treat
31617 arrays of arrays or aggregates more strict. Fix
31618 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31619 * tree.c (array_at_struct_end_p): Adjust prototype.
31620 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31621 * gimple-fold.c (get_range_strlen): Likewise.
31622 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31624 2017-05-04 Richard Biener <rguenther@suse.de>
31626 PR tree-optimization/31130
31627 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31629 (supports_overflow_infinity): Likewise.
31630 (is_negative_overflow_infinity): Likewise.
31631 (is_positive_overflow_infinity): Likewise.
31632 (is_overflow_infinity): Likewise.
31633 (stmt_overflow_infinity): Likewise.
31634 (overflow_infinity_range_p): Likewise.
31635 (usable_range_p): Remove as always returning true.
31636 (make_overflow_infinity): Remove.
31637 (negative_overflow_infinity): Likewise.
31638 (positive_overflow_infinity): Likewise.
31639 (avoid_overflow_infinity): Likewise.
31640 (set_value_range): Adjust accordingly.
31641 (set_value_range_to_nonnegative): Likewise, remove now unused
31642 overflow_infinity arg.
31643 (vrp_operand_equal_p): Adjust.
31644 (update_value_range): Likewise.
31645 (range_int_cst_singleton_p): Likewise.
31646 (operand_less_p): Likewise.
31647 (compare_values_warnv): Likewise.
31648 (extract_range_for_var_from_comparison_expr): Likewise.
31649 (vrp_int_const_binop): Likewise.
31650 (zero_nonzero_bits_from_vr): Likewise.
31651 (extract_range_from_multiplicative_op_1): Likewise.
31652 (extract_range_from_binary_expr_1): Likewise.
31653 (extract_range_from_unary_expr): Likewise.
31654 (extract_range_from_comparison): Likewise.
31655 (extract_range_basic): Likewise.
31656 (adjust_range_with_scev): Likewise.
31657 (compare_ranges): Likewise.
31658 (compare_range_with_value): Likewise.
31659 (dump_value_range): Likewise.
31660 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31662 (simplify_cond_using_ranges): Adjust.
31664 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31666 * brig-builtins.def: Added a builtin for class_f64.
31667 * builtin-types.def: Added a builtin type needed by class_f64.
31669 2017-05-03 Jason Merrill <jason@redhat.com>
31671 * timevar.def: Add TV_CONSTEXPR.
31673 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31675 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31677 2017-05-03 Martin Jambor <mjambor@suse.cz>
31679 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31680 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31681 * ipa-cp.c (ipcp_propagate_stage): Do not call
31682 ipa_update_after_lto_read.
31683 * ipa-inline.c (ipa_inline): Likewise.
31685 2017-05-03 Martin Jambor <mjambor@suse.cz>
31687 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31688 tag. Added a default constructor and a destructor.
31689 (ipa_edge_args_sum_t): New class;
31690 (ipa_edge_args_sum): Declare.
31691 (ipa_edge_args_vector): Remove declaration.
31692 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31693 (ipa_free_edge_args_substructures): Remove declaration.
31694 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31695 (ipa_edge_args_info_available_for_edge_p): Likewise.
31696 * ipa-prop.c (ipa_edge_args_vector): Removed.
31697 (edge_removal_hook_holder): Likewise.
31698 (edge_duplication_hook_holder): Likewise.
31699 (ipa_edge_args_sum): New variable.
31700 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31701 ipa_edge_args_vector.
31702 (ipa_free_edge_args_substructures): Likewise.
31703 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31704 ipa_edge_args_vector.
31705 (ipa_edge_removal_hook): Turned into method
31706 ipa_edge_args_sum_t::remove.
31707 (ipa_edge_duplication_hook): Turned into method
31708 ipa_edge_args_sum_t::duplicate.
31709 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31710 registering edge hooks.
31711 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31712 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31713 ipa_edge_args_sum instead of ipa_edge_args_vector.
31714 * ipa-profile.c (ipa_profile): Likewise.
31716 2017-05-03 Martin Jambor <mjambor@suse.cz>
31718 * symbol-summary.h (function_summary): New method exists.
31719 (function_summary::symtab_removal): Deallocate through release.
31720 (call_summary): New class.
31721 (gt_ggc_mx): New overload.
31722 (gt_pch_nx): Likewise.
31723 (gt_pch_nx): Likewise.
31725 2017-05-03 Jeff Law <law@redhat.com>
31727 PR tree-optimization/78496
31728 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31729 from simplify_cond_using_ranges. Split off code to walk
31730 backwards through casts into ...
31731 (simplify_cond_using_ranges_2): New function.
31732 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31733 (execute_vrp): After identifying jump threads, call
31734 simplify_cond_using_ranges_2.
31736 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31739 * ipa-inline.h (inline_summary): Add ctor.
31740 (create_ggc): Do not use ggc_cleared_alloc.
31742 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31743 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31745 * gcc.c (handle_braces): Support escaping in switch matching
31747 * doc/invoke.texi (Spec Files): Document it.
31748 Remove superfluous @code markup in items.
31750 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31752 * diagnostic-show-locus.c (struct column_range): New struct.
31753 (get_affected_columns): New function.
31754 (get_printed_columns): New function.
31755 (struct correction): New struct.
31756 (correction::ensure_capacity): New function.
31757 (correction::ensure_terminated): New function.
31758 (struct line_corrections): New struct.
31759 (line_corrections::~line_corrections): New dtor.
31760 (line_corrections::add_hint): New function.
31761 (layout::print_trailing_fixits): Reimplement in terms of the new
31763 (selftest::test_overlapped_fixit_printing): New function.
31764 (selftest::diagnostic_show_locus_c_tests): Call it.
31766 2017-05-03 Nathan Sidwell <nathan@acm.org>
31768 Canonicalize canonical type hashing
31769 * tree.h (type_hash_canon_hash): Declare.
31770 * tree.c (type_hash_list, attribute_hash_list): Move into
31771 type_hash_canon_hash.
31772 (build_type_attribute_qual_variant): Break out hash code calc into
31773 type_hash_canon_hash.
31774 (type_hash_canon_hash): New. Generic type hash computation.
31775 (build_range_type_1, build_array_type_1, build_function_type,
31776 build_method_type_directly, build_offset_type, build_complex_type,
31777 make_vector_type): Call it.
31779 2017-05-03 Richard Biener <rguenther@suse.de>
31781 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31782 When all DRs have unknown misaligned do not always peel
31783 when there is a store but apply the same costing model as if
31784 there were only loads.
31786 2017-05-03 Richard Biener <rguenther@suse.de>
31789 PR tree-optimization/80492
31790 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31791 compare_base_decls returning dont-know properly.
31793 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
31795 * config/arm/iterators.md (CCSI): New mode iterator.
31796 (arch): New mode attribute.
31797 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
31798 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
31799 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
31800 code iterator for success result mode.
31801 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
31802 the corresponding new insn generators.
31804 2017-05-03 Bin Cheng <bin.cheng@arm.com>
31807 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31808 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31810 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
31812 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
31813 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
31814 (DDR_A): Wrap DDR argument in brackets.
31815 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
31816 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
31817 (DDR_REVERSED_P): Likewise.
31819 2017-05-03 Jakub Jelinek <jakub@redhat.com>
31821 PR tree-optimization/79472
31822 * tree-switch-conversion.c (struct switch_conv_info): Add
31823 contiguous_range and default_case_nonstandard fields.
31824 (collect_switch_conv_info): Compute contiguous_range and
31825 default_case_nonstandard fields, don't clear final_bb if
31826 contiguous_range and only the default case doesn't have the required
31828 (check_all_empty_except_final): Set default_case_nonstandard instead
31829 of failing if contiguous_range and the default case doesn't have empty
31831 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
31832 and only the default case doesn't have the required constants. Skip
31834 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
31835 if default_case_nonstandard.
31836 (build_constructors): Build constant 1 just once. Assert that default
31837 values aren't inserted in between cases if contiguous_range. Skip
31839 (build_arrays): Skip virtual phis.
31840 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
31841 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
31842 Handle virtual phis.
31843 (gen_inbound_check): Handle default_case_nonstandard case.
31844 (process_switch): Adjust check_final_bb caller. Call
31845 gather_default_values with the first non-default case instead of
31846 default case if default_case_nonstandard.
31848 2017-05-02 Nathan Sidwell <nathan@acm.org>
31850 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
31851 check. Fix formatting.
31853 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
31855 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
31856 errors when comparing specialized and unspecialized times.
31858 2017-05-02 David Malcolm <dmalcolm@redhat.com>
31860 * diagnostic-show-locus.c
31861 (layout::should_print_annotation_line_p): Make private.
31862 (layout::print_annotation_line): Make private.
31863 (layout::annotation_line_showed_range_p): Make private.
31864 (layout::show_ruler): Make private.
31865 (layout::print_source_line): Make private. Pass in line and
31866 line_width, rather than calling location_get_source_line. Drop
31868 (layout::print_leading_fixits): New method.
31869 (layout::print_any_fixits): Rename to...
31870 (layout::print_trailing_fixits): ...this, and make private.
31871 Don't print newline fixits.
31872 (diagnostic_show_locus): Move logic for printing one row into...
31873 (layout::print_line): ...this new function. Move the
31874 location_get_source_line call and error-handling from
31875 print_source_line to here. Call print_leading_fixits, and rename
31876 print_any_fixits to print_trailing_fixits.
31877 (selftest::test_fixit_insert_containing_newline): Update now that
31878 newlines are partially supported.
31879 (selftest::test_fixit_insert_containing_newline_2): New test.
31880 (selftest::test_fixit_replace_containing_newline): Update comments.
31881 (selftest::diagnostic_show_locus_c_tests): Call the new test.
31882 * edit-context.c (class added_line): New class.
31883 (class edited_line): Describe newline handling in comment.
31884 (edited_line::actually_edited_p): New method.
31885 (edited_line::print_content): Delete redundant decl.
31886 (edited_line::m_predecessors): New field.
31887 (edited_file::print_content): Call edited_line::print_content.
31888 (edited_file::print_diff): Update to support newlines.
31889 (edited_file::print_diff_hunk): Likewise.
31890 (edited_file::print_run_of_changed_lines): New function.
31891 (edited_file::print_diff_line): Convert to...
31892 (print_diff_line): ...this.
31893 (edited_file::get_effective_line_count): New function.
31894 (edited_line::edited_line): Initialize new field m_predecessors.
31895 (edited_line::~edited_line): Clean up m_predecessors.
31896 (edited_line::apply_fixit): Handle newlines.
31897 (edited_line::get_effective_line_count): New function.
31898 (edited_line::print_content): New function.
31899 (edited_line::print_diff_lines): New function.
31900 (selftest::test_applying_fixits_insert_containing_newline): New
31902 (selftest::test_applying_fixits_replace_containing_newline): New
31904 (selftest::insert_line): New function.
31905 (selftest::test_applying_fixits_multiple_lines): Add example of
31907 (selftest::edit_context_c_tests): Call the new tests.
31909 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31911 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
31912 parameter cand. Update dump information.
31913 (get_computation_cost): Update uses.
31915 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31917 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
31918 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
31919 (get_computation_at, rewrite_use_address): Update use of
31920 get_computation_aff.
31922 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31924 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
31925 (get_computation): Delete.
31926 (get_computation_cost): Implement like get_computation_cost_at.
31927 Use get_computation_at.
31928 (get_computation_cost_at): Delete.
31929 (rewrite_use_nonlinear_expr): Use get_computation_at.
31930 (rewrite_use_compare, remove_unused_ivs): Ditto.
31932 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31934 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
31936 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31938 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
31939 (ivopts_global_cost_for_size): Rename parameter and update uses.
31940 (iv_ca_recount_cost): Update uses.
31941 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
31942 candidates seperately in n_invs and n_cands.
31943 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
31945 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31947 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
31948 (find_inv_vars_cb): New.
31949 (find_depends): Renamed to ...
31950 (find_inv_vars): ... this.
31951 (add_candidate_1, force_var_cost): Call find_inv_vars.
31952 (split_address_cost, determine_group_iv_cost_cond): Ditto.
31954 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31956 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
31957 inv_vars. Add inv_exprs.
31958 (struct iv_cand): Rename depends_on to inv_vars.
31959 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
31960 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
31961 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
31962 (dump_cand): Dump inv_vars.
31963 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
31964 (record_invariant, find_depends, add_candidate_1): Ditto.
31965 (set_group_iv_cost, force_var_cost): Ditto.
31966 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
31967 (get_computation_cost_at, get_computation_cost): Ditto.
31968 (determine_group_iv_cost_generic): Ditto.
31969 (determine_group_iv_cost_address): Ditto.
31970 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
31971 (determine_group_iv_costs): Ditto.
31972 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
31973 (iv_ca_set_remove_invariants): Renamed to ...
31974 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
31975 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
31976 (iv_ca_set_add_invariants): Renamed to ...
31977 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
31978 (iv_ca_set_cp): Use iv_ca_set_add_invs.
31979 (iv_ca_has_deps): Support inv_vars and inv_exprs.
31980 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
31981 (create_new_ivs): Remove useless dump.
31983 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31985 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
31987 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
31988 (iv_ca_set_no_cp, create_new_iv): Ditto.
31990 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31992 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31994 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31996 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
31997 function tree_check2.
31999 2017-05-02 Martin Liska <mliska@suse.cz>
32001 * doc/gcov.texi: Add missing preposition.
32002 * gcov.c (function_info::function_info): Properly fill up
32003 all member variables.
32005 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32007 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32009 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32011 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32013 2017-05-02 Martin Liska <mliska@suse.cz>
32016 * lto-streamer-in.c (lto_read_tree_1): Remove
32017 LTO_STREAMER_DEBUG.
32018 * lto-streamer.c (struct tree_hash_entry): Likewise.
32019 (struct tree_entry_hasher): Likewise.
32020 (tree_entry_hasher::hash): Likewise.
32021 (tree_entry_hasher::equal): Likewise.
32022 (lto_streamer_init): Likewise.
32023 (lto_orig_address_map): Likewise.
32024 (lto_orig_address_get): Likewise.
32025 (lto_orig_address_remove): Likewise.
32026 * lto-streamer.h: Likewise.
32027 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32028 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32030 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32032 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32033 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32034 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32035 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32036 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32037 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32038 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32039 (mm_maskz_sub_ss): New intrinsics.
32040 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32041 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32042 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32043 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32044 (__builtin_ia32_subss_mask_round): New builtins.
32045 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32046 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32047 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32049 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32050 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32052 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32055 2017-05-02 Martin Jambor <mjambor@suse.cz>
32057 PR tree-optimization/78687
32058 * tree-sra.c (access): New field parent.
32059 (process_subtree_disqualification): New function.
32060 (disqualify_candidate): Call it.
32061 (build_accesses_from_assign): Reset write flag if creating an
32063 (build_access_subtree): Fill in parent field and also prpagate
32064 down grp_write flag.
32065 (create_artificial_child_access): New parameter set_grp_write, set
32066 grp_write to its value.
32067 (propagate_subaccesses_across_link): Also propagate grp_write flag
32069 (propagate_all_subaccesses): Push the closest parent back to work
32070 queue if add_access_to_work_queue returned true.
32072 2017-05-02 Richard Biener <rguenther@suse.de>
32074 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32075 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32076 -fstrict-overflow documentation.
32077 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32078 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32079 flag_strict_overflow.
32080 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32081 * lto-opts.c (lto_write_options): Do not stream it.
32082 * lto-wrapper.c (merge_and_complain): Do not handle it.
32083 * opts.c (default_options_table): Do not set -fstrict-overflow.
32084 (finish_options): Likewise do not clear it when sanitizing.
32085 * simplify-rtx.c (simplify_const_relational_operation): Do not
32086 test flag_strict_overflow.
32088 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32090 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32091 using enabled attribute.
32092 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32093 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32094 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32095 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32096 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32097 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32098 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32099 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32100 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32101 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32103 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32105 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32107 2017-05-02 Richard Biener <rguenther@suse.de>
32109 PR tree-optimization/80591
32111 2017-04-10 Richard Biener <rguenther@suse.de>
32113 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32116 2017-05-02 Richard Biener <rguenther@suse.de>
32118 PR tree-optimization/80549
32119 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32120 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32121 headers if they do not have a preheader.
32123 2017-05-02 Martin Liska <mliska@suse.cz>
32126 * common.opt: Fix typo.
32127 * doc/invoke.texi: Likewise.
32129 2017-05-01 Jan Beulich <jbeulich@suse.com>
32131 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32132 swapping, add (x,x,m,x,n) alternative.
32134 2017-05-01 Nathan Sidwell <nathan@acm.org>
32136 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32137 unnecessary unadjusted_alignment check.
32139 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32142 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32144 * gimplify.c (gimplify_cilk_detach): New function.
32145 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32146 * tree-core.h: Document EXPR_CILK_SPAWN.
32147 * tree.h (EXPR_CILK_SPAWN): Define.
32149 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32151 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32152 to use new fixit_hint representation, using the "replace" logic.
32153 (get_line_span_for_fixit_hint): Likewise.
32154 (layout::print_any_fixits): Likewise.
32155 (selftest::test_one_liner_many_fixits): Rename to...
32156 (selftest::test_one_liner_many_fixits_1): ...this, and update
32157 comment and expected output to reflect that the multiple fix-it
32158 hints are now consolidated into one insertion.
32159 (selftest::test_one_liner_many_fixits_2): New test.
32160 (selftest::test_diagnostic_show_locus_one_liner): Update for
32162 (selftest::test_fixit_consolidation): Update for fix-it API
32164 * diagnostic.c (print_parseable_fixits): Likewise.
32165 * edit-context.c (edited_line::m_line_events): Convert from
32166 auto_vec <line_event *> to auto_vec <line_event>.
32167 (class line_event): Convert from abstract base class to a concrete
32168 class, taking over the role of replace_event.
32169 (class insert_event): Delete.
32170 (class replace_event): Rename to class line_event. Convert to
32172 (edit_context::add_fixits): Reimplement.
32173 (edit_context::apply_insert): Delete.
32174 (edit_context::apply_replace): Rename to...
32175 (edit_context::apply_fixit): ...this. Convert to half-open range.
32176 (edited_file::apply_insert): Delete.
32177 (edited_file::apply_replace): Rename to...
32178 (edited_file::apply_fixit): ...this.
32179 (edited_line::~edited_line): Drop deletion of events.
32180 (edited_line::apply_insert): Delete.
32181 (edited_line::apply_replace): Rename to...
32182 (edited_line::apply_fixit): ...this. Convert to half-open range.
32183 Update for change to type of m_line_events.
32184 * edit-context.h (edit_context::apply_insert): Delete.
32185 (edit_context::apply_replace): Rename to...
32186 (edit_context::apply_fixit): ...this.
32188 2017-05-01 Martin Sebor <msebor@redhat.com>
32190 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32193 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32196 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32197 __get_cpuid_max returns 0.
32198 (__get_cpuid_count): Ditto.
32200 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32202 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32203 replacement expression is another instance of one of its arguments.
32205 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32208 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32209 check for stack push/pop autoinc.
32210 * config/i386/i386.c (ix86_agi_dependent): Return false
32211 if the only reason why modified_in_p returned true is that
32212 addr is SP based and set_insn is a push or pop.
32214 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32216 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32219 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32222 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32223 (account_size_time): Use two predicates - exec_pred and
32225 (evaluate_conditions_for_known_args): Compute both clause and
32227 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32228 (inline_summary_t::duplicate): Update.
32229 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32231 (compute_inline_parameters): Likewise.
32232 (estimate_edge_size_and_time): Update caluclation of time.
32233 (estimate_node_size_and_time): Compute both time and nonspecialized
32235 (estimate_ipcp_clone_size_and_time): Update.
32236 (inline_merge_summary): Update.
32237 (do_estimate_edge_time): Update.
32238 (do_estimate_edge_size): Update.
32239 (do_estimate_edge_hints): Update.
32240 (inline_read_section, inline_write_summary): Stream both new predicates.
32241 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32243 (compute_inlined_call_time): Cleanup.
32244 (big_speedup_p): Update.
32245 (edge_badness): Update.
32246 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32247 (size_time_entry): Replace predicate by exec_predicate and
32248 nonconst_predicate.
32249 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32250 (estimate_edge_time): Return also nonspec_time.
32251 (reset_edge_growth_cache): Update.
32253 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32255 PR rtl-optimization/80491
32256 * ifcvt.c (noce_process_if_block): When looking for x setter
32257 with missing else_bb, don't check only the insn right before
32258 cond_earliest, but look for the last insn that x is modified in
32259 within the same bb.
32261 PR rtl-optimization/80491
32262 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32264 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32266 PR tree-optimization/80487
32267 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32269 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32271 PR tree-optimization/79697
32272 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32273 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32274 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32276 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32277 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32279 2017-04-28 Martin Sebor <msebor@redhat.com>
32281 PR tree-optimization/80523
32282 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32283 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32285 (maybe_warn, format_directive, parse_directive): Use new functions.
32286 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32288 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32290 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32292 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32294 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32295 target_header_dir): Set correctly.
32296 * configure: Regenerated.
32297 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32298 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32299 instead of SYSTEM_HEADER_DIR.
32301 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32303 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32304 (estimate_local_effects): Likewise.
32305 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32306 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32307 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32308 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32309 do_estimate_edge_time, estimate_edge_time): Likewise.
32310 * ipa-inline-analysis.c (estimate_node_size_and_time,
32311 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32312 (estimate_time_after_inlining): Remove.
32314 2017-04-28 Martin Liska <mliska@suse.cz>
32316 * doc/gcov.texi: Enhance documentation of gcov.
32318 2017-04-28 Martin Liska <mliska@suse.cz>
32320 * doc/gcov.texi: Sort options in alphabetic order.
32321 * doc/gcov-dump.texi: Likewise.
32322 * doc/gcov-tool.texi: Likewise.
32323 * gcov.c (print_usage): Likewise.
32324 * gcov-dump.c (print_usage): Likewise.
32325 * gcov-tool.c (print_merge_usage_message): Likewise.
32326 (print_rewrite_usage_message): Likewise.
32327 (print_overlap_usage_message): Likewise.
32329 2017-04-28 Martin Liska <mliska@suse.cz>
32331 PR gcov-profile/53915
32332 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32334 2017-04-28 Martin Liska <mliska@suse.cz>
32336 PR gcov-profile/79891
32337 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32338 is marked by compiler as living on a line.
32339 (get_cycles_count): Remove usage of the union.
32340 (output_intermediate_file): Likewise.
32341 (find_source): Fix GNU coding style.
32342 (accumulate_line_counts): Remove old non-all block mode.
32343 (output_lines): Remove usage of the union.
32344 * profile.c (output_location): Include all BBs, even if
32345 belonging to a same line (and file) as a previous BB.
32347 2017-04-28 Martin Liska <mliska@suse.cz>
32349 * gcov.c (process_args): Handle new argument 'w'.
32350 (read_graph_file): Assign ID to BBs.
32351 (output_branch_count): Display BB # if verbose flag is set.
32352 (output_lines): Likewise for arcs.
32353 (print_usage): Add '--verbose' option help.
32354 * doc/gcov.texi: Document --verbose (-w) option.
32356 2017-04-28 Martin Liska <mliska@suse.cz>
32358 * gcov.c (struct block_location_info): New struct.
32359 (process_file): Fill up the new structure.
32360 (read_graph_file): Replace usage of encoding by the newly added
32362 (add_line_counts): Likewise.
32363 (accumulate_line_counts): Remove usage of the union.
32364 (function_info::function_info): New function.
32365 (function_info::~function_info): Likewise.
32366 (process_file): Call delete instead of release_function.
32367 (release_function): Release the function.
32368 (release_structures): Call delete instead of release_function.
32369 (solve_flow_graph): Replace usage of num_blocks.
32370 (find_exception_blocks): Likewise.
32371 (output_lines): Fix GNU coding style.
32373 2017-04-28 Martin Liska <mliska@suse.cz>
32376 * coverage.c (coverage_remove_note_file): New function.
32377 * coverage.h: Declare the function.
32378 * toplev.c (finalize): Clean if an error has been seen.
32380 2017-04-28 Martin Liska <mliska@suse.cz>
32382 PR gcov-profile/80031
32383 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32384 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32385 * gcov.c (read_graph_file): Read just number of blocks.
32386 * profile.c (branch_prob): Do not stream 0 flags per a basic
32389 2017-04-28 Martin Liska <mliska@suse.cz>
32391 * gcov-dump.c (tag_*): Add new argument to declarations.
32392 (dump_gcov_file): Likewise.
32393 (tag_blocks): Add and use new argument depth.
32394 (tag_arcs): Likewise.
32395 (tag_lines): Likewise.
32396 (tag_counters): Likewise.
32397 (tag_summary): Likewise.
32398 (dump_working_sets): Use depth to do a proper indentation.
32400 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32403 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32404 * symtab.c (symtab_node::debug_symtab): Move definition here.
32406 2017-04-28 Richard Biener <rguenther@suse.de>
32408 * lto-streamer.h (LTO_major_version): Bump to 7.
32410 2017-04-28 Richard Biener <rguenther@suse.de>
32412 * tree-vrp.c (assert_info): New struct.
32413 (add_assert_info): New helper.
32414 (register_edge_assert_for_2): Refactor to add asserts to a vector
32416 (register_edge_assert_for_1): Likewise.
32417 (register_edge_assert_for): Likewise.
32418 (finish_register_edge_assert_for): New helper actually registering
32419 asserts where live on edge.
32420 (find_conditional_asserts): Adjust.
32421 (find_switch_asserts): Likewise.
32422 (evrp_dom_walker::try_find_new_range): Generalize.
32423 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32425 2017-04-27 Marek Polacek <polacek@redhat.com>
32428 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32429 arg10 and arg11 to itype.
32431 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32433 * doc/extend.texi (Object Size Checking): Improve grammar.
32435 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32438 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32439 that the logic for permitting reciprocal estimates matches that
32442 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32445 * tree.c (type_cache_hasher::equal): Only compare
32446 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32447 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32448 non-aggregate element types.
32449 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32450 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32452 2017-04-27 Richard Biener <rguenther@suse.de>
32454 PR middle-end/80533
32455 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32456 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32457 keeping a reference to a trailing array.
32459 2017-04-27 Richard Biener <rguenther@suse.de>
32461 PR middle-end/80539
32462 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32463 being in loop-closed SSA form conservatively.
32464 (chrec_fold_multiply_poly_poly): Likewise.
32466 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32468 PR middle-end/79665
32469 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32470 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32472 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32475 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32476 (aarch64_function_arg_alignment): Return unsigned int again, but still
32477 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32478 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32479 Don't emit -Wpsabi note.
32480 (aarch64_function_arg_boundary): Likewise.
32481 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32484 2017-04-26 Nathan Sidwell <nathan@acm.org>
32486 * tree.h (crc32_unsigned_n): Declare.
32487 (crc32_unsigned, crc32_unsigned): Make inline.
32488 * tree.c (crc32_unsigned_bits): Replace with ...
32489 (crc32_unsigned_n): ... this.
32490 (crc32_unsigned, crc32_byte): Remove.
32491 (crc32_string): Remove unnecessary braces.
32493 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32495 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32496 * ipa-inline-analysis.c (MAX_TIME): Remove.
32497 (account_size_time): Use sreal for time.
32498 (dump_inline_summary): Update.
32499 (estimate_function_body_sizes): Update.
32500 (estimate_edge_size_and_time): Update.
32501 (estimate_calls_size_and_time): Update.
32502 (estimate_node_size_and_time): Update.
32503 (inline_merge_summary): Update.
32504 (inline_update_overall_summary): Update.
32505 (estimate_time_after_inlining): Update.
32506 (inline_read_section): Update.
32507 (inline_write_summary): Update.
32508 * ipa-inline.c (compute_uninlined_call_time): Update.
32509 (compute_inlined_call_time): Update.
32510 (recursive_inlining): Update.
32511 (inline_small_functions): Update.
32512 (dump_overall_stats): Update.
32513 * ipa-inline.h: Include sreal.h.
32514 (size_time_entry): Turn time to sreal.
32515 (inline_summary): Turn self_time nad time to sreal.
32517 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32519 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32521 (sreal::stream_out, sreal::stream_in): New.
32522 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32524 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32526 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32529 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32532 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32534 (dimode_scalar_chain::compute_convert_gain): Ditto.
32535 (dimode_scalar_chain::make_vector_copies): Ditto.
32536 (dimode_scalar_chain::convert_reg): Ditto.
32537 (dimode_scalar_chain::convert_insn): Ditto.
32538 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32539 (VI248_AVX512BW_1): New mode iterator.
32540 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32541 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32544 2017-04-25 Martin Sebor <msebor@redhat.com>
32546 PR tree-optimization/80497
32547 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32548 constants are representable in HOST_WIDE_INT.
32549 (parse_directive): Ditto.
32551 2017-04-25 Martin Sebor <msebor@redhat.com>
32554 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32555 (new_zero_array): Adjust signature.
32556 (dom_info::dom_init): Used unsigned rather that size_t.
32557 (dom_info::dom_info): Same.
32559 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32560 Jakub Jelinek <jakub@redhat.com>
32563 * config/arm/arm.c: Include gimple.h.
32564 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32565 returns negative, increment ncrn only if it returned positive.
32566 (arm_needs_doubleword_align): Return int instead of bool,
32567 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32568 members, but if there is any such non-FIELD_DECL
32569 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32570 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32571 returns negative, increment nregs only if it returned positive.
32572 (arm_setup_incoming_varargs): Likewise.
32573 (arm_function_arg_boundary): Emit -Wpsabi note if
32574 arm_needs_doubleword_align returns negative, return
32575 DOUBLEWORD_ALIGNMENT only if it returned positive.
32577 2017-04-25 Marek Polacek <polacek@redhat.com>
32580 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32581 first argument to type.
32583 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32586 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32587 type checks to test for compatibility instead of equality.
32589 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32590 Jakub Jelinek <jakub@redhat.com>
32593 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32595 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32596 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32597 the alignment computation, but return their maximum in warn_alignment.
32598 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32599 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32601 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32602 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32605 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32607 * config/arc/simdext.md (dmpyh): Fix typo.
32609 2017-04-25 Richard Biener <rguenther@suse.de>
32611 PR tree-optimization/80492
32612 * alias.c (compare_base_decls): Handle registers with asm
32613 specification conservatively.
32614 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32615 compare_base_decls returning dont-know properly.
32617 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32619 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32620 (legitimate_offset_address_p): New function.
32621 (arc_legitimate_address_p): Use above function.
32623 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32625 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32627 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32629 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32630 ACCH registers whenever they are available.
32632 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32634 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32635 double regs fix when not used.
32637 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32639 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32641 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32642 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32644 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32646 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32647 when emitting short ADD instructions.
32649 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32651 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32653 (cmpsi_cc_c_insn): Likewise.
32654 (cbranchsi4_scratch): Compute proper instruction length using
32655 compact_hreg_operand.
32656 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32658 2017-04-25 Richard Biener <rguenther@suse.de>
32660 PR middle-end/80509
32661 * passes.c (pass_manager::pass_manager): Initialize
32662 m_name_to_pass_map.
32664 2017-04-25 Richard Biener <rguenther@suse.de>
32666 PR tree-optimization/79201
32667 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32669 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32672 * config/s390/vector.md: Split MEM->GPR vector moves for
32673 non-s_operand addresses.
32675 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32678 * config/s390/predicates.md (reload_const_wide_int_operand): New
32680 * config/s390/s390.md ("movti"): Remove d/P alternative.
32681 ("movti_bigconst"): New pattern definition.
32683 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32686 * s390-protos.h (s390_expand_cs_hqi): Removed.
32687 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32688 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32689 modes as well as CCZ1mode and CCZmode.
32690 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32691 signature of s390_emit_compare_and_swap.
32692 (s390_expand_cs_hqi): Likewise, make static.
32693 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32694 compare-and-swap, in some cases.
32695 (s390_expand_cs): Wrapper function.
32696 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32698 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32699 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32700 patterns for small and large integers. Forbid symref memory operands.
32701 Move expander to s390.c. Require cc register.
32702 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32703 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32704 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32705 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32706 symref memory operands. Remove CC mode and call s390_match_ccmode
32708 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32710 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32712 * config/s390/s390.md (define_peephole2): New peephole to help
32713 combining the load-and-test pattern with volatile memory.
32715 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32717 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32718 with CCZmode for TARGET_Z196.
32720 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32722 PR rtl-optimization/80501
32723 * combine.c (make_compound_operation_int): Set subreg_code to SET
32724 even for AND with mask of the sign bit of mode.
32726 PR rtl-optimization/80500
32727 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32728 sum's initial value.
32730 2017-04-25 Julian Brown <julian@codesourcery.com>
32731 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32733 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32735 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32737 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32739 2017-04-25 Julian Brown <julian@codesourcery.com>
32740 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32742 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32743 (thunderx2t99_sha): New Reservation.
32745 2017-04-25 Julian Brown <julian@codesourcery.com>
32746 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32748 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32749 type for 1-element load.
32751 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32753 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32755 2017-04-24 Martin Jambor <mjambor@suse.cz>
32757 PR tree-optimization/80293
32758 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32759 char arrays not totally scalarizable if it is false.
32760 (analyze_all_variable_accesses): Pass correct value in the new
32761 parameter. Add a statistics counter.
32763 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32765 PR middle-end/79931
32766 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32768 2017-04-24 Richard Biener <rguenther@suse.de>
32770 PR tree-optimization/80494
32771 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32772 out for complex types.
32774 2017-04-24 Richard Biener <rguenther@suse.de>
32776 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32777 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32778 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32779 (DFS): Adjust and never fail.
32780 (sccvn_dom_walker::fail): Remove.
32781 (sccvn_dom_walker::before_dom_children): Adjust.
32782 (run_scc_vn): Likewise and never fail.
32783 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32784 (pass_fre::execute): Likewise.
32786 2017-04-24 Richard Biener <rguenther@suse.de>
32788 PR tree-optimization/79725
32789 * tree-ssa-sink.c (statement_sink_location): Return whether
32790 failure reason was zero uses. Move that check later.
32791 (sink_code_in_bb): Deal with zero uses by removing the stmt
32794 2017-04-24 Richard Biener <rguenther@suse.de>
32797 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
32798 pointer-based references.
32800 2017-04-24 Richard Biener <rguenther@suse.de>
32803 * pass_manager.h (pass_manager::operator new): Remove.
32804 (pass_manager::operator delete): Likewise.
32805 * passes.c (pass_manager::operator new): Remove.
32806 (pass_manager::operator delete): Likewise.
32807 (pass_manager::pass_manager): Zero individual pass members.
32809 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
32812 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
32813 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
32814 Check "XEXP (src, 1)" operand here.
32815 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
32816 Check "XEXP (src, 1)" operand here.
32817 (dimode_scalar_chain::make_vector_copies): Detect count register
32818 of a shift instruction. Zero extend count register from QImode
32819 to DImode to satisfy vector shift pattern count operand predicate.
32820 Substitute vector shift count operand with a DImode copy.
32821 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
32824 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
32826 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
32827 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32828 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
32829 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32830 (UNSPEC_NOREX_MEM): Remove definition.
32832 2017-04-21 Richard Biener <rguenther@suse.de>
32834 PR tree-optimization/79547
32835 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
32836 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
32837 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
32838 without any constraints.
32840 2017-04-21 Richard Biener <rguenther@suse.de>
32842 PR tree-optimization/78847
32843 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
32845 2017-04-21 Richard Biener <rguenther@suse.de>
32847 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
32848 (build_distinct_type_copy): Likewise.
32849 (build_variant_type_copy): Likewise.
32850 * tree.c (build_qualified_type): Pass down mem-stat info.
32851 (build_distinct_type_copy): Likewise.
32852 (build_variant_type_copy): Likewise.
32854 2017-04-21 Richard Biener <rguenther@suse.de>
32856 PR tree-optimization/80237
32857 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
32859 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
32860 for a simplified result.
32862 2016-04-21 Richard Biener <rguenther@suse.de>
32864 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
32865 sth as strict as a simple_iv but a chrec without symbols and an
32866 operand defined in the loop we are peeling (and not some subloop).
32867 (propagate_constants_for_unrolling): Propagate all constants.
32869 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32872 * config/i386/i386.c (print_reg): Remove assert for disalowed
32873 regno values, call output_operand_lossage instead.
32875 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32878 * config/i386/constraints.md (Yc): New register constraint.
32879 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
32880 Use Yc constraint for alternative 2 of operand 0. Remove
32881 preferred_for_speed attribute.
32883 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32885 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
32886 lastprivate clauses in SIMT case.
32888 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
32890 * doc/invoke.texi (-Wextra-semi): Document new warning option.
32892 2017-04-20 Richard Biener <rguenther@suse.de>
32894 PR tree-optimization/57796
32895 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
32896 as N scalar stores.
32897 (vect_model_load_cost): Cost gathers as N scalar loads.
32899 2017-04-20 Richard Biener <rguenther@suse.de>
32901 * ggc-page.c (ggc_allocated_p): Rename to ...
32902 (safe_lookup_page_table_entry): ... this and return the lookup
32904 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
32906 2017-04-20 Richard Biener <rguenther@suse.de>
32908 PR tree-optimization/80453
32909 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
32910 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
32911 from the conditions.
32912 (vn_phi_eq): Pass them down.
32913 (vn_phi_lookup): Record them.
32914 (vn_phi_insert): Likewise.
32916 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
32918 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
32919 uninitialized variable warning to avoid buffer overrun.
32921 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32924 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
32925 is suppressed for '{ 0 }' in C.
32927 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32929 * BASE-VER: Set to 8.0.0.
32931 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
32933 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
32934 priority .init_array and .fini_array section with SECTION_NOTYPE
32937 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32939 PR middle-end/80423
32940 * tree.h (build_array_type): Add typeless_storage default argument.
32941 * tree.c (type_cache_hasher::equal): Also compare
32942 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
32943 (build_array_type): Add typeless_storage argument, set
32944 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
32946 (build_nonshared_array_type): Adjust build_array_type_1 caller.
32947 (build_array_type): Likewise. Add typeless_storage argument.
32949 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32950 Jakub Jelinek <jakub@redhat.com>
32952 PR tree-optimization/80426
32953 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
32954 operation on symbolic operands, also compute the overflow for the
32955 invariant part when the operation degenerates into a negation.
32957 2017-04-19 Jakub Jelinek <jakub@redhat.com>
32960 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
32961 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
32964 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
32966 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
32969 * config/avr/avr.c (tree.h): Include it.
32970 (cgraph.h): Include it.
32971 (avr_encode_section_info): Don't warn for uninitialized progmem
32972 variable if it's just an alias.
32974 2017-04-19 Richard Biener <rguenther@suse.de>
32977 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
32978 when needed by AutoPGO.
32980 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
32983 * doc/lto.texi: Remove an extra 'that'.
32985 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
32987 PR rtl-optimization/80429
32988 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
32989 are only used in debug insns.
32991 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32992 Vladimir Makarov <vmakarov@redhat.com>
32994 * config/sparc/predicates.md (input_operand): Add comment. Return
32995 true for any memory operand when LRA is in progress.
32996 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
32998 2017-04-18 Jeff Law <law@redhat.com>
33001 * mips.md ({return,simple_return}_internal): Do not overwrite
33004 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33006 PR tree-optimization/80443
33007 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33008 instead of adding 1, subtract -1 and similarly instead of subtracting
33011 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33013 PR rtl-optimization/80357
33014 * haifa-sched.c (tmp_bitmap): New variable.
33015 (model_recompute): Handle duplicate use records.
33016 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33017 (free_global_sched_pressure_data): Free it.
33019 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33022 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33023 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33024 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33025 instead of SYSTEM_HEADER_DIR.
33027 2017-04-18 Jeff Law <law@redhat.com>
33029 PR middle-end/80422
33030 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33031 predecessors after walking up the insn chain.
33033 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33036 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33037 sizetype type into debug info.
33039 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33042 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33043 unneeded test for TARGET_UPPER_REGS_SF.
33044 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33046 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33049 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33050 instead of gsi_after_labels.
33052 2017-04-18 Jeff Law <law@redhat.com>
33054 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33058 2017-04-13 Jeff Law <law@redhat.com>
33059 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33060 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33062 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33065 * config/avr/avr.c (intl.h): Include it.
33066 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33068 2017-04-18 Martin Liska <mliska@suse.cz>
33070 PR gcov-profile/78783
33071 * gcov-tool.c (gcov_output_files): Validate that destination
33072 file is either removed by the tool or by a user.
33074 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33075 Guy Benyei <guybe@mellanox.com>
33077 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33078 block, and do not negate it, the stored id is already negative.
33080 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33082 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33084 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33087 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33088 masks of options that should be turned off if the VSX vector
33089 options are turned off.
33090 (OTHER_P8_VECTOR_MASKS): Likewise.
33091 (OTHER_VSX_VECTOR_MASKS): Likewise.
33092 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33093 rs6000_disable_incompatible_switches to validate no type switches
33095 (rs6000_incompatible_switch): New function to disallow turning on
33096 other vector options if -mno-vsx, -mno-power8-vector, or
33097 -mno-power9-vector are specified.
33099 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33101 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33103 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33105 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33106 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33107 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33108 (ARG_POINTER_CFA_OFFSET): Likewise.
33110 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33112 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33113 conditions to take advantage of various optimizations.
33115 2017-04-13 Jeff Law <law@redhat.com>
33117 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33118 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33119 (zero_extendsidi2_dext): Likewise.
33121 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33124 * fold-const.c (fold_ternary_loc): Revert
33125 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33128 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33130 PR rtl-optimization/80343
33131 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33132 new scratch pseudo.
33134 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33137 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33138 to ubsan_encode_value.
33140 2017-04-13 Jeff Law <law@redhat.com>
33142 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33143 appearing in DEBUG_INSNs.
33145 2017-04-13 Martin Liska <mliska@suse.cz>
33147 PR gcov-profile/80413
33148 * gcov-io.c (gcov_write_string): Copy to buffer just when
33149 allocated size is greater than zero.
33151 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33154 * dwarf2out.c (decls_for_scope): Ignore declarations of
33155 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33157 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33160 * ipa-visibility.c (non_local_p): Fix typos.
33161 (localize_node): When localizing symbol in same comdat group,
33162 dissolve the group only when we know external symbols are going
33164 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33166 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33168 PR tree-optimization/79390
33169 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33170 order does not result in usable sequence, retry with reversed operand
33176 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33177 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33178 op0 instead of fold_convert_loc (loc, type, arg0).
33180 2017-04-12 Jeff Law <law@redhat.com>
33182 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33183 has a delay slot in the generated code.
33185 * config/cris/cris.md (cris_preferred_reload_class): Return
33186 GENNONACR_REGS rather than GENERAL_REGS.
33188 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33191 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33192 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33193 signedness of the result type.
33195 2017-04-12 Richard Biener <rguenther@suse.de>
33196 Jeff Law <law@redhat.com>
33198 PR tree-optimization/80359
33199 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33200 trim stores to TARGET_MEM_REFs.
33202 2017-04-12 Richard Biener <rguenther@suse.de>
33204 PR tree-optimization/79390
33205 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33206 threading case even more.
33208 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33211 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33212 for quad_address_p for TImode, instead of just not indexed_address.
33214 2017-04-12 Richard Biener <rguenther@suse.de>
33215 Bernd Edlinger <bernd.edlinger@hotmail.de>
33217 PR middle-end/79671
33218 * alias.c (component_uses_parent_alias_set_from): Handle
33219 TYPE_TYPELESS_STORAGE.
33220 (get_alias_set): Likewise.
33221 * tree-core.h (tree_type_common): Add typeless_storage flag.
33222 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33223 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33224 for types containing members with TYPE_TYPELESS_STORAGE.
33225 (place_field): Likewise.
33226 (layout_type): Likewise for ARRAY_TYPE.
33227 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33228 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33229 TYPE_TYPELESS_STORAGE.
33230 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33232 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33235 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33236 first argument to type.
33238 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33242 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33243 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33244 (rs6000_expand_binop_builtin): Likewise.
33245 (rs6000_expand_ternop_builtin): Likewise; also add missing
33246 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33247 vshasigma built-ins.
33248 * doc/extend.texi: Document that vec_xxpermdi's third argument
33249 must be a constant.
33251 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33253 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33254 Use shift_const cost parameter when calculating gain of STV shifts.
33256 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33258 PR rtl-optimization/70478
33259 * lra-constraints.c (process_alt_operands): Check memory for
33260 disfavoring memory insn operand.
33262 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33264 PR middle-end/80100
33265 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33266 left shift in unsigned HOST_WIDE_INT type.
33268 PR rtl-optimization/80385
33269 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33270 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33273 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33274 if they have any depend clauses.
33276 2017-04-11 Martin Liska <mliska@suse.cz>
33279 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33280 * ipa-split.c (split_function): Create a local comdat symbol
33281 if caller is in a comdat group.
33283 2017-04-11 Martin Liska <mliska@suse.cz>
33286 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33289 2017-04-11 Martin Sebor <msebor@redhat.com>
33291 PR middle-end/80364
33292 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33293 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33295 (directive::set_width, directive::set_precision, format_character):
33297 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33300 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33303 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33304 conflict, set target->arch_name instead of target->cpu_name.
33306 2017-04-11 Richard Biener <rguenther@suse.de>
33308 PR tree-optimization/80374
33309 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33310 build_zero_cst, remove fold_convertible_p check again.
33312 2017-04-11 Martin Liska <mliska@suse.cz>
33315 * ubsan.c (instrument_object_size): Do not instrument register
33318 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33321 * config/i386/i386-builtin-types.def
33322 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33323 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33324 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33325 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33326 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33327 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33328 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33329 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33330 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33331 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33332 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33333 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33334 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33335 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33336 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33337 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33338 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33339 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33340 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33341 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33342 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33343 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33344 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33345 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33346 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33347 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33348 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33349 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33350 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33351 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33352 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33353 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33354 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33355 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33356 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33357 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33358 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33359 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33360 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33361 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33362 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33363 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33364 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33365 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33366 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33367 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33369 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33370 flag to second_arg_count, handle 4 argument function type _COUNT
33371 aliases, handle second_arg_count on second argument rather than last.
33373 2017-04-10 Jeff Law <law@redhat.com>
33375 PR tree-optimization/80374
33376 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33377 record anything if we can not convert integer_zero_node to the
33380 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33383 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33384 Enhance special handling given to the TARGET_P9_MINMAX option in
33385 relation to certain other options.
33387 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33389 PR tree-optimization/80153
33390 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33391 remove POINTER_PLUS_EXPR's base part directly, rather than through
33394 2017-04-10 Richard Biener <rguenther@suse.de>
33395 Bin Cheng <bin.cheng@arm.com>
33397 PR tree-optimization/80153
33398 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33399 the first element of pointer type aff_tree. Build result expr in
33401 (add_elt_to_tree): Convert to type unconditionally. Remove other
33402 fold_convert calls.
33403 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33404 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33406 2017-04-10 Richard Biener <rguenther@suse.de>
33408 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33411 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33413 PR rtl-optimization/70478
33414 * lra-constraints.c (curr_small_class_check): New.
33415 (update_and_check_small_class_inputs): New.
33416 (process_alt_operands): Update curr_small_class_check. Disfavor
33417 alternative insn memory operands. Check available regs for small
33420 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33423 * config/mips/mips.opt (-mvirt): Update description.
33424 * doc/invoke.texi (-mvirt): Likewise.
33426 2017-04-10 Richard Biener <rguenther@suse.de>
33428 PR middle-end/80362
33429 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33430 looking for NEGATE_EXPR in -A / -B to A / B folding.
33432 2017-04-10 Martin Liska <mliska@suse.cz>
33434 PR gcov-profile/80224
33435 * gcov.c (print_usage): Fix usage string.
33436 (get_gcov_intermediate_filename): Remove.
33437 (output_gcov_file): Use both for normal and intermediate format.
33438 (generate_results): Do not initialize special file for
33439 intermediate format.
33441 2017-04-10 Richard Biener <rguenther@suse.de>
33443 PR tree-optimization/80304
33444 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33447 2017-04-10 Nathan Sidwell <nathan@acm.org>
33450 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33451 (rs6000_init_builtins): Use it.
33453 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33455 * config/arm/arm.md (<mrc>): Add mode to SET source.
33456 (<mrrc>): Likewise.
33458 2017-04-10 Richard Biener <rguenther@suse.de>
33460 PR middle-end/80344
33461 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33463 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33466 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33467 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33468 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33469 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33470 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33471 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33472 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33473 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33474 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33475 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33476 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33477 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33478 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33479 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33480 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33481 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33482 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33483 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33484 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33485 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33486 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33487 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33488 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33490 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33492 PR rtl-optimization/70478
33493 * lra-constraints.c: Reverse the last patch.
33495 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33497 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33498 Add comment for WCHAR_T.
33500 2017-04-08 Martin Liska <mliska@suse.cz>
33503 2017-04-07 Martin Liska <mliska@suse.cz>
33506 * ipa-split.c (split_function): Add function part to a same comdat
33509 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33512 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33514 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33516 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33517 (vec_pairing): ...to this.
33518 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33519 (rs6000_sched_init): Adjust for name change.
33520 (struct rs6000_sched_context): Likewise.
33521 (rs6000_init_sched_context): Likewise.
33522 (rs6000_set_sched_context): Likewise.
33524 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33530 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33532 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33533 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33534 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33536 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33538 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33540 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33542 PR rtl-optimization/70703
33543 * ira-color.c (update_conflict_hard_regno_costs): Use
33544 int64_t instead of HOST_WIDE_INT.
33546 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33548 PR rtl-optimization/70478
33549 * lra-constraints.c (process_alt_operands): Disfavor alternative
33550 insn memory operands.
33552 2017-04-07 Jeff Law <law@redhat.com>
33554 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33555 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33557 2017-04-07 Martin Liska <mliska@suse.cz>
33560 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33561 Show error message instead of an ICE.
33563 2017-04-07 Martin Liska <mliska@suse.cz>
33566 * ipa-split.c (split_function): Add function part to a same comdat
33569 2017-04-07 Richard Biener <rguenther@suse.de>
33571 PR middle-end/80341
33572 * tree.c (get_unwidened): Also handle ! for_type case for
33574 * convert.c (do_narrow): Split out from ...
33575 (convert_to_integer_1): ... here. Do not pass final truncation
33576 type to get_unwidened for TRUNC_DIV_EXPR.
33578 2017-04-07 Richard Biener <rguenther@suse.de>
33580 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33582 (aff_combination_const): Adjust.
33583 (aff_combination_scale): Likewise.
33584 (aff_combination_add_elt): Likewise.
33585 (aff_combination_add_cst): Likewise.
33586 (aff_combination_convert): Likewise.
33587 (add_elt_to_tree): Likewise. Remove unused argument.
33588 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33590 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33592 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33594 * config/arm/arm.c (arm_default_short_enums): Use
33595 ARM_DEFAULT_SHORT_ENUMS.
33596 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33598 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33601 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33602 members with redundant out-of-class redeclaration.
33604 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33607 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33608 * config/i386/i386.md (*zero_extendsidi2):
33609 Add (?*x,*x) and (?*v,*v) alternatives.
33611 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33614 * config/i386/i386.c (ix86_expand_builtin)
33615 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33616 mode from insn data. Convert operands to insn operand mode.
33617 Copy operands that don't satisfy insn predicate to a register.
33619 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33621 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33624 2017-04-06 Richard Biener <rguenther@suse.de>
33626 PR tree-optimization/80334
33627 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33628 preserve alignment of accesses.
33630 2017-04-06 Richard Biener <rguenther@suse.de>
33632 PR tree-optimization/80262
33633 * tree-sra.c (build_ref_for_offset): Preserve address-space
33635 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33636 Drop useless address-space information on MEM_REF offsets.
33638 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33640 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33642 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33644 PR rtl-optimization/70703
33645 * ira-color.c (update_conflict_hard_regno_costs): Use
33646 HOST_WIDE_INT instead of long.
33648 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33651 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33652 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33654 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33655 for x86_64 target. Handle -m3dnowa option.
33657 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33659 PR rtl-optimization/70703
33660 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33661 (update_conflict_hard_regno_costs): Use long instead of unsigned
33662 arithmetic for cost calculation.
33664 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33665 Bernd Edlinger <bernd.edlinger@hotmail.de>
33668 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33671 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33674 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33675 ptr_mode with Pmode throughout.
33676 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33677 into probe_stack_range and use DImode.
33679 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33682 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33683 call_eh_return is true.
33685 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33687 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33688 Initialize last_match_fntype_index.
33690 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33693 * tree-nvr.c: Include internal-fn.h.
33694 (pass_return_slot::execute): Ignore internal calls without
33697 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33698 Richard Biener <rguenther@suse.de>
33701 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33702 captures used multiple times, except for the last use.
33703 * generic-match-head.c: Include gimplify.h.
33705 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33707 PR tree-optimization/79390
33708 * target.h (struct noce_if_info): Declare.
33709 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33710 * target.def (noce_conversion_profitable_p): New target hook.
33711 * ifcvt.h (struct noce_if_info): New type, moved from ...
33712 * ifcvt.c (struct noce_if_info): ... here.
33713 (noce_conversion_profitable_p): Renamed to ...
33714 (default_noce_conversion_profitable_p): ... this. No longer
33716 (noce_try_store_flag_constants, noce_try_addcc,
33717 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33718 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33719 instead of noce_conversion_profitable_p.
33720 * config/i386/i386.c: Include ifcvt.h.
33721 (ix86_option_override_internal): Don't override
33722 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33723 (ix86_noce_conversion_profitable_p): New function.
33724 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33725 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33726 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33727 * doc/tm.texi: Regenerated.
33729 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33731 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33734 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33737 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33738 instructions for small multiply cores.
33740 2017-04-04 Jeff Law <law@redhat.com>
33742 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33744 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33745 that are not set by the loop over the elements.
33747 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33750 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33751 int mode, convert_modes it to mode as unsigned, otherwise use
33752 lowpart_subreg to mode rather than SImode.
33753 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33754 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33755 Use DImode instead of SImode for the shift count operand.
33756 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33759 2017-04-04 Richard Biener <rguenther@suse.de>
33761 PR middle-end/80281
33762 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33763 arithmetic done for the negate or the plus. Simplify.
33764 (A - (-B) -> A + B): Likewise.
33765 * fold-const.c (split_tree): Make sure to not negate pointers.
33767 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33769 PR rtl-optimization/60818
33770 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33771 a compare of comparisons with the thing compared if this results
33772 in a different machine mode.
33774 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33776 * alias.c (base_alias_check): Fix typo in comment.
33777 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33778 * cgraphunit.c (symbol_table::compile): Likewise.
33779 * collect2.c (maybe_run_lto_and_relink): Likewise.
33780 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33781 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33782 * config/avr/avr.c (avr_map_op_t): Likewise.
33783 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33784 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33785 * config/epiphany/epiphany.md (movcc): Likewise.
33786 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33787 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33789 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33790 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33791 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33792 * config/sh/sh.c (sh_rtx_costs): Likewise.
33793 * fold-const.c (fold_truth_andor): Likewise.
33794 * genautomata.c (collapse_flag): Likewise.
33795 * gengtype.h (struct type::u::s): Likewise.
33796 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
33797 * input.c (FORMAT_AMOUNT): Likewise.
33798 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
33799 (known_aggs_to_agg_replacement_list): Likewise.
33800 * ipa-inline-analysis.c: Likewise.
33801 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
33802 * ipa-polymorphic-call.c
33803 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
33804 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
33805 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
33807 * modulo-sched.c (apply_reg_moves): Likewise.
33808 * omp-expand.c (build_omp_regions_1): Likewise.
33809 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
33810 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
33811 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
33812 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33813 * value-prof.c: Likewise.
33814 * var-tracking.c (val_reset): Likewise.
33816 2017-04-03 Richard Biener <rguenther@suse.de>
33818 PR tree-optimization/80275
33819 * fold-const.c (split_address_to_core_and_offset): Handle
33822 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
33824 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
33825 descriptors is at least equal to that of functions.
33827 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33829 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
33831 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33834 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
33835 (mov<IMOD4:mode>): New expander.
33836 (*mov<IMOD4:mode>_internal): New insn and split pattern.
33838 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
33840 PR rtl-optimization/79405
33841 * fwprop.c (propagations_left): New variable.
33842 (forward_propagate_into): Decrement it.
33843 (fwprop_init): Initialize it.
33844 (fw_prop): If the variable has reached zero, stop propagating.
33845 (fwprop_addr): Ditto.
33847 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33850 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
33851 a FUNCTION_DECL, pass it as decl instead of origin to
33854 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
33856 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
33859 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
33862 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
33863 TARGET_VSX_SMALL_INTEGER.
33865 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33867 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
33868 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
33870 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33872 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
33873 extraction from odd-numbered MSA register.
33875 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33877 PR middle-end/80173
33878 * expmed.c (store_bit_field_1): Don't attempt to create
33879 a word subreg out of hard registers wider than word if they
33880 have HARD_REGNO_NREGS of 1 for their mode.
33882 PR middle-end/80163
33883 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
33884 conversions to integer types wider than word and pointer.
33887 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
33888 (rtx_equal_for_cselib_p): Pass 0 to it.
33889 * cselib.c (cselib_hasher::equal): Likewise.
33890 (rtx_equal_for_cselib_1): Add depth argument. If depth
33891 is 128, don't look up VALUE locs and punt. Increment
33892 depth in recursive calls when walking VALUE locs.
33894 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
33896 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
33897 (make_gcov_file_name): Use the canonical path name for generating
33899 (read_line): Fix handling of files with ascii null bytes.
33901 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
33903 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
33904 to initialise a vector register instead
33905 of using a const_int.
33907 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33909 PR translation/80189
33910 * gimplify.c (omp_default_clause): Use %qs instead of %s in
33911 diagnostic messages.
33913 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
33916 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
33917 (dfp_diex_<mode>): Update mode of operand 1.
33918 * doc/extend.texi (dxex, dxexq): Document change to return type.
33919 (diex, diexq): Document change to argument type.
33921 2017-03-30 Martin Jambor <mjambor@suse.cz>
33924 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
33925 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
33926 it reflects the signature changes performed at the callee side.
33927 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
33928 to cgraph_build_function_type_skip_args.
33929 (build_function_decl_skip_args): Adjust call to the above function.
33931 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33934 * config/i386/sse.md
33935 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
33936 register as dest whenever it is a MEM not rtx_equal_p to the
33937 corresponding dup operand, and when forcing into reg move the
33938 reg into the memory afterwards.
33939 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
33940 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
33941 for the force_reg mode.
33942 (avx512vl_vextractf128<mode>): Use register as dest either
33943 always when a MEM, or when it is a MEM not rtx_equal_p to the
33944 corresponding dup operand, or even not when it is a CONST_VECTOR
33945 depending on the mode and lo vs. hi.
33946 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
33948 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
33949 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
33950 Likewise. Require that operands[2] is even.
33951 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
33952 Remove extraneous parens. Require that operands[2] is a multiple
33954 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
33955 operands[0] is a MEM if <mask_applied>, the predicates/constraints
33956 disallow memory then.
33958 2017-03-30 Richard Biener <rguenther@suse.de>
33960 PR tree-optimization/77498
33961 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
33962 to non-constants over backedges.
33964 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
33966 PR rtl-optimization/80233
33967 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
33968 as last_combined_insn. Do not test for BARRIER_P separately.
33970 2017-03-29 Andreas Schwab <schwab@suse.de>
33973 * calls.c (prepare_call_address): Convert funexp to Pmode before
33974 copying to temp reg.
33976 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33978 PR tree-optimization/80158
33979 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
33980 Handle possible future case of more than one alternate
33982 (replace_rhs_if_not_dup): Likewise.
33983 (replace_one_candidate): Likewise.
33985 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
33987 PR rtl-optimization/80193
33988 * ira.c (ira): Do not check allocation for LRA.
33990 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
33992 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
33993 (nvptx_output_simt_exit): Declare.
33994 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
33995 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
33996 (init_softstack_frame): Move initialization of crtl->is_leaf to...
33997 (nvptx_declare_function_name): ...here. Emit declaration of local
33998 memory space buffer for omp_simt_enter insn.
33999 (nvptx_output_unisimt_switch): New.
34000 (nvptx_output_softstack_switch): New.
34001 (nvptx_output_simt_enter): New.
34002 (nvptx_output_simt_exit): New.
34003 * config/nvptx/nvptx.h (struct machine_function): New fields
34004 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34005 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34006 (UNSPECV_SIMT_EXIT): Ditto.
34007 (omp_simt_enter_insn): New insn.
34008 (omp_simt_enter): New expansion.
34009 (omp_simt_exit): New insn.
34010 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34012 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34013 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34014 (expand_GOMP_SIMT_EXIT): New.
34015 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34016 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34017 (GOMP_SIMT_EXIT): Ditto.
34018 * target-insns.def (omp_simt_enter): New insn.
34019 (omp_simt_exit): Ditto.
34020 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34022 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34023 (lower_rec_input_clauses): Likewise.
34024 (lower_lastprivate_clauses): Handle SIMT privatization.
34026 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34027 (ompdevlow_adjust_simt_enter): New.
34028 (find_simtpriv_var_op): New.
34029 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34030 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34032 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34033 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34034 (copy_decl_for_dup_finish): Ditto.
34036 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34038 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34041 * config/i386/i386.c (ix86_option_override_internal): Always
34042 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34044 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34046 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34048 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34050 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34051 mark new edge's irreducible flag accordign to it.
34052 (vect_do_peeling): Check loop preheader edge's irreducible flag
34053 and pass it to function slpeel_add_loop_guard.
34055 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34057 PR tree-optimization/80218
34058 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34059 Update block frequencies and counts.
34061 2017-03-28 Richard Biener <rguenther@suse.de>
34063 PR tree-optimization/78644
34064 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34065 of a simplification result we may not use it at all.
34067 2017-03-28 Richard Biener <rguenther@suse.de>
34070 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34071 without arguments, generate default definition of a SSA name.
34073 2017-03-28 Richard Biener <rguenther@suse.de>
34075 PR middle-end/80222
34076 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34077 TYPE_REF_CAN_ALIAS_ALL references.
34078 * fold-const.c (fold_indirect_ref_1): Likewise.
34080 2017-03-28 Martin Liska <mliska@suse.cz>
34083 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34084 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34086 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34087 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34089 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34090 (EXTRA_SPECS): Define.
34091 (SUBTARGET_EXTRA_SPECS): Likewise.
34092 (SUBTARGET_CPP_SPEC): Likewise.
34093 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34094 SUBTARGET_EXTRA_SPECS.
34095 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34097 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34099 * config/arc/simdext.md (vst64_insn): Update pattern.
34100 (vld32wh_insn): Likewise.
34101 (vld32wl_insn): Likewise.
34102 (vld64_insn): Likewise.
34103 (vld32_insn): Likewise.
34105 2017-03-28 Marek Polacek <polacek@redhat.com>
34108 * fold-const.c (fold_comparison): Use protected_set_expr_location
34109 instead of SET_EXPR_LOCATION.
34111 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34113 * tree.c (add_expr): Avoid name lookup warning.
34115 2017-03-27 Jeff Law <law@redhat.com>
34117 PR tree-optimization/80216
34118 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34119 function name. Limit recursion depth.
34120 (record_temporary_equivalences): Corresponding changes.
34122 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34124 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34127 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34130 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34132 * cfgcleanup.c (reg_note_cfa_p): New array.
34133 (insns_have_identical_cfa_notes): New function.
34134 (old_insns_match_p): Don't cross-jump in between /f
34135 and non-/f instructions. If both i1 and i2 are frame related,
34136 verify all CFA notes, their order and content.
34138 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34141 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34142 HImode and SImode with zero extend to DImode to one insn.
34143 (bswap<mode>2_extenddi): Likewise.
34144 (bswapsi2_extenddi): Likewise.
34145 (bswaphi2_extendsi): Likewise.
34146 (bswaphi2): Combine bswap HImode and SImode into one insn.
34147 Separate memory insns from swapping register.
34148 (bswapsi2): Likewise.
34149 (bswap<mode>2): Likewise.
34150 (bswaphi2_internal): Delete, no longer used.
34151 (bswapsi2_internal): Likewise.
34152 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34153 store, and gpr<-gpr swap insns.
34154 (bswap<mode>2_store): Likewise.
34155 (bswaphi2_reg): Register only splitter, combine with the splitter.
34156 (bswaphi2 splitter): Likewise.
34157 (bswapsi2_reg): Likewise.
34158 (bswapsi2 splitter): Likewise.
34159 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34160 the insns into load, store, and register/register insns.
34161 (bswapdi2_ldbrx): Likewise.
34162 (bswapdi2_load): Likewise.
34163 (bswapdi2_store): Likewise.
34164 (bswapdi2_reg): Likewise.
34166 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34168 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34169 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34171 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34174 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34176 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34177 special handling for target option conflicts between dform
34178 options (-mpower9-dform, -mpower9-dform-vector,
34179 -mpower9-dform-scalar) and -mno-direct-move.
34181 2017-03-27 Richard Biener <rguenther@suse.de>
34183 PR tree-optimization/80181
34184 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34186 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34188 * config/arc/predicates.md (move_double_src_operand): Replace the
34189 call to move_double_src_operand with a call to address_operand.
34191 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34193 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34194 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34195 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34197 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34199 * config/arc/predicates.md (long_immediate_loadstore_operand):
34200 Consider scaled addresses cases.
34202 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34204 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34205 restored when in interrupt.
34206 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34207 doesn't have delay slot.
34209 2017-03-27 Richard Biener <rguenther@suse.de>
34212 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34213 inlined thunk clones.
34215 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34218 * asan.c (instrument_derefs): Copy over last operand from
34219 original COMPONENT_REF to the new COMPONENT_REF with
34220 DECL_BIT_FIELD_REPRESENTATIVE.
34221 * ubsan.c (instrument_object_size): Likewise.
34223 2017-03-27 Richard Biener <rguenther@suse.de>
34225 PR tree-optimization/80170
34226 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34227 sure DR/SCEV didnt fold in constants we do not see when looking
34228 at the reference base alignment.
34230 2017-03-27 Richard Biener <rguenther@suse.de>
34232 PR middle-end/80171
34233 * gimple-fold.c (fold_ctor_reference): Properly guard against
34234 NULL return value from canonicalize_constructor_val.
34236 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34239 * config/i386/i386.c (ix86_expand_builtin)
34240 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34241 flags reg setting and flags reg using instructions.
34242 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34243 clobbering instructions to zero extend op2.
34245 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34247 * doc/install.texi (Configuration) <--with-aix-soname>:
34248 Update link to AIX ld.
34250 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34252 PR rtl-optimization/80160
34253 PR rtl-optimization/80159
34254 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34255 reg_alternate_class into account.
34257 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34260 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34261 to consider in curr_insn_transform.
34263 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34265 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34266 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34267 and emit_mode_inner.
34269 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34271 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34272 argument to the overloaded builtin variants. Use the new flag to
34273 deprecate certain builtin variants.
34274 * config/s390/s390-builtin-types.def: Add new builtin types.
34275 * config/s390/s390-builtins.h: Support new flags field for
34276 overloaded builtins.
34277 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34278 (s390_macro_to_expand): Enable vector float data type.
34279 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34280 builtins by incrementing the __VEC__ version number.
34281 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34283 (s390_resolve_overloaded_builtin): Emit error messages depending
34284 on the builtin flags.
34285 * config/s390/s390.c (s390_expand_builtin): Support additional
34286 flags argument. Change error message to match the messages
34287 emitted in s390-c.c.
34288 * config/s390/s390.md: New UNSPEC_* constants.
34289 (op_type): Add new instruction types.
34290 * config/s390/vecintrin.h: Add new builtins and test data class
34292 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34293 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34294 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34295 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34296 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34297 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34299 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34300 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34301 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34302 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34304 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34305 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34306 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34307 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34309 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34310 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34311 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34312 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34313 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34314 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34315 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34317 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34318 ("vec_scatter_element<V_HW_4:mode>_DI")
34319 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34320 ("vec_fpint<mode>", "vflls")
34321 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34322 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34323 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34324 ("*vec_cmphe<mode>_cc"): ... these.
34326 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34327 mode constant instead of magic value.
34329 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34331 * config/s390/s390.c (s390_expand_vec_compare): Support other
34332 vector floating point modes than just V2DF.
34333 (s390_expand_vcond): Likewise.
34334 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34335 (s390_cannot_change_mode_class): Prevent mode changes between TF
34336 and V1TF in vector registers.
34337 * config/s390/s390.md (DF, SF): New mode attributes.
34338 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34339 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34340 SFmode support for VRs.
34341 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34343 (VFT, VF_HW): New mode iterators.
34344 (vw, sdx): New mode attributes.
34345 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34346 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34347 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34348 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34349 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34350 also the new vector floating point modes. Renaming to ...
34352 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34353 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34354 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34355 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34356 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34357 ("vec_unordered<mode>"): ... these.
34359 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34360 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34361 ("*vec_extendv2df"): New insn definitions.
34363 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34365 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34366 ("mulditi3_2", "*muldi3_sign"): New patterns.
34367 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34368 rename the pattern definition.
34370 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34372 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34374 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34376 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34378 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34379 instruction if possible.
34380 * config/s390/vector.md (vec_halfnumelts): New mode
34382 ("*vec_vllezlf<mode>"): New pattern.
34384 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34386 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34387 ("popcountv4si2", "popcountv2di2"): Rename to ...
34388 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34389 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34391 ("popcount<mode>2_vxe"): New pattern.
34393 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34395 * common/config/s390/s390-common.c (processor_flags_table): Add
34397 * config.gcc: Add arch12.
34398 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34399 Default to arch12 for unknown CPU model numbers.
34400 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34401 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34402 PROCESSOR_max sanity check.
34403 * config/s390/s390-opts.h (enum processor_type): Add
34405 * config/s390/s390.c (processor_table): Add arch12.
34406 (s390_expand_builtin): Add check for B_VXE flag.
34407 (s390_issue_rate): Add PROCESSOR_ARCH12.
34408 (s390_get_sched_attrmask): Likewise.
34409 (s390_get_unit_mask): Likewise.
34410 (s390_sched_score): Enable z13 scheduling for arch12.
34411 (s390_sched_reorder): Likewise.
34412 (s390_sched_variable_issue): Likewise.
34413 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34415 (s390_tune_attr): Use z13 scheduling also for arch12.
34416 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34417 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34418 (TARGET_VXE_P): New macros.
34419 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34420 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34421 * config/s390/s390.opt: Add arch12 as processor_type.
34423 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34425 * config/s390/s390.md
34426 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34427 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34428 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34430 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34431 Rename expanders to ...
34433 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34434 ("fixuns_truncdddi2_emu"): ... these.
34436 ("fixuns_trunc<mode>si2_emu"): New expander.
34438 ("*fixuns_truncdfdi2_z13"): Rename to ...
34439 ("*fixuns_truncdfdi2_vx"): ... this.
34441 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34443 * config/s390/2964.md: Remove the single element vector compare
34444 instructions which are no longer used.
34445 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34447 (s390_canonicalize_comparison): Remove handling of DFmode
34449 (s390_expand_vec_compare_scalar): Remove function.
34450 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34451 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34453 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34455 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34457 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34459 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34460 will anyway by matched by mov<mode>_64dfp.
34462 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34464 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34465 vlef/vstef. Add missing operand to vleif.
34467 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34469 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34470 pair for all vector types with 64 bit elements.
34471 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34472 * config/s390/vector.md (V_HW_64): ... here.
34473 (V_128_NOSINGLE): New mode iterator.
34474 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34475 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34476 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34477 ("*vec_load_pairv2di"): Change to ...
34478 ("*vec_load_pair<mode>"): ... this one.
34480 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34482 * config/s390/constraints.md: Add comments.
34483 (jKK): Reject element sizes > 8 bytes.
34484 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34486 * config/s390/s390.md: Add the s_operand checks formerly in
34487 s390_split_ok_p to various splitters where they are still
34489 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34490 for 128 bit vectors. Plus two splitters.
34492 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34494 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34497 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34500 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34501 error if the boundary argument is not constant.
34503 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34505 PR rtl-optimization/80112
34506 * loop-doloop.c (doloop_condition_get): Don't check condition
34507 if cmp isn't SET with IF_THEN_ELSE src.
34509 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34511 PR tree-optimization/80158
34512 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34513 replacing a candidate statement, also replace it for the
34514 candidate's alternate interpretation.
34515 (replace_rhs_if_not_dup): Likewise.
34516 (replace_one_candidate): Likewise.
34518 2017-03-24 Richard Biener <rguenther@suse.de>
34520 PR tree-optimization/80167
34521 * graphite-isl-ast-to-gimple.c
34522 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34524 (translate_isl_ast_to_gimple::get_rename): Likewise.
34526 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34528 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34529 handling of certain combinations of target options, including the
34530 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34531 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34533 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34536 * config/arm/arm.md (*load_multiple): Add reload_completed to
34537 matching condition.
34539 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34540 Richard Biener <rguenth@suse.de>
34542 PR tree-optimization/79908
34543 PR tree-optimization/80136
34544 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34545 been cast away, gimplify_and_add suffices.
34547 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34549 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34551 2017-03-23 Richard Biener <rguenther@suse.de>
34553 PR tree-optimization/80032
34554 * gimplify.c (gimple_push_cleanup): Forced unconditional
34555 cleanups still have to go to the conditional_cleanups
34558 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34560 PR tree-optimization/80072
34561 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34563 (next_operand_entry_id): Change type to unsigned int.
34564 (sort_by_operand_rank): Make sure to return the right return value
34565 even if unsigned fields are bigger than INT_MAX.
34566 (struct oecount): Change cnt and id type to unsigned int.
34567 (oecount_hasher::equal): Formatting fix.
34568 (oecount_cmp): Make sure to return the right return value
34569 even if unsigned fields are bigger than INT_MAX.
34570 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34573 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34574 TREE_READONLY on result if writing it more than once.
34577 * doc/invoke.texi (-fsanitize=thread): Document that with
34578 -fnon-call-exceptions atomics are not able to throw
34582 * tsan.c: Include tree-eh.h.
34583 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34584 maybe_clean_or_replace_eh_stmt where needed.
34585 (instrument_memory_accesses): Add cfg_changed argument.
34586 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34587 if it returned true.
34588 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34590 PR rtl-optimization/63191
34591 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34592 wrapper function, moved the whole old content into ...
34593 (ix86_delegitimize_address_1): ... this. New inline function.
34594 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34595 true as last argument instead of ix86_delegitimize_address.
34597 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34599 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34600 cortexa57_branch_cost.
34602 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34604 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34606 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34609 * doc/md.texi (Constraints): Document wA constraint.
34610 * config/rs6000/constraints.md (wA): New.
34611 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34612 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34613 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34614 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34616 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34619 * gimplify.c (is_oacc_declared): New function.
34620 (oacc_default_clause): Use it to set default flags for acc declared
34621 variables inside parallel regions.
34622 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34623 declared variables.
34624 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34625 declare attribute to any decl as necessary.
34627 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34630 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34631 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34632 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34633 (arm_arch_lpae): This.
34634 * config/arm/arm.c (arm_arch7ve): Rename into ...
34635 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34636 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34639 2017-03-22 Martin Liska <mliska@suse.cz>
34642 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34643 error message instead of an ICE.
34645 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34647 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34649 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34651 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34653 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34656 2017-03-21 Martin Sebor <msebor@redhat.com>
34658 * doc/extend.texi: Use "cannot" instead of "can't."
34659 * doc/hostconfig.texi: Same.
34660 * doc/install.texi: Same.
34661 * doc/invoke.texi: Same.
34662 * doc/loop.texi: Same.
34663 * doc/md.texi: Same.
34664 * doc/objc.texi: Same.
34665 * doc/rtl.texi: Same.
34666 * doc/tm.texi: Same.
34667 * doc/tm.texi.in: Same.
34668 * doc/trouble.texi: Same.
34670 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34673 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34674 (collect_checksum_attributes): Set it.
34675 (die_checksum_ordered): Use it.
34677 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34679 PR tree-optimization/79908
34680 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34681 change: For a VA_ARG whose LHS has been cast away, use
34682 force_gimple_operand to construct the side effects.
34684 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34686 PR translation/80001
34687 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34688 more amenable to translation.
34689 (oacc_loop_auto_partitions): Likewise.
34691 2017-03-21 Marek Polacek <polacek@redhat.com>
34692 Martin Sebor <msebor@redhat.com>
34694 PR tree-optimization/80109
34695 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34696 on INTEGRAL_TYPE_P.
34698 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34699 Segher Boessenkool <segher@kernel.crashing.org>
34702 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34703 check reg_used_between_p between insn and one of succ or succ2
34704 depending on if succ is artificial insn not inserted into insn
34707 2017-03-21 Martin Liska <mliska@suse.cz>
34709 PR gcov-profile/80081
34710 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34711 * doc/gcc.texi: Include gcov-dump stuff.
34712 * doc/gcov-dump.texi: New file.
34714 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34716 PR rtl-optimization/79150
34717 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34718 conditional jump, if the jump is the last insn of the loop.
34720 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34721 Richard Biener <rguenth@suse.de>
34723 PR tree-optimization/79908
34724 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34725 been cast away, use force_gimple_operand to construct the side
34728 2017-03-21 Martin Liska <mliska@suse.cz>
34730 PR libfortran/79956
34731 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34734 2017-03-21 Brad Spengler <spender@grsecurity.net>
34737 * plugin.c (htab_hash_plugin): New function.
34738 (add_new_plugin): Use it and adjust.
34739 (parse_plugin_arg_opt): Adjust.
34740 (init_one_plugin): Likewise.
34742 2017-03-21 Richard Biener <rguenther@suse.de>
34744 PR tree-optimization/80032
34745 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34746 if set force the cleanup to happen unconditionally.
34747 (gimplify_target_expr): Push inserted clobbers with force_uncond
34748 to avoid them being removed by control-dependent DCE.
34750 2017-03-21 Richard Biener <rguenther@suse.de>
34752 PR tree-optimization/80122
34753 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34754 va_arg_pack_len when the inlined call stmt requires pack
34756 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34758 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34761 * tsan.c (instrument_builtin_call): If the memory model argument
34762 is not a constant, assume it is valid.
34765 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34768 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34770 PR rtl-optimization/79910
34771 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34772 if its dest is used by an insn before I2 (other than the combined
34773 insns themselves, which are properly handled already).
34775 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34778 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34780 * combine.c (record_used_regs): New static function.
34781 (try_combine): Handle situations where there is an additional
34782 instruction between I2 and I3 which needs to have a LOG_LINK
34786 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34788 * combine.c (try_combine): Delete redundant i1 test. Call
34789 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34791 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34794 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
34795 alternatives 13/14.
34797 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34799 PR tree-optimization/80054
34800 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
34801 the optimization if a PHI or any of its arguments is not dominated
34802 by the candidate's basis. Use gphi* rather than gimple* as
34804 (replace_profitable_candidates): Clean up a gimple* variable that
34805 should be a gphi* variable.
34807 2017-03-20 Martin Sebor <msebor@redhat.com>
34810 * doc/extend.texi (attribute constructor): Document present limitation.
34812 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
34815 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
34816 __POWER9_VECTOR__ #ifdef control, change template definition to
34817 use Power9-specific built-in function.
34818 (vec_any_eq): Likewise.
34819 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
34820 to control outcomes from this test.
34821 (vector_ae_<mode>p): For VEC_F modes, likewise.
34823 2017-03-20 Ian Lance Taylor <iant@google.com>
34825 * config/i386/i386.c (ix86_function_regparm): Save an extra
34826 register for -fsplit-stack with DECL_STATIC_CHAIN.
34828 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34831 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
34832 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
34834 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34836 * config/riscv/riscv.c (riscv_print_operand): Use "fence
34838 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
34841 2017-03-20 Marek Polacek <polacek@redhat.com>
34844 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
34846 2017-03-20 Richard Biener <rguenther@suse.de>
34848 PR tree-optimization/80113
34849 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
34850 allocate extra SSA name for PHI def.
34851 (add_close_phis_to_outer_loops): Likewise.
34852 (add_close_phis_to_merge_points): Likewise.
34853 (copy_loop_close_phi_args): Likewise.
34854 (copy_cond_phi_nodes): Likewise.
34856 2017-03-20 Martin Liska <mliska@suse.cz>
34858 PR middle-end/79753
34859 * tree-chkp.c (chkp_build_returned_bound): Do not build
34860 returned bounds for a LHS that's not a BOUNDED_P type.
34862 2017-03-20 Martin Liska <mliska@suse.cz>
34866 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
34867 COMPLEX_CST and VECTOR_CST.
34869 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34872 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
34873 target operand. A new splitter adds the clobber statement in case
34874 the target operand is dead anyway.
34876 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
34878 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
34879 to age-old versions of binutils and glibc.
34881 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
34883 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
34885 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34887 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
34889 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34891 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
34892 requirement for binutils 2.13.
34894 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34896 * combine.c (try_combine): Delete redundant i1 test. Call
34897 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34899 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
34901 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
34902 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
34904 <riscv64-*-elf>: Re-arrange section
34905 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
34906 <riscv32-*-linux>: Likewise.
34907 <riscv64-*-elf>: Likewise
34908 <riscv64-*-linux>: Likewise.
34910 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
34913 * aarch64.opt(verbose-cost-dump): Fix typo.
34915 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
34918 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
34919 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
34921 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34923 * reload.c (find_reloads): When reloading a nonoffsettable address,
34924 use RELOAD_OTHER for it and its address reloads.
34926 PR rtl-optimization/79910
34927 * combine.c (record_used_regs): New static function.
34928 (try_combine): Handle situations where there is an additional
34929 instruction between I2 and I3 which needs to have a LOG_LINK
34932 2017-03-17 Jeff Law <law@redhat.com>
34934 PR tree-optimization/71437
34935 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
34936 conditional in the hash table first.
34937 (vrp_dom_walker::before_dom_children): Extract condition from
34938 ASSERT_EXPR. Record condition, its inverion and any implied
34939 conditions as well.
34941 2017-03-17 Marek Polacek <polacek@redhat.com>
34942 Markus Trippelsdorf <markus@trippelsdorf.de>
34944 PR tree-optimization/80079
34945 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
34948 2017-03-17 Richard Biener <rguenther@suse.de>
34950 PR middle-end/80075
34951 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
34952 Properly verify the LHS before the RHS possibly claims to be
34954 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
34957 2017-03-17 Martin Jambor <mjambor@suse.cz>
34959 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
34960 (List of -O2 options): Likewise.
34961 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
34964 2017-03-17 Tom de Vries <tom@codesourcery.com>
34966 * gcov-dump.c (print_usage): Print bug_report_url.
34968 2017-03-17 Richard Biener <rguenther@suse.de>
34970 PR middle-end/80050
34971 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
34972 (parser::peek): Likewise.
34974 2017-03-17 Richard Biener <rguenther@suse.de>
34976 PR tree-optimization/80048
34977 * sese.c (free_sese_info): Properly release rename_map and
34978 copied_bb_map elements.
34980 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
34982 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
34983 Add linked-list forward and backlinks. Insert on
34984 construction, remove on destruction.
34985 (class pass_store_merging): Add m_stores_head field.
34986 (pass_store_merging::terminate_and_process_all_chains):
34987 Iterate over m_stores_head list.
34988 (pass_store_merging::terminate_all_aliasing_chains):
34990 (pass_store_merging::execute): Check for debug stmts first.
34991 Push new chains onto the m_stores_head stack.
34993 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
34996 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
34997 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
34998 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35000 2017-03-16 Jeff Law <law@redhat.com>
35002 PR tree-optimization/71437
35003 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35004 member function. Implementation moved into after_dom_children
35005 member function and into the threader's thread_outgoing_edges
35007 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35008 some code into new thread_outgoing_edges.
35009 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35010 definition. Simplify marker handling (do it here). Assume we always
35011 have the available expression and the const/copies tables.
35012 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35014 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35015 * tree-vrp.c (equiv_stack): No longer file scoped.
35016 (vrp_dom_walker): New class.
35017 (vrp_dom_walker::before_dom_children): New member function.
35018 (vrp_dom_walker::after_dom_children): Likewise.
35019 (identify_jump_threads): Setup domwalker. Use it rather than
35020 walking edges in a random order by hand. Simplify setup/finalization.
35021 (finalize_jump_threads): Remove.
35022 (vrp_finalize): Do not call identify_jump_threads here.
35023 (execute_vrp): Do it here instead and call thread_through_all_blocks
35026 PR tree-optimization/71437
35027 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35029 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35031 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35032 (dom_opt_dom_walker::thread_across_edge): Remove
35033 handle_dominating_asserts argument. All callers changed.
35034 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35035 changes. Remove calls to lhs_of_dominating_assert. Other
35036 uses of handle_dominating_asserts turn into unconditional code
35037 (simplify_control_stmt_condition_1): Likewise.
35038 (simplify_control_stmt_condition): Likewise.
35039 (thread_through_normal_block, thread_across_edge): Likewise.
35040 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35041 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35042 object if it is not an SSA_NAME.
35043 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35044 before calling into the VRP specific simplifiers.
35045 (identify_jump_threads): Remove handle_dominating_asserts
35048 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35051 * tree-diagnostic.c (default_tree_printer): No longer static.
35052 * tree-diagnostic.h (default_tree_printer): New prototype.
35054 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35056 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35057 Change ins into fmov.
35059 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35061 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35062 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35063 Use h_con constraint for operand 1.
35064 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35065 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35067 2017-03-15 Jeff Law <law@redhat.com>
35069 PR tree-optimization/71437
35070 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35071 (record_temporary_equivalences): Use it.
35073 PR tree-optimization/71437
35074 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35075 tree-ssa-scopedtables.
35076 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35077 (record_conditions, record_cond, vuse_eq): Likewise.
35078 (record_edge_info): Adjust to API tweak of record_conditions.
35079 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35080 (record_temporary_equivalences, optimize_stmt): Likewise.
35081 (eliminate_redundant_computations): Likewise.
35082 (record_equivalences_from_stmt): Likewise.
35083 * tree-ssa-scopedtables.c: Include options.h and params.h.
35084 (vuse_eq): New function, moved from tree-ssa-dom.c
35085 (build_and_record_new_cond): Likewise.
35086 (record_conditions): Likewise. Accept vector of conditions rather
35087 than edge_equivalence structure for first argument.
35088 for the first argument.
35089 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35090 from tree-ssa-dom.c.
35091 (avail_exprs_stack::record_cond): Likewise.
35092 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35093 from tree-ssa-dom.c.
35094 (avail_exprs_stack): Add new member functions lookup_avail_expr
35096 (record_conditions): Declare.
35098 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35101 * lra-constraints.c (process_alt_operands): Increase reject for
35102 reloading an input/output operand.
35104 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35107 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35108 insns to convert from signed/unsigned char/short to IEEE 128-bit
35110 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35112 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35115 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35116 subreg of inner mode for values already in registers.
35118 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35120 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35121 iteration reg is used after the loop.
35123 2017-03-14 Martin Sebor <msebor@redhat.com>
35125 PR tree-optimization/79800
35126 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35127 precision in negative-positive range.
35128 (format_floating): Call non-const overload with adjusted precision.
35130 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35133 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35136 2017-03-14 Martin Sebor <msebor@redhat.com>
35138 PR middle-end/80020
35139 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35140 * builtins.def (aligned_alloc): Use it.
35143 * Makefile.in (GTFILES): Add calls.c.
35144 * calls.c: Include "gt-calls.h".
35146 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35148 PR rtl-optimization/79728
35149 * regs.h (struct target_regs): New field
35150 x_contains_allocatable_regs_of_mode.
35151 (contains_allocatable_regs_of_mode): New macro.
35152 * reginfo.c (init_reg_sets_1): Initialize it, and change
35153 contains_reg_of_mode so it includes global regs as well.
35154 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35155 rather than contains_regs_of_mode.
35157 2017-03-14 Martin Liska <mliska@suse.cz>
35159 * doc/invoke.texi: Document options that can't be combined with
35160 -fcheck-pointer-bounds.
35162 2017-03-14 Martin Liska <mliska@suse.cz>
35164 PR middle-end/79831
35165 * doc/invoke.texi (-Wchkp): Document the option.
35167 2017-03-14 Martin Liska <mliska@suse.cz>
35169 * Makefile.in: Install gcov-dump.
35171 2017-03-14 Martin Liska <mliska@suse.cz>
35173 * multiple_target.c (expand_target_clones): Bail out for
35174 an invalid attribute.
35176 2017-03-14 Richard Biener <rguenther@suse.de>
35178 * alias.c (struct alias_set_entry): Pack properly.
35179 * cfgloop.h (struct loop): Likewise.
35180 * cse.c (struct set): Likewise.
35181 * ipa-utils.c (struct searchc_env): Likewise.
35182 * loop-invariant.c (struct invariant): Likewise.
35183 * lra-remat.c (struct cand): Likewise.
35184 * recog.c (struct change_t): Likewise.
35185 * rtl.h (struct address_info): Likewise.
35186 * symbol-summary.h (function_summary): Likewise.
35187 * tree-loop-distribution.c (struct partition): Likewise.
35188 * tree-object-size.c (struct object_size_info): Likewise.
35189 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35190 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35191 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35192 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35193 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35194 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35195 (struct _stmt_vec_info): Likewise.
35197 2017-03-14 Martin Liska <mliska@suse.cz>
35200 * multiple_target.c (create_dispatcher_calls): Check that
35201 a target can create a function dispatcher.
35203 2017-03-14 Martin Liska <mliska@suse.cz>
35206 * multiple_target.c (expand_target_clones): Drop local.local
35207 flag for default implementation.
35209 2017-03-14 Richard Biener <rguenther@suse.de>
35211 PR tree-optimization/80030
35212 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35214 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35216 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35217 gcc_fallthrough() instead of __attribute__((fallthrough));
35219 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35221 * doc/gcc.texi: Remove "up" link to (DIR).
35222 * doc/gccint.texi: Ditto.
35224 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35226 * doc/install.texi (Specific) <avr>: Remove reference to
35229 2017-03-13 Jeff Law <law@redhat.com>
35231 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35232 attribute rather than comments.
35234 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35235 match_scratch operand is highest.
35237 2017-03-13 Martin Liska <mliska@suse.cz>
35239 PR middle-end/78339
35240 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35241 is a CHKP clone, use original declaration.
35243 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35245 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35246 (arc_conditional_register_usage): Use a different allocation order
35247 when optimizing for size.
35248 * common/config/arc/arc-common.c (arc_option_optimization_table):
35249 Section anchors default on when optimizing for size.
35251 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35253 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35255 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35257 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35258 * config/arc/arc.md (cpu_facility): Add cd variant.
35259 (*movqi_insn): Add code density variant.
35260 (*movhi_insn): Likewise.
35261 (*movqi_insn): Likewise.
35262 (*addsi3_mixed): Likewise.
35263 (subsi3_insn): Likewise.
35265 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35267 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35269 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35271 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35272 expressions with MINUS and UNARY ops.
35274 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35277 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35279 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35280 between vec_select and vector argument.
35281 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35282 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35283 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35284 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35285 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35286 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35288 2017-03-13 Richard Biener <rguenther@suse.de>
35291 * params.def (vect-max-peeling-for-alignment): Fix typo.
35293 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35295 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35296 issue that only occurred with binutils below 2.18.
35298 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35300 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35301 refer to binutils 2.11/2.12 minimum.
35303 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35305 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35306 ftp.kernel.org and simplify binutils requirement.
35308 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35310 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35312 (Optimize Options): Ditto. Also remove redundancy.
35314 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35316 PR translation/79848
35317 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35319 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35320 to G_ to avoid double translation.
35322 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35324 PR translation/79923
35325 * auto-profile.c (get_combined_location): Convert leading
35326 character of diagnostics to lower case and remove trailing period.
35327 (read_profile): Likewise for various diagnostics.
35328 * config/arm/arm.c (arm_option_override): Remove trailing period
35329 from various diagnostics.
35330 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35331 (msp430_expand_delay_cycles): Likewise.
35333 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35336 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35337 full command-line argument, rather than just "str".
35338 (aarch64_validate_march): Likewise.
35339 (aarch64_validate_mtune): Likewise.
35341 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35343 PR rtl-optimization/78911
35344 * lra-assigns.c (must_not_spill_p): New function.
35345 (spill_for): Use it.
35347 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35349 PR tree-optimization/79981
35350 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35351 ATOMIC_COMPARE_EXCHANGE ifn result.
35352 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35353 IFN_ATOMIC_COMPARE_EXCHANGE.
35355 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35358 * opts.c (parse_sanitizer_options): Add missing question mark to
35359 "did you mean" message.
35361 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35363 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35365 (VMULEUH_UNS): Likewise.
35366 (VMULOUB_UNS): Likewise.
35367 (VMULOUH_UNS): Likewise.
35368 * config/rs6000/rs6000.c (builtin_function_type): Remove
35369 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35371 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35374 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35375 x with result of extra_parsing_for_operand_code_0.
35376 (function_reader::extra_parsing_for_operand_code_0): Convert
35377 return type from void to rtx, returning x. When reading
35378 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35379 larger size containing struct block_symbol.
35381 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35383 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35384 -mfloat128-hardware without -m64.
35386 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35389 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35390 entries to the case statement that marks unsigned arguments to
35391 overloaded functions.
35393 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35395 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35396 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35398 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35401 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35402 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35404 2017-03-10 Martin Liska <mliska@suse.cz>
35408 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35409 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35412 2017-03-10 Olivier Hainque <hainque@adacore.com>
35414 * tree-switch-conversion (array_value_type): Start by resetting
35415 candidate type to it's main variant.
35417 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35419 PR rtl-optimization/79909
35420 * combine.c (try_combine): Use simplify_replace_rtx on individual
35421 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35422 of the whole CALL_INSN_FUNCTION_USAGE.
35424 PR tree-optimization/79972
35425 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35426 get_range_info on SSA_NAMEs. Formatting fixes.
35428 2017-03-10 Richard Biener <rguenther@suse.de>
35429 Jakub Jelinek <jakub@redhat.com>
35431 PR tree-optimization/77975
35432 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35433 edge to be constant.
35434 (get_val_for): For constant x return it. Formatting fix.
35435 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35436 would use the same bases as the current one.
35438 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35440 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35441 instead of vec_select for V1TImode.
35442 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35444 (VSX_LE_128): Add V1TI to this mode iterator.
35445 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35446 (*vsx_le_perm_store_<mode>): Likewise.
35447 (pre-reload splitter for VSX stores): Likewise.
35448 (post-reload splitter for VSX stores): Likewise.
35449 (*vsx_xxpermdi2_le_<mode>): Likewise.
35450 (*vsx_lxvd2x2_le_<mode>): Likewise.
35451 (*vsx_stxvd2x2_le_<mode>): Likewise.
35453 2017-03-09 Michael Eager <eager@eagercon.com>
35455 Correct failures with --enable-checking=yes,rtl.
35457 * config/microblaze/microblaze.c (microblaze_expand_shift):
35458 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35459 test for const0_rtx.
35460 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35461 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35463 2017-03-09 Richard Biener <rguenther@suse.de>
35465 PR tree-optimization/79977
35466 * graphite-scop-detection.c (scop_detection::merge_sese):
35467 Handle the case of extra exits to blocks dominating the entry.
35469 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35471 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35474 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35476 PR rtl-optimization/79949
35477 * lra-constraints.c (process_alt_operands): Check memory when
35478 trying to predict a cycle. Print about the overall increase.
35480 2017-03-09 Richard Biener <rguenther@suse.de>
35482 PR middle-end/79971
35483 * gimple-expr.c (useless_type_conversion_p): Preserve
35484 TYPE_SATURATING for fixed-point types.
35486 2017-03-09 Richard Biener <rguenther@suse.de>
35489 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35490 alignment of BLKmode params.
35492 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35495 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35496 (VALL_NO_V2Q): Likewise.
35498 * config/aarch64/aarch64-simd.md
35499 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35501 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35502 VALL_NO_V2Q mode iterator.
35503 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35505 2017-03-09 Martin Liska <mliska@suse.cz>
35507 PR tree-optimization/79631
35508 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35509 tree_int_cst_sign_bit just for INTEGER constants.
35511 2017-03-09 Martin Liska <mliska@suse.cz>
35515 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35518 2017-03-09 Marek Polacek <polacek@redhat.com>
35521 * tree.c (inchash::add_expr): Handle TREE_VEC.
35523 2017-03-09 Martin Liska <mliska@suse.cz>
35526 (chkp_narrow_size_and_offset): New function.
35527 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35528 (void chkp_parse_bit_field_ref): New function.
35529 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35530 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35532 2017-03-09 Martin Liska <mliska@suse.cz>
35535 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35536 (chkp_find_bounds_1): Remove gcc_unreachable.
35538 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35541 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35542 BUILT_IN_SYNC*, determine the access type from the size suffix and
35543 always build a MEM_REF with that type. Handle forgotten
35544 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35547 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35548 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35549 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35550 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35551 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35552 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35553 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35554 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35555 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35556 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35557 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35558 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35559 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35560 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35561 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35562 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35563 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35564 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35565 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35566 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35567 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35568 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35569 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35570 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35571 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35572 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35573 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35574 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35575 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35576 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35577 definitions outside of __OPTIMIZE__ guarded section.
35580 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35581 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35582 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35583 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35586 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35588 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35589 ("vfenez<mode>"): Add missing constraints.
35591 2017-03-08 Martin Sebor <msebor@redhat.com>
35594 * config/nds32/nds32.c (nds32_option_override):
35595 Fix misspelled diagnostic.
35597 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35600 * gimplify.c (gimplify_omp_for): Replace index var in outer
35601 taskloop statement with an artificial variable and add
35602 OMP_CLAUSE_PRIVATE clause for it.
35604 2017-03-08 Richard Biener <rguenther@suse.de>
35606 PR tree-optimization/79955
35607 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35608 for accesses that are completely outside of the variable.
35610 2017-03-08 Andrew Haley <aph@redhat.com>
35612 PR tree-optimization/79943
35613 * tree-ssa-loop-split.c (compute_new_first_bound): When
35614 calculating the new upper bound, (END-BEG) should be added, not
35617 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35619 * config/avr/avr.md (setmemhi): Make sure match_dup
35620 operand number comes before match_scratch.
35622 2017-03-08 Richard Biener <rguenther@suse.de>
35624 PR tree-optimization/79920
35625 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35626 with ncopies == 1 to ...
35627 (vect_transform_slp_perm_load): ... here. Properly compute
35628 all element loads by iterating VF times over the group. Do
35629 not handle ncopies (computed in a broken way) in
35630 vect_create_mask_and_perm.
35632 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35635 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35636 is a uniform vector, use uniform_vector_p return value instead of
35637 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35639 2017-03-07 Marek Polacek <polacek@redhat.com>
35641 PR middle-end/79809
35642 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35643 (alloca_call_type): Likewise.
35645 2017-03-07 Martin Liska <mliska@suse.cz>
35647 * gcov.c (process_args): Put comment to correct location.
35649 2017-03-07 Martin Liska <mliska@suse.cz>
35651 PR middle-end/68270
35652 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35653 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35654 (chkp_narrow_bounds_for_field): Likewise.
35655 (chkp_parse_array_and_component_ref): Pass one more argument to
35658 2017-03-07 Richard Biener <rguenther@suse.de>
35660 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35663 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35665 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35666 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35668 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35671 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35672 to end of description.
35673 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35675 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35677 PR rtl-optimization/79901
35678 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35680 (*avx512f_<code><mode>3<mask_name>): ... this.
35681 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35682 iterator instead of VI8_AVX2_AVX512BW.
35684 PR rtl-optimization/79901
35685 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35686 min/max expander, expand it using expand_vec_cond_expr.
35689 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35692 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35695 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35696 to void * for PCH reasons.
35697 * dwarf2out.c (output_loc_operands, output_die): Cast
35698 v.val_vec.array to unsigned char *.
35700 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35703 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35706 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35708 PR rtl-optimization/79571
35709 * lra-constraints.c (process_alt_operands): Calculate static
35710 reject and subtract it from overall when only addresses will be
35713 2017-03-06 Julia Koval <julia.koval@intel.com>
35716 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35717 incoming stack boundary to 128 for 64-bit targets.
35719 2017-03-06 Richard Biener <rguenther@suse.de>
35721 PR tree-optimization/79894
35722 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35723 to NULL after folding it.
35725 2017-03-06 Richard Biener <rguenther@suse.de>
35727 PR tree-optimization/79824
35728 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35729 check disabling peeling for gaps.
35731 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35733 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35734 attributes): Document gettimeofday.
35736 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35738 * config/s390/s390.c (s390_option_override_internal): Set
35739 PARAM_MIN_VECT_LOOP_BOUND
35741 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35743 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35744 * config/s390/s390.md: Likewise.
35746 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35749 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35750 (<avx2_avx512>_perm<mode>): Rename to ...
35751 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35753 (<avx512>_perm<mode>_mask): Rename to ...
35754 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35756 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35757 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35758 instead of VI8F_256_512.
35759 (avx512f_perm<mode>): New define_expand.
35760 (avx512f_perm<mode>_mask): Likewise.
35761 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35762 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35764 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35766 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35767 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35769 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35771 2017-03-06 Martin Liska <mliska@suse.cz>
35774 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35775 when having a SSA NAME w/o VAR_DECL assigned to it.
35777 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35779 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35780 msa_dpsub_<su>_d): Fix MODE for vec_select.
35782 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35784 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35786 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35788 2017-03-06 Richard Biener <rguenther@suse.de>
35790 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35791 * plugin.c (register_plugin_info): Likewise.
35792 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
35794 2017-03-05 Jakub Jelinek <jakub@redhat.com>
35796 * config/i386/sse.md (sse_storehps, sse_storelps,
35797 avx_<castmode><avxsizesuffix>_<castmode>,
35798 avx512f_<castmode><avxsizesuffix>_<castmode>,
35799 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
35800 in condition that at least one operand is not a MEM.
35802 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35804 PR middle-end/79805
35805 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
35806 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
35808 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
35809 gimple_call_nothrow_p flag based on whether original builtin can throw.
35810 If it can, emit following stmts on the fallthrough edge.
35811 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
35812 don't create new bb if inserting just debug stmts on the edge, try to
35813 insert them on the fallthru bb or just reset debug stmts.
35815 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
35818 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
35819 restore recog_data (including the operand rtxes inside it) around
35820 the call to get_insn_template.
35822 2017-03-03 Martin Sebor <msebor@redhat.com>
35824 PR tree-optimization/79699
35825 * context.c (context::~context): Free MPFR caches to avoid
35826 a memory leak on program exit.
35828 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35830 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
35831 Use wide_int::ulow () instead of .elt (0).
35833 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35835 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
35836 (*pushxf): Limit oF constraint to 32bit targets and add oC
35837 constraint for 64bit targets.
35838 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
35839 (*pushdf): Change rmF constraint to rmC.
35841 2017-03-03 Martin Liska <mliska@suse.cz>
35843 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
35844 Remove unused variable.
35846 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35849 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
35850 is a memory operand, increase num_memory.
35851 (ix86_expand_args_builtin): Likewise.
35853 2017-03-03 Jan Hubicka <jh@suse.cz>
35856 * ipa-devirt.c (maybe_record_node): Properly handle
35857 __cxa_pure_virtual visibility.
35859 2017-03-03 Martin Liska <mliska@suse.cz>
35861 PR tree-optimization/79803
35862 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
35864 (pass_loop_prefetch::execute): Disabled optimization if an
35865 assumption about L1 cache size is not met.
35867 2017-03-03 Martin Liska <mliska@suse.cz>
35869 PR rtl-optimization/79574
35870 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
35871 (hash_scan_set): Likewise.
35872 (dump_hash_table): Likewise.
35873 (hoist_code): Likewise.
35875 2017-03-03 Richard Biener <rguenther@suse.de>
35877 * fixed-value.c (fixed_from_string): Restore use of elt (1)
35878 in place of uhigh ().
35879 (fixed_convert_from_real): Likewise.
35881 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35884 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
35886 2017-03-03 Richard Biener <rguenther@suse.de>
35888 PR middle-end/79818
35889 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
35890 TYPE_OVERFLOW_UNDEFINED check.
35892 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35894 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
35896 (vector_ae_<mode>_p): Likewise.
35897 (vector_nez_<mode>_p): Likewise.
35898 (vector_ne_v2di_p): Likewise.
35899 (vector_ae_v2di_p): Likewise.
35900 (vector_ne_<mode>_p): Likewise.
35901 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
35903 (vsx_tsqrt<mode>2_fe): Likewise.
35905 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
35908 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
35910 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35912 PR rtl-optimization/79780
35913 * cprop.c (one_cprop_pass): When second and further conditional trap
35914 in a single basic block is turned into an unconditional trap, turn it
35915 into a deleted note to avoid RTL verification failures.
35917 2017-03-02 Richard Biener <rguenther@suse.de>
35919 * fold-const.c (const_binop): Use ulow () instead of elt (0).
35921 2017-03-02 Richard Biener <rguenther@suse.de>
35923 PR tree-optimization/79345
35925 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
35926 param and abort the walk, returning -1 if it is hit.
35927 (walk_aliased_vdefs): Take a limit param and pass it on.
35928 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
35929 defaulting to 0 and return a signed int.
35930 * tree-ssa-uninit.c (struct check_defs_data): New struct.
35931 (check_defs): New helper.
35932 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
35933 about uninitialized memory.
35934 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
35935 bogus uninitialized warning.
35936 (fixed_convert_from_real): Likewise.
35938 2017-03-02 Bin Cheng <bin.cheng@arm.com>
35940 PR tree-optimization/66768
35941 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
35942 iv_use if base object can't be determined.
35944 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35946 PR tree-optimization/79345
35947 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
35948 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
35949 (get_pattern_stats): Initialize it.
35950 * genemit.c (gen_expand): Verify match_scratch numbers come after
35951 match_operand/match_dup numbers.
35952 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
35953 match_scratch numbers.
35954 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
35956 * config/s390/s390.md (trunctdsd2): Likewise.
35958 2017-03-02 Richard Biener <rguenther@suse.de>
35960 * wide-int.h (wide_int_storage::operator=): Implement in terms
35963 2017-03-02 Richard Biener <rguenther@suse.de>
35965 PR tree-optimization/79777
35966 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
35967 the to insert expression to sth existing.
35969 2017-03-01 Martin Sebor <msebor@redhat.com>
35971 PR middle-end/79692
35972 * gimple-ssa-sprintf.c
35973 (directive::known_width_and_precision): New function.
35974 (format_integer): Use it.
35975 (get_mpfr_format_length): Consider the full range of precision
35976 when computing %g output with the # flag. Set the likely byte
35977 count to 3 rather than 1 when precision is indeterminate.
35978 (format_floating): Correct the lower bound of precision.
35980 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35982 * doc/invoke.texi: Document default code model for 64-bit Linux.
35984 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35987 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
35988 udiv rather than div since input pattern is unsigned.
35990 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
35992 * config/i386/i386.c (print_reg): Warn for values of
35993 unsupported size in integer register.
35995 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
35998 * config/rs6000/predicates.md (current_file_function_operand): Do
35999 not allow self calls to be local if the function is replaceable.
36001 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36004 * config/rs6000/altivec.h (vec_ctz and others): Change the
36005 preprocessor macro that controls conditional compilation from
36006 _ARCH_PWR9 to __POWER9_VECTOR__.
36007 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36008 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36009 control (instead of _ARCH_PWR9 control) so that template
36010 definition uses power9-specific function.
36011 (vec_any_eq): Likewise.
36012 (vec_all_ne): Change macro definition to use a power9-specific
36013 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36014 _ARCH_PWR9 control).
36015 (vec_any_eq) Likewise.
36016 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36017 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36018 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36019 support for xvcmpnedp instruction.
36020 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36021 macro expansion so that Power9 implementation of vec_all_ne does
36022 not use the AltiVec predicate framework.
36023 (VCMPNEH_P): Likewise.
36024 (VCMPNEW_P): Likewise.
36025 (VCMPNED_P): Likewise.
36026 (VCMPNEFP_P): Likewise.
36027 (VCMPNEDP_P): Likewise.
36028 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36029 implementation of vec_any_eq to not use AltiVec predicate
36031 (VCMPAEH_P): Likewise.
36032 (VCMPAEW_P): Likewise.
36033 (VCMPAED_P): Likewise.
36034 (VCMPAEFP_P): Likewise.
36035 (VCMPAEDP_P): Likewise.
36036 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36037 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36038 not use the AltiVec predicate framework.
36039 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36040 of vec_any_eq to not use AltiVec predicate framework.
36041 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36042 support for predefined __POWER9_VECTOR__ macro to indicate that
36043 Power9 instruction selection is enabled.
36044 (altivec_overloaded_builtins): Remove extraneous
36045 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36046 function argument types RS6000_BTI_bool_V16QI and
36047 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36048 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36049 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36050 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36051 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36052 Power9 for implementations of vec_cmpne. Change the signature for
36053 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36054 (representing vec_all_ne) to remove the previously described first
36055 argument of type RS6000_BTI_INTSI, as this was an artifact of
36056 reliance on the AltiVec predicate framework, which is no longer
36057 used in the implementation of these functions. Add
36058 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36059 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36060 since, unlike the AltiVec predicate framework implementation, we
36061 do not share function descriptors between vec_alle and vec_anyeq.
36062 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36063 set of modes that receive special treatment even when
36064 TARGET_P9_VECTOR is true. The special treatment emits code that
36065 does not depend on Power9 instructions.
36066 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36067 define_expand to not rely on AltiVec predicate framework.
36068 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36070 (vector_ne_v2di_p): Change this define_expand to not rely on
36071 AltiVec predicate framework.
36072 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36074 (vector_ne_<mode>_p): Change this define_expand to not rely on
36075 AltiVec predicate framework.
36076 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36078 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36079 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36080 define_insn pattern.
36081 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36082 define_insn pattern because the xvcmpne<VSs>. instruction is not
36084 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36085 instruction is not supported.
36087 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36089 * config/nvptx/nvptx.c: Include intl.h.
36091 2017-03-01 Martin Jambor <mjambor@suse.cz>
36094 * ipa-prop.h (ipa_bits): Removed field known.
36095 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36096 to pointers. Adjusted their comments to warn about their sharing.
36097 (ipcp_transformation_summary): Change bits to a vector of pointers.
36098 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36099 (ipa_get_ipa_bits_for_value): Declare.
36100 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36101 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36102 (ipa_bits_hash_table): Likewise.
36103 (ipa_vr_ggc_hash_traits): Likewise.
36104 (ipa_vr_hash_table): Likewise.
36105 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36106 being pointers and vr_known being removed.
36107 (ipa_set_jf_unknown): Likewise.
36108 (ipa_get_ipa_bits_for_value): New function.
36109 (ipa_set_jfunc_bits): Likewise.
36110 (ipa_get_value_range): New overloaded functions.
36111 (ipa_set_jfunc_vr): Likewise.
36112 (ipa_compute_jump_functions_for_edge): Use the above functions to
36113 construct bits and vr parts of jump functions.
36114 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36115 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36117 (ipcp_grow_transformations_if_necessary): Also allocate
36118 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36120 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36121 them. Fix too long lines.
36122 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36123 vr_known being removed.
36124 (ipa_read_jump_function): Use new setter functions to construct bits
36125 and vr parts of jump functions or set them to NULL.
36126 (write_ipcp_transformation_info): Adjust for bits being pointers.
36127 (read_ipcp_transformation_info): Likewise.
36128 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36130 Include gt-ipa-prop.h.
36131 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36133 (ipcp_store_bits_results): Likewise.
36134 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36135 Do not write to existing jump functions but use a temporary instead.
36137 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36140 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36141 attempt to use its first operand as BIT_FIELD_REF base.
36143 2017-03-01 Richard Biener <rguenther@suse.de>
36145 PR middle-end/79721
36146 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36147 interpolating formula in wrapping arithmetic.
36148 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36150 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36152 PR tree-optimization/79734
36153 * tree-vect-generic.c (expand_vector_condition): Optimize
36154 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36155 Handle VEC_COND_EXPR where comparison has different inner width from
36156 type's inner width.
36158 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36160 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36161 markup, and similar issues. Remove @opindex entries for things
36162 that aren't options. Add missing -mmpy-option entries.
36164 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36166 PR tree-optimization/79737
36167 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36168 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36169 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36170 instead of byte_size. Formatting fix.
36171 (shift_bytes_in_array_right): Formatting fix.
36173 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36176 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36177 condition on optimize for the leaf function test.
36179 2017-02-28 Martin Liska <mliska@suse.cz>
36182 * read-rtl-function.c (function_reader::handle_unknown_directive):
36183 Bail out when one uses -flto.
36185 2017-02-28 Martin Liska <mliska@suse.cz>
36187 * common.opt: Replace space with tabular for options of <number>
36189 * config/i386/i386.opt: Show <number> value for
36190 -mlarge-data-threshold.
36191 * opts.c (print_filtered_help): Do not display number in hexadecimal
36194 2017-02-28 Martin Liska <mliska@suse.cz>
36196 * common.opt: Fix --help=option -Q for options which are of
36199 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36201 * config/i386/i386.c (print_reg): Error out for values
36202 of 8-bit size in invalid integer register.
36204 2017-02-28 Martin Sebor <msebor@redhat.com>
36206 PR tree-optimization/79691
36207 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36209 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36212 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36213 gcc_unreachable with output_operand_lossage.
36215 2017-02-28 Richard Biener <rguenther@suse.de>
36217 PR tree-optimization/79740
36218 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36220 (visit_nary_op): Insert the nary into the hashtable if we
36221 pattern-matched sth.
36222 * tree-ssa-pre.c (eliminate_insert): Robustify.
36224 2017-02-28 Richard Biener <rguenther@suse.de>
36226 PR middle-end/79731
36227 * fold-const.c (decode_field_reference): Reject out-of-bound
36230 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36232 * config/i386/i386.c: Include intl.h.
36233 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36234 instead of just cond ? "..." : "...".
36235 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36236 * coverage.c (read_counts_file): Likewise.
36237 * omp-offload.c: Include intl.h.
36238 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36239 of just cond ? "..." : "...".
36240 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36241 of just cond ? "..." : "...".
36243 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36246 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36248 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36249 'tune for' CPU name.
36250 * config/arm/arm-cpu-data.h: Regenerated.
36252 2017-02-28 Richard Biener <rguenther@suse.de>
36254 PR tree-optimization/79732
36255 * tree-inline.c (expand_call_inline): Do not shadow var.
36257 2017-02-28 Richard Biener <rguenther@suse.de>
36259 PR tree-optimization/79723
36260 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36261 address-space properly.
36263 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36265 * doc/optinfo.texi (Optimization groups): Fix option used for
36267 * doc/invoke.texi (-fopt-info): Document "omp".
36268 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36269 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36270 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36271 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36272 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36274 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36276 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36277 OPTGROUP_OMP with "omp".
36279 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36282 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36283 for arithmetic shift of unsigned V2DI.
36285 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36287 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36288 arc/linux.h headers.
36289 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36290 (LINK_SPEC): Likewise.
36291 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36292 (EXTRA_SPECS): Likewise.
36293 (STARTFILE_SPEC): Likewise.
36294 (ENDFILE_SPEC): Likewise.
36295 (LIB_SPEC): Likewise.
36296 (TARGET_SDATA_DEFAULT): Likewise.
36297 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36298 (MULTILIB_DEFAULTS): Likewise.
36299 (DWARF2_UNWIND_INFO): Likewise.
36300 * config/arc/big.h: New file.
36301 * config/arc/elf.h: Likewise.
36302 * config/arc/linux.h: Likewise.
36303 * config/arc/t-uClibc: Remove.
36305 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36307 PR tree-optimization/77536
36308 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36309 (tree_transform_and_unroll_loop): Use above function to compute the
36310 estimated niter of unrolled loop and use it when scaling profile.
36311 Also use count info rather than frequency if it's non-zero.
36312 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36313 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36314 (vect_transform_loop): Call above function.
36316 2017-02-27 Richard Biener <rguenther@suse.de>
36318 PR tree-optimization/45397
36319 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36320 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36321 (visit_nary_op): Add pattern matching for CSEing sign-changed
36322 or truncated operations with wider ones.
36324 2017-02-27 Richard Biener <rguenther@suse.de>
36326 PR tree-optimization/79690
36327 * tree-vect-stmts.c (vectorizable_store): Use vector type
36328 built from the DR with address-space.
36330 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36332 * doc/invoke.texi (Optimize Options): Refine the description
36333 of asan-use-after-return.
36335 2017-02-25 Alan Modra <amodra@gmail.com>
36337 PR rtl-optimization/79584
36338 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36339 base, not ad->base_term, the reg within base. Remove assertion
36340 that ad->base == ad->base_term. Replace gen_int_mode using
36341 bogus mode with const0_rtx.
36343 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36345 PR middle-end/79396
36346 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36347 FMA_EXPR like tcc_binary or tcc_unary.
36349 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36352 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36354 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36355 (output_loc_operands): Handle DW_OP_call_ref and
36356 DW_OP_GNU_variable_value.
36357 (struct variable_value_struct): New type.
36358 (struct variable_value_hasher): Likewise.
36359 (variable_value_hash): New variable.
36360 (string_types): Remove.
36361 (copy_loc_descr): New function.
36362 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36363 (prepend_loc_descr_to_each): New function.
36364 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36365 instead of add_loc_descr_to_each if the first argument is single
36366 location list and the second has multiple.
36367 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36368 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36369 when looking for variable value which doesn't have other location info.
36370 (loc_list_from_tree): Formatting fix.
36371 (gen_array_type_die): Simplify DW_AT_string_length handling.
36372 (adjust_string_types): Remove.
36373 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36374 string_types. Call resolve_variable_values.
36375 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36376 (resolve_addr_in_expr): Likewise. Add A argument.
36377 (copy_deref_exprloc): Remove deref argument. Adjust for the
36378 original expression being DW_OP_GNU_variable_value with optionally
36379 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36380 optionally after it.
36381 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36382 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36383 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36384 (variable_value_hasher::hash, variable_value_hasher::equal): New
36386 (resolve_variable_value_in_expr, resolve_variable_value,
36387 resolve_variable_values, note_variable_value_in_expr,
36388 note_variable_value): New functions.
36389 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36392 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36395 * opts.h (handle_generated_option): Add GENERATED_P argument.
36396 * opts-common.c (handle_option): Adjust function comment.
36397 (handle_generated_option): Add GENERATED_P argument, pass it to
36399 (control_warning_option): Pass false to handle_generated_option
36401 * opts.c (maybe_default_option): Pass true to handle_generated_option
36403 * optc-gen.awk: Likewise.
36405 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36407 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36408 a REG, look at the REG it is a SUBREG of.
36409 (splitter for cmpeqsi_t): Ditto.
36411 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36413 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36414 the special USEs with the pattern of the insn, not the insn itself.
36416 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36419 * doc/invoke.texi: Document -mload-store-pairs.
36421 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36422 Sandra Loosemore <sandra@codesourcery.com>
36424 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36425 argument isn't a CONST_INT.
36426 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36427 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36428 (nios2_validate_compare): Bypass alternate compare logic if *op2
36429 is not a CONST_INT.
36430 (ldstwm_operation_p): Return false if first_base is not a REG or
36431 if first_offset is not a CONST_INT.
36433 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36435 * config/cris/cris.md: Use correct operand in a define_peephole2.
36437 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36439 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36441 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36443 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36444 this_insn if it is an INSN or JUMP_INSN.
36445 (force_offsettable): Look at base, not at addr.
36446 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36447 on things that aren't necessarily CONST_INTs.
36449 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36451 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36452 -mfpmath=sse is the default also for x86-32 targets with SSE2
36453 instruction set when @option{-ffast-math} is enabled
36455 2017-02-24 Jeff Law <law@redhat.com>
36457 PR rtl-optimizatoin/79286
36458 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36461 2017-02-24 Richard Biener <rguenther@suse.de>
36463 PR tree-optimization/79389
36464 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36467 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36469 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36470 function comment to reflect reality.
36471 (loop_exits_before_overflow): Fix typo in function description.
36473 2017-02-24 Richard Biener <rguenther@suse.de>
36475 PR tree-optimization/79389
36476 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36477 properly that a threading opportunity exists. Detect conditional
36478 copy/constant propagation opportunities.
36480 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36482 * config/visium/visium.md (type): Add trap.
36483 (b): New mode attribute.
36484 (*btst): Rename into...
36485 (*btst<mode>): ...this and adjust.
36486 (*cbranchsi4_btst_insn): Rename into...
36487 (*cbranch<mode>4_btst_insn): ...this and adjust.
36488 (trap): New define_insn.
36490 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36492 PR tree-optimization/79389
36493 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36494 (noce_reversed_cond_code): New function.
36495 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36496 reversed_comparison_code. Formatting fix.
36497 (noce_try_store_flag): Test rev_cond != NULL in addition to
36498 reversed_comparison_code.
36499 (noce_try_store_flag_constants): Likewise.
36500 (noce_try_store_flag_mask): Likewise.
36501 (noce_try_addcc): Use rev_cond if non-NULL instead of
36502 reversed_comparison_code.
36503 (noce_try_cmove_arith): Likewise. Formatting fixes.
36504 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36505 (noce_find_if_block): Initialize rev_cond.
36506 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36507 instead of false as last argument never attempt to reverse it
36510 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36512 PR tree-optimization/79663
36513 * tree-predcom.c (combine_chains): Process refs in reverse order
36514 only for ZERO length chains, and add explaining comment.
36516 2017-02-23 Jeff Law <law@redhat.com>
36518 PR tree-optimization/79578
36519 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36520 in call to operand_equal_p.
36522 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36525 * config/i386/cpuid.h: Fix another undefined behavior.
36527 2017-02-23 Richard Biener <rguenther@suse.de>
36529 PR tree-optimization/79683
36530 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36531 vector types for data-refs.
36533 2017-02-23 Martin Liska <mliska@suse.cz>
36535 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36537 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36539 PR middle-end/79665
36540 * internal-fn.c (get_range_pos_neg): Moved to ...
36541 * tree.c (get_range_pos_neg): ... here. No longer static.
36542 * tree.h (get_range_pos_neg): New prototype.
36543 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36544 are known to be in between 0 and signed maximum inclusive, try to
36545 expand both unsigned and signed divmod and use the cheaper one from
36548 2017-02-22 Jeff Law <law@redhat.com>
36550 PR tree-optimization/79578
36551 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36552 to compare base operands.
36554 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36557 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36558 gpc_reg_operand instead of fpr_reg_operand.
36560 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36562 * config/mips/mips.c (mips_return_in_memory): Force FP
36563 vector types to be returned in memory for o32 ABI.
36565 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36567 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36568 instead of DW_TAG_member for static data member declarations and don't
36569 set no_linkage_name for static inline data members.
36570 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36573 2017-02-22 Martin Liska <mliska@suse.cz>
36575 * doc/invoke.texi: Replace inequality signs with square brackets
36578 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36581 * lra-constraints.c (simplify_operand_subreg): Handle
36582 WORD_REGISTER_OPERATIONS targets.
36584 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36587 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36588 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36589 elimination by swapping fld*.
36591 2017-02-22 Richard Biener <rguenther@suse.de>
36593 PR tree-optimization/79673
36594 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36595 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36596 irrelevant address-space qualifiers and avoiding a
36597 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36599 2017-02-22 Richard Biener <rguenther@suse.de>
36601 PR tree-optimization/79666
36602 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36603 to not symbolically negate if that may introduce undefined
36606 2017-02-22 Martin Liska <mliska@suse.cz>
36609 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36610 * data-streamer-out.c (streamer_write_gcov_count_stream):
36612 * value-prof.c (stream_out_histogram_value): Make assert more
36613 precise based on type of counter.
36615 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36618 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36619 Use nonimmediate_operand instead of memory_operand for operand 1.
36620 (float-extend standard_x87sse_constant_load splitter): Ditto.
36622 2017-02-21 Jeff Law <law@redhat.com>
36624 PR tree-optimization/79621
36625 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36626 blocks with edges to themselves.
36628 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36631 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36632 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36633 Use gimple_call_builtin_p.
36636 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36637 on temporarily removed DEBUG_INSNs.
36639 PR tree-optimization/79649
36640 * tree-loop-distribution.c (classify_partition): Give up on
36641 non-generic address space loads/stores.
36643 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36645 * doc/loop.texi (Loop manipulation): Remove nonexistent
36646 tree_ssa_loop_version from the documentation.
36647 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36649 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36652 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36653 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36654 * config/rs6000/rs6000.c: Include except.h.
36655 (rs6000_expand_split_stack_prologue): Call
36656 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36658 2017-02-21 Martin Jambor <mjambor@suse.cz>
36661 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36662 have been analyzed.
36664 2017-02-21 Martin Jambor <mjambor@suse.cz>
36666 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36667 for backward compatibility only.
36668 * doc/invoke.texi (Option Summary): Remove all references to
36669 -fipa-cp-alignment.
36671 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36675 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36677 * lra-constraints.c (curr_insn_transform): Handle
36678 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36680 2017-02-21 Martin Liska <mliska@suse.cz>
36682 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36684 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36686 PR translation/79638
36687 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36689 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36692 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36693 (arm_function_ok_for_sibcall): Return false for an indirect call by
36694 descriptor if all the argument registers are used.
36695 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36696 alignment of the function.
36698 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36700 PR tree-optimization/61441
36701 * simplify-rtx.c (simplify_const_unary_operation): For
36702 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36703 the sNaN unmodified.
36705 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36707 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36708 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36709 instead of SYSTEM_HEADER_DIR.
36711 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36712 Martin Liška <mliska@suse.cz>
36714 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36715 Fix typos and grammar, use active voice, and clarify.
36717 2017-02-20 Marek Polacek <polacek@redhat.com>
36719 PR middle-end/79537
36720 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36723 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36725 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36728 * config/i386/i386.c (ix86_expand_builtin): Handle
36729 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36730 ix86_builtins_isa[fcode].isa as a requirement of those
36731 flags and any other flag in the bitmask.
36732 (ix86_init_mmx_sse_builtins): Use 0 instead of
36733 ~OPTION_MASK_ISA_64BIT as mask.
36734 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36735 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36736 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36737 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36739 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36742 * lra-constraints.c (split_reg): Check requested split mode
36743 is supported by the register.
36745 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36747 * lra-constraints.c (simplify_operand_subreg): Remove early
36750 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36753 * lra-constraints.c (curr_insn_transform): Tighten condition
36754 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36756 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36759 * lra-constraints.c (curr_insn_transform): Handle
36760 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36762 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36765 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36767 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36769 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36772 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36775 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36777 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36780 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36782 * final.c (last_columnnum, override_columnnum): New variables.
36783 (final_start_function): Set last_columnnum, pass it to begin_prologue
36784 hook and pass 0 to dwarf2out_begin_prologue.
36785 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36786 to source_line debug hook.
36787 (notice_source_line): Compute last_columnnum and for debug_column_info
36788 return true on column changes.
36789 * debug.h (struct gcc_debug_hooks): Add column argument to
36790 source_line and begin_prologue hooks.
36791 (debug_nothing_int_charstar_int_bool): Remove prototype.
36792 (debug_nothing_int_int_charstar,
36793 debug_nothing_int_int_charstar_int_bool): New prototypes.
36794 (dwarf2out_begin_prologue): Add column argument.
36795 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36796 begin_prologue hooks.
36797 (debug_nothing_int_charstar_int_bool): Remove.
36798 (debug_nothing_int_int_charstar,
36799 debug_nothing_int_int_charstar_int_bool): New functions.
36800 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
36801 through to dwarf2out_source_line.
36802 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
36803 (dwarf2out_source_line): Add column argument, emit it if requested.
36804 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
36806 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36807 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36808 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
36809 through to dwarf2out_begin_prologue.
36810 (vmsdbgout_source_line): Add column argument, pass it through to
36811 dwarf2out_source_line.
36812 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
36813 dbxout_source_line caller.
36814 (dbxout_source_line): Add column argument.
36816 * common.opt (gno-column-info, gcolumn-info): New options.
36817 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
36818 (check_die): Also test for multiple DW_AT_decl_column attributes.
36819 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
36820 DW_AT_decl_column if requested.
36821 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
36823 (gen_variable_die): Likewise.
36824 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
36825 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
36828 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
36829 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
36830 (ix86_handle_option): Handle OPT_m3dnowa.
36831 * doc/invoke.texi (-m3dnowa): Document.
36832 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
36833 -m3dnowa instead of -m3dnow -march=athlon.
36836 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
36837 instead of gcc_assert for K, r and R code checks. Formatting fixes.
36839 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36842 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
36843 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
36844 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
36845 generator for vsx_xxpermdi_<mode>_be.
36846 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
36847 force big-endian semantics.
36848 (vsx_xxpermdi_<mode>_be): New define_expand with same
36849 implementation as previous version of vsx_xxpermdi_<mode>.
36851 2017-02-17 Jakub Jelinek <jakub@redhat.com>
36853 PR tree-optimization/79327
36854 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
36855 variable, its initialization and use.
36857 2017-02-17 Julia Koval <julia.koval@intel.com>
36859 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
36860 (OPTION_MASK_ISA_PKU_UNSET): New.
36861 (ix86_handle_option): Handle -mrdpid.
36862 * config/i386/cpuid.h (bit_RDPID): New.
36863 * config/i386/driver-i386.c (host_detect_local_cpu):
36864 Detect RDPID feature.
36865 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
36866 * config/i386/i386-c.c (ix86_target_macros_internal):
36868 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
36869 (ix86_valid_target_attribute_inner_p): Add "rdpid".
36870 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
36871 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
36872 * config/i386/i386.md (define_insn "rdpid"): New.
36873 * config/i386/i386.opt Add -mrdpid.
36874 * config/i386/immintrin.h (_rdpid_u32): New.
36876 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
36878 PR rtl-optimization/79541
36879 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
36880 instead of transforming it into USE.
36882 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
36884 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
36885 If HONOR_SNANS (SFmode) force the input to a register.
36886 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
36887 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
36888 an frsp or similar insn.
36890 2017-02-17 Martin Liska <mliska@suse.cz>
36892 PR rtl-optimization/79577
36893 * params.def (selsched-max-sched-times): Increase minimum to 1.
36895 2017-02-17 Martin Liska <mliska@suse.cz>
36897 PR rtl-optimization/79574
36898 * gcse.c (want_to_gcse_p): Prevent integer overflow.
36900 2017-02-17 Martin Liska <mliska@suse.cz>
36902 PR tree-optimization/79529
36903 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
36904 ssa_defined_default_def_p to handle cases which are implicitly
36906 * tree-ssa.c (ssa_defined_default_def_p): New function.
36907 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
36908 which are implicitly defined.
36909 * tree-ssa.h (ssa_defined_default_def_p): Declare.
36911 2017-02-17 Richard Biener <rguenther@suse.de>
36913 PR middle-end/79576
36914 * params.def (max-ssa-name-query-depth): Limit to 10.
36916 2017-02-17 Richard Biener <rguenther@suse.de>
36918 PR tree-optimization/79552
36919 * tree-ssa-structalias.c (visit_loadstore): Properly verify
36922 2017-02-17 Richard Biener <rguenther@suse.de>
36925 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
36927 2017-02-17 Marek Polacek <polacek@redhat.com>
36929 PR middle-end/79536
36930 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
36931 (fold_negate_expr): New wrapper.
36933 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
36935 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
36936 Correct terminology and de-emphasize pre-standard behavior.
36938 2017-02-16 Alan Modra <amodra@gmail.com>
36940 PR rtl-optimization/79286
36941 * ira.c (def_dominates_uses): New function.
36942 (update_equiv_regs): Don't create an equivalence for insns that
36943 may trap where the register def does not dominate the use.
36945 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
36947 PR rtl-optimization/78127
36948 * lra.c (lra): Call lra_eliminate before finish the loop after
36951 2017-02-16 Richard Biener <rguenther@suse.de>
36953 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
36955 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
36956 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
36957 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
36958 (isl_val_int_from_wi): New function.
36959 (extract_affine_gmp): Rename to ...
36960 (extract_affine_wi): ... this, take a widest_int.
36961 (extract_affine_int): Just wrap extract_affine_wi.
36962 (add_param_constraints): Use isl_val_int_from_wi.
36963 (add_loop_constraints): Likewise, and extract_affine_wi.
36965 2017-02-15 Jeff Law <law@redhat.com>
36967 PR middle-end/79521
36968 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
36969 ira_init_register_move_cost_if_necessary.
36971 2017-02-15 Martin Sebor <msebor@redhat.com>
36973 PR middle-end/32003
36974 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
36975 removed in a prior commit.
36977 2017-02-15 Bin Cheng <bin.cheng@arm.com>
36979 PR tree-optimization/79347
36980 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
36981 counters during peeling.
36983 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
36985 * Makefile.in (site.exp): Remove "set ISLVER".
36987 2017-02-15 Jakub Jelinek <jakub@redhat.com>
36990 * real.c (real_from_integer): Call real_convert even for decimal.
36992 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
36995 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
36997 2017-02-14 Andrew Pinski <apinski@cavium.com>
36999 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37000 cores and change the partno/implementer to be correct.
37001 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37002 the 'B" as the implementer.
37003 * config/aarch64/aarch64-tune.md: Regenerate.
37005 2017-02-14 Carl Love <cel@us.ibm.com>
37007 * config/rs6000/rs6000.c: Add case statement entry to make the
37008 xvcvuxdsp built-in argument unsigned.
37009 * config/rs6000/vsx.md: Fix the source and return operand types so they
37010 match the instruction definitions from the ISA document. Fix typo
37011 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37014 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37017 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37018 member early_clobber_alts.
37019 * lra-lives.c (reg_early_clobber_p): New.
37020 (process_bb_lives): Use it.
37021 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37022 (debug_operand_data): Initialize early_clobber_alts.
37023 (setup_operand_alternative): Set up early_clobber_alts.
37024 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37025 alternatives to new_insn_reg.
37026 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37028 (lra_update_insn_regno_info): Pass the new arg.
37030 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37032 PR middle-end/79505
37033 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37034 (new_oacc_loop_raw): Don't clear already cleared fields.
37037 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37038 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37039 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37041 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37044 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37046 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37049 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37050 the extra instruction to the right place to store 128-bit constant
37053 2017-02-14 Martin Sebor <msebor@redhat.com>
37055 PR middle-end/79448
37056 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37057 warning for strings of unknown length.
37059 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37061 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37063 2017-02-14 Jeff Law <law@redhat.com>
37066 * ira-costs.c (scan_one_insn): Initialize register move costs
37067 for pseudos seen in USE/CLOBBER insns.
37069 PR tree-optimization/79095
37070 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37071 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37072 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37073 if the operands are known to be not equal, then the resulting range
37075 (intersect_ranges): If the new range is ~[0,0] and the old range is
37076 wide, then prefer ~[0,0].
37077 * tree-vrp.c (overflow_comparison_p_1): New function.
37078 (overflow_comparison_p): New function.
37079 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37080 if NAME is used in an overflow test.
37081 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37082 overflow check that can be expressed as an equality test, then adjust
37083 ops to be that equality test.
37085 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37087 * config/s390/s390-builtin-types.def: Remove flags argument.
37088 * config/s390/s390.c (s390_init_builtins): Likewise.
37090 2017-02-14 Martin Liska <mliska@suse.cz>
37092 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37093 vector. Fix trailing white spaces.
37095 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37097 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37100 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37102 PR rtl-optimization/68664
37103 * config/arm/arm.c (arm_sched_can_speculate_insn):
37104 New function. Declare prototype.
37105 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37107 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37109 PR rtl-optimization/68664
37110 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37112 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37114 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37116 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37117 max skip bytes for function, loop and jump.
37119 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37121 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37122 ABS_EXPR for gimple dump.
37124 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37127 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37129 PR tree-optimization/79408
37130 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37131 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37132 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37133 also if rhs1 is INTEGER_CST.
37135 2017-02-14 Richard Biener <rguenther@suse.de>
37137 PR middle-end/79432
37138 * tree-into-ssa.c (insert_phi_nodes): When the function can
37139 have abnormal edges rewrite SSA names with broken use-def
37140 dominance out of SSA and register them for PHI insertion.
37142 2017-02-13 Martin Sebor <msebor@redhat.com>
37144 PR middle-end/79496
37145 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37146 clearing info.nowrite flag when snprintf size argument is a range.
37148 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37150 * cprop.c (cprop_jump): Add missing space in string literal.
37151 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37152 (get_constraint_for_component_ref): Likewise.
37153 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37154 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37155 * lra-constraints.c (process_alt_operands): Likewise.
37156 * ipa-inline.c (inline_small_functions): Likewise.
37157 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37158 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37159 * trans-mem.c (diagnose_tm_1_op): Likewise.
37160 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37161 (grid_parallel_clauses_gridifiable): Likewise.
37163 * config/nvptx/mkoffload.c (process): Add space in between
37166 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37167 "MOD4_SSE_REGS" and "ALL_REGS".
37169 * spellcheck.c (test_data): Add , in between "foo" and "food".
37171 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37174 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37175 boundary crossing check and subsequent code generation agree.
37177 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37179 * config/aarch64/aarch64.c (has_memory_op): Delete.
37180 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37183 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37185 PR rtl-optimization/79388
37186 PR rtl-optimization/79450
37187 * combine.c (distribute_notes): When removing TEM_INSN for which
37188 corresponding dest has last value recorded, invalidate that last
37191 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37193 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37194 of explicit '@'. Add missing assembly comment marker on branch costs
37197 2017-02-13 Nathan Sidwell <nathan@acm.org>
37199 * gengtype-lex.l (<in_struct>): Add '/'.
37201 2017-02-13 Martin Liska <mliska@suse.cz>
37204 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37206 2017-02-13 Richard Biener <rguenther@suse.de>
37208 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37210 * configure: Re-generate.
37211 * config.in: Likewise.
37212 * graphite-dependences.c: Simplify as if
37213 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37214 * graphite-isl-ast-to-gimple.c: Likewise.
37215 * graphite-optimize-isl.c: Likewise.
37216 * graphite-poly.c: Likewise.
37217 * graphite-sese-to-poly.c: Likewise.
37218 * graphite.h: Likewise.
37219 * toplev.c: Include isl/version.h and use isl_version () for
37220 printing the ISL version.
37221 * doc/install.texi: Update ISL requirement.
37223 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37225 * doc/standards.texi (Standards): Update reference to
37228 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37230 * doc/extend.texi (Named Address Spaces): sourceware.org now
37232 * doc/install.texi (Binaries): Ditto.
37235 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37237 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37238 standard terminology "stringize"/"stringizing" throughout.
37239 * doc/cppinternals.texi: Likewise.
37241 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37243 * doc/extend.texi: Fix some spelling mistakes and typos.
37244 * doc/invoke.texi: Likewise.
37246 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37249 * params.def (inline-min-speedup) Change from 10 to 8.
37251 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37253 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37256 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37259 * ipa-inline-analysis.c (get_minimal_bb): New function.
37260 (record_modified): Use it.
37261 (remap_edge_change_prob): Handle also ancestor functions.
37263 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37265 * doc/contrib.texi (Contributors): Remove broken link into
37266 the Mauve CVS repository.
37268 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37270 PR middle-end/79454
37271 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37272 result computation whenever lhs doesn't have vector mode, not
37273 just when it has BLKmode.
37275 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37277 * doc/makefile.texi (profiledbootstrap): Refer to the
37278 installation instructions only in textual form.
37280 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37283 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37285 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37287 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37288 (Specific): Update mingw-w64 reference.
37290 (Specific): Remove broken link to Renesas RX processor.
37292 2017-02-10 Richard Biener <rguenther@suse.de>
37294 * toplev.c (process_options): Do not mention obsolete graphite
37295 options when printing sorry message about missing graphite support.
37296 Mention -floop-nest-optimize.
37298 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37300 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37301 (vtst_p16): Likewise.
37302 (vtstq_p8): Likewise.
37303 (vtstq_p16): Likewise.
37305 (vtstq_p64): Likewise.
37306 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37307 (vset_lane_p64): New.
37308 (vsetq_lane_p64): New.
37310 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37312 PR tree-optimization/79411
37313 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37314 stmt operands are SSA_NAMEs used in abnormal phis.
37315 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37318 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37321 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37324 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37326 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37328 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37330 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37334 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37335 not arbitrary TREE_CONSTANT.
37338 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37339 "omp declare target link" attribute unless is_global_var.
37340 * omp-offload.c (find_link_var_op): Likewise.
37342 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37343 Chung-Lin Tang <cltang@codesourcery.com>
37345 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37347 (gimplify_adjust_omp_clauses): Don't delete TILE.
37348 (gimplify_omp_for): Deal with TILE.
37349 * internal-fn.c (expand_GOACC_TILE): New function.
37350 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37352 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37353 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37355 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37356 avoid DIV for outermost collapse var.
37357 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37358 Remove out of date comments, fix whitespace.
37359 * omp-general.c (omp_extract_for_data): Deal with tiling.
37360 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37361 adjust OLF_DIM_BASE value.
37362 (struct omp_for_data): Add tiling field.
37363 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37364 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37365 for auto loops. Remove default auto determining, moved to
37366 oacc_loop_fixed_partitions.
37367 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37368 stmts, add e_mask field.
37369 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37370 (oacc_thread_numbers): Use oacc_dim_call.
37371 (oacc_xform_tile): New.
37372 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37373 (finish_oacc_loop): Adjust for ifns vector.
37374 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37375 add case for GOACC_TILE fns.
37376 (oacc_loop_xform_loop): Delete.
37377 (oacc_loop_process): Iterate over call list directly, and add
37378 handling for GOACC_TILE fns.
37379 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37381 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37382 vector partitioning to outer loops. Assign 2 partitions to loops
37383 when available. Add TILE handling.
37384 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37385 (execite_oacc_device_lower): Process GOACC_TILE fns,
37386 ignore unknown specs.
37387 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37388 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37389 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37391 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37393 * configure.ac (ACX_BUGURL): Update.
37394 * configure: Regenerate.
37396 2017-02-09 Richard Biener <rguenther@suse.de>
37398 PR tree-optimization/69823
37399 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37400 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37402 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37404 * config/arc/arc-c.def: Add __NPS400__ definition.
37405 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37406 (TARGET_NPS400): Define.
37408 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37410 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37412 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37413 pointer, arch_info.
37414 (arc_cpu_types): Fill the arch_info field with a pointer into the
37415 arc_arch_types table.
37416 (arc_selected_cpu): Declare.
37417 * config/arc/arc.c (arc_selected_cpu): Make global.
37418 (arc_selected_arch): Delete.
37419 (arc_base_cpu): Delete.
37420 (arc_override_options): Remove references to deleted variables,
37421 update access to arch information.
37422 (ARC_OPT): Update access to arch information.
37423 (ARC_OPTX): Likewise.
37424 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37425 (TARGET_ARC600): Update access to arch information.
37426 (TARGET_ARC601): Likewise.
37427 (TARGET_ARC700): Likewise.
37428 (TARGET_EM): Likewise.
37429 (TARGET_HS): Likewise.
37430 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37433 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37436 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37437 condition/operands for integer GE/LE/GEU/LEU operations.
37439 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37441 PR translation/79397
37442 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37445 2017-02-08 Martin Jambor <mjambor@suse.cz>
37448 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37449 whether allocation happened.
37450 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37451 nothing was allocated.
37453 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37455 PR tree-optimization/79408
37456 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37457 constant, but SSA_NAME with a known integer range, use the minimum
37458 of that range instead of op1 to determine if modulo can be replaced
37459 with its first operand.
37461 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37463 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37465 2017-02-08 Richard Biener <rguenther@suse.de>
37467 PR tree-optimization/71824
37468 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37469 Check all loops contained in the merged region.
37471 2017-02-07 Andrew Pinski <apinski@cavium.com>
37473 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37475 2017-02-07 Andrew Pinski <apinski@cavium.com>
37477 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37478 (thunderxt88): Likewise.
37479 (thunderxt81): Disable LSE and change v8.1 to v8.
37480 (thunderxt83): Likewise.
37482 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37483 Richard Biener <rguenther@suse.de>
37485 PR middle-end/79399
37486 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37487 type from int to size_t.
37488 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37490 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37492 PR rtl-optimization/79386
37493 * cprop.c (bypass_conditional_jumps): Initialize
37494 bypass_last_basic_block already before splitting bbs after
37495 unconditional traps...
37496 (bypass_conditional_jumps): ... rather than here.
37499 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37500 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37501 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37502 fix -masm=intel patterns.
37504 2017-02-07 Richard Biener <rguenther@suse.de>
37506 PR tree-optimization/79256
37507 PR middle-end/79278
37508 * builtins.c (get_object_alignment_2): Use min_align_of_type
37509 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37510 and ADJUST_FIELD_ALIGN.
37512 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37514 * doc/tm.texi: Regenerate.
37515 * stor-layout.c (layout_decl): Adjust.
37516 (update_alignment_for_field): Likewise.
37517 (place_field): Likewise.
37518 (min_align_of_type): Likewise.
37519 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37520 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37521 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37522 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37523 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37524 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37525 * config/i386/i386.c (x86_field_alignment): Likewise.
37526 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37527 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37528 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37529 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37530 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37531 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37535 2017-01-30 Richard Biener <rguenther@suse.de>
37537 PR tree-optimization/79256
37538 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37539 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37542 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37544 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37545 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37546 builtins to SImode and emit a zero-extend, if necessary.
37548 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37550 * docs/invoke.texi (RISC-V Options): Alphabetize.
37552 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37554 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37557 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37559 * config/riscv/riscv.c: New file.
37560 * common/config/riscv/riscv-common.c: Likewise.
37561 * config.gcc: Likewise.
37562 * config/riscv/constraints.md: Likewise.
37563 * config/riscv/elf.h: Likewise.
37564 * config/riscv/generic.md: Likewise.
37565 * config/riscv/linux.h: Likewise.
37566 * config/riscv/multilib-generator: Likewise.
37567 * config/riscv/peephole.md: Likewise.
37568 * config/riscv/pic.md: Likewise.
37569 * config/riscv/predicates.md: Likewise.
37570 * config/riscv/riscv-builtins.c: Likewise.
37571 * config/riscv/riscv-c.c: Likewise.
37572 * config/riscv/riscv-ftypes.def: Likewise.
37573 * config/riscv/riscv-modes.def: Likewise.
37574 * config/riscv/riscv-opts.h: Likewise.
37575 * config/riscv/riscv-protos.h: Likewise.
37576 * config/riscv/riscv.h: Likewise.
37577 * config/riscv/riscv.md: Likewise.
37578 * config/riscv/riscv.opt: Likewise.
37579 * config/riscv/sync.md: Likewise.
37580 * config/riscv/t-elf-multilib: Likewise.
37581 * config/riscv/t-linux: Likewise.
37582 * config/riscv/t-linux-multilib: Likewise.
37583 * config/riscv/t-riscv: Likewise.
37584 * configure.ac: Likewise.
37585 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37586 Waterman as RISC-V maintainers.
37587 * doc/install.texi: Add RISC-V entries.
37588 * doc/invoke.texi: Add RISC-V options section.
37589 * doc/md.texi: Add RISC-V constraints section.
37590 * configure: Regenerated.
37592 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37595 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37596 false values to be constant vectors with all 0 or all 1 bits set.
37597 (vcondu<mode><mode>): Likewise.
37598 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37600 (fpmask_comparison_operator): Update comment.
37601 (vecint_comparison_operator): New predicate.
37602 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37603 vector conditionals when the true and false values are constant
37604 vectors with all 0 bits or all 1 bits set.
37606 2017-02-06 Martin Sebor <msebor@redhat.com>
37608 PR tree-optimization/79376
37609 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37611 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37613 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37614 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37615 to simplify split condition.
37617 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37619 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37620 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37623 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37625 PR rtl-optimization/68664
37626 * target.def (can_speculate_insn): New hook.
37627 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37628 * doc/tm.texi: Regenerate.
37629 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37630 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37631 (rs6000_sched_can_speculate_insn): New function.
37633 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37635 PR tree-optimization/79284
37636 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37637 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37638 vectorizable_mask_load_store, vectorizable_operation,
37639 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37640 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37641 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37642 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37643 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37644 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37645 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37646 is_gimple_assign (stmt). Replace another such test with
37647 is_gimple_assign (stmt).
37649 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37652 * config/avr/avr.c (rtl-iter.h): Include it.
37653 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37654 (avr_legitimate_combined_insn): ...and implementation.
37656 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37658 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37659 * config/s390/s390.c (s390_const_operand_ok)
37660 (s390_canonicalize_comparison, s390_extract_part)
37661 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37662 (s390_contiguous_bitmask_p, s390_rtx_costs)
37663 (legitimize_pic_address): Likewise.
37664 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37665 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37666 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37667 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37668 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37670 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37672 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37673 REGNO($0) == REGNO($1).
37675 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37677 * config/s390/linux.h(SIZE_TYPE): Add comment.
37679 2017-02-06 Julian Brown <julian@codesourcery.com>
37680 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37681 Virendra Pathak <virendra.pathak@broadcom.com>
37683 * config/aarch64/aarch64-cores.def: Change the scheduler
37685 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37686 * config/aarch64/thunderx2t99.md: New file.
37688 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37690 * doc/standards.texi (Go Language): Update link to language
37693 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37695 * tree-eh.c (lower_resx): Sanitize profile.
37696 (cleanup_empty_eh_move_lp): Likewise.
37698 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37701 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37703 * cfgloopmanip.h (loop_version): Update prototype.
37704 * modulo-sched.c (sms_schedule): Update call of loop_version.
37705 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37706 * tree-parloops.c (gen_parallel_loop): Likewise.
37707 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37708 * tree-ssa-loop-split.c (split_loop): Likewise.
37709 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37710 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37712 2017-02-05 Martin Liska <mliska@suse.cz>
37715 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37717 (print_operand_address): Initialize a struct to zero.
37719 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37721 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37722 garbage collector only in textual form.
37724 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37726 * doc/extend.texi (x86 specific memory model extensions for
37727 transactional memory): Simplify a phrase.
37729 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37732 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37733 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37734 (atomic_storedi_1): Likewise.
37736 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37738 PR tree-optimization/79338
37739 * tree-parloops.c (gather_scalar_reductions): Don't call
37740 vect_analyze_loop_form for loop->inner before destroying loop's
37743 2017-02-03 Martin Sebor <msebor@redhat.com>
37745 PR tree-optimization/79327
37746 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37747 when precision has resulted in leading zeros.
37748 (format_integer): Adjust the likely counter to assume an unknown
37749 argument that may be zero is non-zero.
37751 2017-02-03 Jason Merrill <jason@redhat.com>
37754 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37755 avoid copying non-taken branch.
37757 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37759 PR tree-optimization/79340
37760 * tree-vect-loop.c (vectorizable_reduction): Release
37761 vec_defs elements after safe_splicing them into other vectors.
37764 PR tree-optimization/79327
37765 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37766 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37768 (format_integer): Use wide_int_to_tree instead of build_int_cst
37769 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37770 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37771 of shortest and longest sequence.
37773 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37775 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37776 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37778 2017-02-03 Walter Lee <walt@tilera.com>
37781 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37782 after initial stackframe link reg save.
37783 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37785 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37788 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37789 wu for stxssp alternative.
37791 2017-02-03 Martin Sebor <msebor@redhat.com>
37793 PR tree-optimization/79352
37794 * gimple-fold.c (get_range_strlen): Add argument.
37795 (get_range_strlen): Change return type to bool.
37796 (get_maxval_strlen): Pass in a dummy argument.
37797 * gimple-fold.h (get_range_strlen): Change return type to bool.
37798 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
37799 * tree.h (array_at_struct_end_p): Add argument.
37800 * tree.c (array_at_struct_end_p): Handle it.
37802 2017-02-03 Martin Liska <mliska@suse.cz>
37805 * multiple_target.c (create_dispatcher_calls): Redirect edge
37806 from a caller of a dispatcher.
37807 (expand_target_clones): Make the clones local.
37808 (ipa_target_clone): Do both target clones and resolvers.
37809 (ipa_dispatcher_calls): Remove the pass.
37810 (pass_dispatcher_calls::gate): Likewise.
37811 (make_pass_dispatcher_calls): Likewise.
37812 * passes.def (pass_target_clone): Put as very first IPA early
37815 2017-02-03 Martin Liska <mliska@suse.cz>
37817 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
37818 in case of a function with ifunc attribute.
37820 2017-02-03 Martin Liska <mliska@suse.cz>
37822 * cgraph.c (cgraph_node::dump): Dump function version info.
37823 * symtab.c (symtab_node::dump_base): Add missing new line.
37825 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37827 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
37828 (ifcombine_ifandif): Use it.
37830 2017-02-03 Martin Liska <mliska@suse.cz>
37832 * doc/invoke.texi: Document default value for
37833 use-after-scope-direct-emission-threshold.
37835 2017-02-03 Martin Liska <mliska@suse.cz>
37837 PR tree-optimization/79339
37838 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
37839 (format_floating): Likewise.
37841 2017-02-03 Martin Liska <mliska@suse.cz>
37844 * ipa-prop.c (ipa_node_params_t::insert): Remove current
37846 (ipa_node_params_t::remove): Likewise.
37847 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
37848 initialization from removed ipa_node_params_t::insert.
37849 (ipa_node_params::~ipa_node_params): Move from removed
37850 ipa_node_params_t::release.
37851 * symbol-summary.h (symbol_summary::m_released): New member.
37852 Do not release a summary twice. Do not allow to call finalizer
37853 for types of a summary that live in GGC memory.
37855 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37857 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
37860 2017-02-02 Martin Sebor <msebor@redhat.com>
37862 PR middle-end/79275
37863 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
37864 (format_string): Tighten up the range of output for non-constant
37865 strings and correct the expected range for wide non-constant strings.
37867 2017-02-02 Martin Sebor <msebor@redhat.com>
37869 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
37871 PR middle-end/32003
37872 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
37874 (-fdump-tree-@var): Add to index and document how to come up
37875 with pass-specific option and dump file names.
37876 (-fdump-passes): Clarify where to look for output.
37878 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37880 PR middle-end/77445
37881 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
37882 statistics of the analyzed path; allow threading for speed when
37883 any of BBs along the path are optimized for speed.
37885 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
37887 PR middle-end/78468
37888 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
37889 settings of the virtual registers.
37892 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
37894 * explow.c (get_dynamic_stack_size): Take known alignment of stack
37895 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
37898 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37900 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
37901 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
37903 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37905 * config/s390/s390.md: Add missing comments with the expanded
37907 * config/s390/vector.md: Likewise.
37908 * config/s390/vx-builtins.md: Likewise.
37910 2017-02-02 Jakub Jelinek <jakub@redhat.com>
37913 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
37914 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
37915 conditions on a single line.
37917 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37919 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37920 __S390_VX__ to __VX__.
37922 2017-02-01 Andrew Pinski <apinski@cavium.com>
37924 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
37925 stmt_info to record_stmt_cost.
37926 (vect_get_known_peeling_cost): Pass stmt_info if known to
37928 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
37929 cpu_vector_cost field into
37930 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
37931 field into vec_int_stmt_cost and vec_fp_stmt_cost.
37932 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
37933 splitting of scalar_stmt_cost and vec_stmt_cost.
37934 (thunderx_vector_cost): Likewise.
37935 (cortexa57_vector_cost): LIkewise.
37936 (exynosm1_vector_cost): Likewise.
37937 (xgene1_vector_cost): Likewise.
37938 (thunderx2t99_vector_cost): Improve after the splitting of the two
37940 (aarch64_builtin_vectorization_cost): Update for the splitting of
37941 scalar_stmt_cost and vec_stmt_cost.
37943 2017-02-01 Torvald Riegel <triegel@redhat.com>
37944 Richard Henderson <rth@redhat.com>
37946 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
37947 conditional on existance of a fast atomic load.
37948 * optabs-query.c (can_atomic_load_p): New function.
37949 * optabs-query.h (can_atomic_load_p): Declare it.
37950 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
37951 no fast atomic load is available for the particular size of access.
37952 (expand_atomic_compare_and_swap): Likewise.
37953 (expand_atomic_load): Likewise.
37954 (expand_atomic_store): Likewise.
37955 (expand_atomic_fetch_op): Likewise.
37956 * testsuite/lib/target-supports.exp
37957 (check_effective_target_sync_int_128): Remove x86 because it provides
37958 no fast atomic load.
37959 (check_effective_target_sync_int_128_runtime): Likewise.
37961 2017-02-01 Richard Biener <rguenther@suse.de>
37963 * graphite.c: Include tree-vectorizer.h for find_loop_location.
37964 (graphite_transform_loops): Provide opt-info for optimized nests.
37965 * tree-parloop.c (parallelize_loops): Provide opt-info for
37966 parallelized loops.
37968 2017-02-01 Richard Biener <rguenther@suse.de>
37970 PR middle-end/79315
37971 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
37972 was not set before.
37974 2017-02-01 Richard Biener <rguenther@suse.de>
37976 PR tree-optimization/71824
37977 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37978 Verify the loops are valid in the merged SESE region.
37979 (scop_detection::can_represent_loop_1): Check analyzing the
37980 evolution of the number of iterations in the region succeeds.
37982 2017-01-31 Ian Lance Taylor <iant@golang.org>
37984 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
37985 REG_ARGS_SIZE note to 32-bit push insns and call insn.
37987 2017-01-31 David Malcolm <dmalcolm@redhat.com>
37989 PR preprocessor/79210
37990 * input.c (get_substring_ranges_for_loc): Replace line_width
37991 assertion with error-handling.
37993 2017-01-31 Richard Biener <rguenther@suse.de>
37995 PR tree-optimization/77318
37996 * graphite-sese-to-poly.c (extract_affine): Fix assert.
37997 (create_pw_aff_from_tree): Take loop parameter.
37998 (add_condition_to_pbb): Pass loop of the condition to
37999 create_pw_aff_from_tree.
38001 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38003 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38004 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38006 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38010 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38012 (convert_int_to_float128): Likewise.
38013 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38014 (convert_int_to_float128): Likewise.
38015 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38016 (UNSPEC_IEEE128_CONVERT): Likewise.
38017 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38018 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38019 Use local variables for IBM extended format.
38020 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38021 (fix_trunc<mode>si2_fprs): Likewise.
38022 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38023 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38024 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38025 to know that we can now have integers of all sizes in vector
38027 (fix<uns>_<mode>di2_hw): Likewise.
38028 (float<uns>_<mode>si2_hw): Likewise.
38029 (fix_<mode>si2_hw): Likewise.
38030 (fixuns_<mode>si2_hw): Likewise.
38031 (float<uns>_<mode>di2_hw): Likewise.
38032 (float_<mode>di2_hw): Likewise.
38033 (float_<mode>si2_hw): Likewise.
38034 (floatuns_<mode>di2_hw): Likewise.
38035 (floatuns_<mode>si2_hw): Likewise.
38036 (xscvqp<su>wz_<mode>): Delete, no longer used.
38037 (xscvqp<su>dz_<mode>): Likewise.
38038 (xscv<su>dqp_<mode>): Likewise.
38039 (ieee128_mfvsrd_64bit): Likewise.
38040 (ieee128_mfvsrd_32bit): Likewise.
38041 (ieee128_mfvsrwz): Likewise.
38042 (ieee128_mtvsrw): Likewise.
38043 (ieee128_mtvsrd_64bit): Likewise.
38044 (ieee128_mtvsrd_32bit): Likewise.
38046 2017-01-31 Martin Liska <mliska@suse.cz>
38049 * ipa-prop.c (ipa_free_all_node_params): Call release method
38050 instead of ~sumbol_summary to not to trigger double times
38053 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38055 PR tree-optimization/71691
38056 * bitmap.h (class auto_bitmap): New.
38057 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38058 is_maybe_undefined instead of ssa_undefined_value_p.
38060 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38062 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38063 __S390_ARCH_LEVEL__ to __ARCH__.
38065 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38067 PR tree-optimization/79267
38068 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38069 if should_remove_lhs_p is true.
38071 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38074 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38075 (add_alignment_attribute): New.
38076 (base_type_die): Add alignment attribute.
38077 (subrange_type_die): Likewise.
38078 (modified_type_die): Likewise.
38079 (gen_array_type_die): Likewise.
38080 (gen_descr_array_type_die: Likewise.
38081 (gen_enumeration_type_die): Likewise.
38082 (gen_subprogram_die): Likewise.
38083 (gen_variable_die): Likewise.
38084 (gen_field_die): Likewise.
38085 (gen_ptr_to_mbr_type_die): Likewise.
38086 (gen_struct_or_union_type_die): Likewise.
38087 (gen_subroutine_type_die): Likewise.
38088 (gen_typedef_die): Likewise.
38089 (base_type_cmp): Compare alignment attribute.
38091 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38094 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38095 (setb_unsigned) New pattern for setb with CCUNS.
38096 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38097 subfc./subfe sequence to avoid overflow problems. Generate a
38098 shorter sequence with cmpld/setb for power9.
38099 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38100 for generating subfc. instruction.
38101 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38102 now uses this instruction.
38104 2017-01-30 Ian Lance Taylor <iant@google.com>
38107 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38108 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38110 2017-01-30 Martin Sebor <msebor@redhat.com>
38112 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38113 Move constant to the right of a relational operator.
38114 (get_mpfr_format_length, format_character, format_string): Ditto.
38115 (should_warn_p, maybe_warn): Same.
38117 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38119 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38122 * asan.c (get_translation_unit_decl): Remove function.
38123 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38125 2017-01-30 Martin Liska <mliska@suse.cz>
38127 PR gcov-profile/79259
38128 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38129 -fprofile-generate.
38131 2017-01-30 Martin Liska <mliska@suse.cz>
38134 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38135 Initialize variables with NULL value.
38137 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38140 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38142 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38144 2017-01-30 Richard Biener <rguenther@suse.de>
38146 PR tree-optimization/79276
38147 * tree-vrp.c (process_assert_insertions): Properly adjust common
38148 when removing a duplicate.
38150 2017-01-30 Richard Biener <rguenther@suse.de>
38152 PR tree-optimization/79256
38153 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38154 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38156 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38158 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38161 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38162 ("*r<noxa>sbg_<mode>_sll_bitmask")
38163 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38164 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38165 Use contiguous_bitmask_nowrap_operand.
38167 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38170 * config/rs6000/altivec.h (vec_xl): Revise #define.
38171 (vec_xst): Likewise.
38173 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38175 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38177 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38179 PR rtl-optimization/79194
38180 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38181 traps before call to bypass_conditional_jumps.
38183 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38185 PR tree-optimization/71374
38186 * lra-constraints.c (check_conflict_input_operands): New.
38187 (match_reload): Use it.
38189 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38192 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38193 account to calculate conflict_set.
38195 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38197 PR rtl-optimization/78559
38198 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38199 other_insn in combine.
38201 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38203 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38204 uint16_type_node for BT_UINT16.
38206 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38208 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38209 "RTL Tests" to menu.
38210 (GIMPLE Tests): New node.
38211 (RTL Tests): New node.
38213 2017-01-27 Richard Biener <rguenther@suse.de>
38215 PR tree-optimization/79245
38216 * tree-loop-distribution.c (distribute_loop): Apply cost
38217 modeling also to detected patterns.
38219 2017-01-27 Richard Biener <rguenther@suse.de>
38221 PR tree-optimization/71433
38222 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38223 (compare_assert_loc): New function.
38224 (process_assert_insertions): Sort and optimize assert locations
38225 to remove duplicates and push down identical assertions on
38226 edges to their destination block.
38228 2017-01-27 Richard Biener <rguenther@suse.de>
38230 PR tree-optimization/79244
38231 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38232 out SSA names even if abnormal.
38234 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38236 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38237 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38238 instead of MPFR_RNDN.
38240 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38243 * arm.c (arm_option_override): Don't call build_target_option_node
38244 until after doing all option overrides.
38245 (arm_valid_target_attribute_tree): Likewise.
38247 2017-01-27 Martin Liska <mliska@suse.cz>
38249 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38250 for {cd}tors and C++ {cd}tors.
38252 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38254 * config/s390/s390.md ("*setmem_long_and")
38255 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38257 2017-01-26 Martin Sebor <msebor@redhat.com>
38259 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38262 2017-01-26 Martin Sebor <msebor@redhat.com>
38264 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38265 HAVE_DFmode before using XFmode or DFmode.
38266 (parse_directive): Avoid using the z length modifier to avoid
38267 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38269 PR middle-end/78703
38270 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38271 to accept adjustment as an array.
38272 (get_int_range): New function.
38273 (struct directive): Make width and prec arrays.
38274 (directive::set_width, directive::set_precision): Call get_int_range.
38275 (format_integer, format_floating): Handle width and precision ranges.
38276 (format_string, parse_directive): Same.
38278 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38281 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38282 comdat_type_p set, just clone them, but keep the children in the
38286 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38287 which have direct callers with -fvar-tracking-assignments enabled
38289 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38290 inside of type units.
38292 2017-01-26 Martin Sebor <msebor@redhat.com>
38294 PR middle-end/78703
38295 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38297 (struct format_result): Replace number_chars, number_chars_min,
38298 and number_chars_max with a single member of struct result_range.
38300 (format_result::operator+=): Adjust.
38301 (struct fmtresult): Remove bounded. Handle likely and unlikely
38303 (fmtresult::adjust_for_width_or_precision): New function.
38304 (fmtresult:type_max_digits): New function.
38305 (bytes_remaining): Handle likely and unlikely counters.
38306 (min_bytes_remaining): Remove.
38307 (format_percent): Simplify.
38308 (format_integer, format_floating): Set likely and unlikely counters.
38309 (get_string_length, format_character, format_string): Same.
38310 (format_plain, should_warn_p): New function.
38311 (maybe_warn): Call should_warn_p. Update diagnostic messages
38312 and handle those for all directives, including plain strings.
38313 (format_directive): Handle likely and unlikely counters.
38314 Remove unnecessary quoting from diagnostics. Add an informational
38316 (add_bytes): Remove.
38317 (pass_sprintf_length::compute_format_length): Simplify.
38318 (try_substitute_return_value): Handle likely and unlikely counters.
38320 2017-01-26 Carl Love <cel@us.ibm.com>
38322 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38323 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38325 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38328 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38329 endianess for subregs into account.
38330 * lra-constraints.c (lra_constraints): Do risky transformations
38331 always on the first iteration.
38332 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38333 last_call_used_reg_set.
38334 (process_bb_lives): Define and use last_call_used_reg_set.
38335 * lra.c (lra): Always continue after lra_constraints on the first
38338 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38340 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38342 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38344 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38346 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38347 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38348 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38349 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38350 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38351 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38352 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38353 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38354 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38356 2017-01-26 Marek Polacek <polacek@redhat.com>
38359 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38360 for the third operand.
38362 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38364 PR middle-end/79236
38365 * omp-low.c (struct omp_context): Add simt_stmt field.
38366 (scan_omp_for): Return omp_context *.
38367 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38368 context to the _simt_ SIMD stmt.
38369 (lower_omp_for): For combined SIMD with sibling _simt_
38370 SIMD, make sure to use the same decls in _looptemp_
38371 clauses as in the sibling.
38373 2017-01-26 David Sherwood <david.sherwood@arm.com>
38375 PR middle-end/79212
38376 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38379 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38382 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38383 emit fld b; fld a; if possible.
38385 * brig-builtins.def: Update copyright years.
38386 * config/arm/arm_acle_builtins.def: Update copyright years.
38388 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38391 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38392 constraint instead of o for the stxsd instruction.
38394 2017-01-25 Carl Love <cel@us.ibm.com>
38396 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38397 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38399 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38401 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38403 2017-01-25 Richard Biener <rguenther@suse.de>
38405 PR tree-optimization/69264
38406 * target.def (vector_alignment_reachable): Improve documentation.
38407 * doc/tm.texi: Regenerate.
38408 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38410 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38411 earlier changes with respect to TYPE_USER_ALIGN.
38412 (vector_alignment_reachable_p): Likewise. Improve dumping.
38414 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38417 * config/arm/arm.md (xordi3): Force constant operand into a register
38420 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38422 * doc/invoke.texi (-fstore-merging): Correct default optimization
38423 levels at which it is enabled.
38424 (-O): Move -fstore-merging from list to...
38427 2017-01-25 Richard Biener <rguenther@suse.de>
38430 * omp-expand.c: Include debug.h.
38431 (expand_omp_taskreg): Make sure to generate early debug before
38432 outlining anything from a function.
38433 (expand_omp_target): Likewise.
38434 (grid_expand_target_grid_body): Likewise.
38436 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38439 * asan.c (get_translation_unit_decl): New function.
38440 (asan_add_global): Extract modules file name from globals
38441 TRANSLATION_UNIT_DECL name.
38443 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38446 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38447 for long calls with APCS frame and VFP.
38449 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38451 * cfg.c (original_copy_tables_initialized_p): New function.
38452 * cfg.h (original_copy_tables_initialized_p): New decl.
38453 * cfgrtl.c (relink_block_chain): Guard the call to
38454 free_original_copy_tables with a call to
38455 original_copy_tables_initialized_p.
38456 * cgraph.h (symtab_node::native_rtl_p): New decl.
38457 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38458 (symtab_node::needed_p): Don't assert for early assembly output
38459 for __RTL functions.
38460 (cgraph_node::finalize_function): Set "force_output" for __RTL
38462 (cgraph_node::analyze): Bail out early for __RTL functions.
38463 (analyze_functions): Update assertion to support __RTL functions.
38464 (cgraph_node::expand): Bail out early for __RTL functions.
38465 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38467 * function.h (struct function): Update comment for field
38469 * gimple-expr.c: Include "tree-pass.h".
38470 (gimple_has_body_p): Return false for __RTL functions.
38471 * Makefile.in (OBJS): Add run-rtl-passes.o.
38472 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38474 (gcc::pass_manager::get_clean_slate): New accessor.
38475 * passes.c: Include "insn-addr.h".
38476 (should_skip_pass_p): Add logging. Update logic for running
38477 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38478 property-provider override so it is only done for gimple passes.
38480 (skip_pass): New function.
38481 (execute_one_pass): Call skip_pass when skipping passes.
38482 * read-md.c (md_reader::read_char): Support filtering
38483 the input to a subset of line numbers.
38484 (md_reader::md_reader): Initialize fields
38485 m_first_line and m_last_line.
38486 (md_reader::read_file_fragment): New function.
38487 * read-md.h (md_reader::read_file_fragment): New decl.
38488 (md_reader::m_first_line): New field.
38489 (md_reader::m_last_line): New field.
38490 * read-rtl-function.c (function_reader::create_function): Only
38491 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38492 curr_properties. Set DECL_INITIAL to a dummy block.
38493 (read_rtl_function_body_from_file_range): New function.
38494 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38496 * run-rtl-passes.c: New file.
38497 * run-rtl-passes.h: New file.
38499 2017-01-24 Jeff Law <law@redhat.com>
38501 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38504 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38506 PR tree-optimization/79159
38507 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38508 (record_nonwrapping_iv): Improve boundary using above function if no
38509 value range information.
38511 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38512 Martin Jambor <mjambor@suse.cz>
38514 * brig-builtins.def: New file.
38515 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38516 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38517 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38518 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38519 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38520 * builtin-types.def (BT_INT8): New.
38521 (BT_INT16): Likewise.
38522 (BT_UINT8): Likewise.
38523 (BT_UINT16): Likewise.
38524 (BT_FN_ULONG): Likewise.
38525 (BT_FN_UINT_INT): Likewise.
38526 (BT_FN_UINT_ULONG): Likewise.
38527 (BT_FN_UINT_LONG): Likewise.
38528 (BT_FN_UINT_PTR): Likewise.
38529 (BT_FN_ULONG_PTR): Likewise.
38530 (BT_FN_INT8_FLOAT): Likewise.
38531 (BT_FN_INT16_FLOAT): Likewise.
38532 (BT_FN_UINT32_FLOAT): Likewise.
38533 (BT_FN_UINT16_FLOAT): Likewise.
38534 (BT_FN_UINT8_FLOAT): Likewise.
38535 (BT_FN_UINT64_FLOAT): Likewise.
38536 (BT_FN_UINT16_UINT32): Likewise.
38537 (BT_FN_UINT32_UINT16): Likewise.
38538 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38539 (BT_FN_INT_PTR_INT): Likewise.
38540 (BT_FN_UINT_PTR_UINT): Likewise.
38541 (BT_FN_LONG_PTR_LONG): Likewise.
38542 (BT_FN_ULONG_PTR_ULONG): Likewise.
38543 (BT_FN_VOID_UINT64_UINT64): Likewise.
38544 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38545 (BT_FN_INT8_INT8_INT8): Likewise.
38546 (BT_FN_INT16_INT16_INT16): Likewise.
38547 (BT_FN_INT_INT_INT): Likewise.
38548 (BT_FN_UINT_FLOAT_UINT): Likewise.
38549 (BT_FN_FLOAT_UINT_UINT): Likewise.
38550 (BT_FN_ULONG_UINT_UINT): Likewise.
38551 (BT_FN_ULONG_UINT_PTR): Likewise.
38552 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38553 (BT_FN_UINT_UINT_UINT): Likewise.
38554 (BT_FN_VOID_UINT_PTR): Likewise.
38555 (BT_FN_UINT_UINT_PTR: Likewise.
38556 (BT_FN_UINT32_UINT64_PTR): Likewise.
38557 (BT_FN_INT_INT_UINT_UINT): Likewise.
38558 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38559 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38560 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38561 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38562 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38563 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38564 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38565 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38566 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38567 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38568 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38569 * doc/frontends.texi: List BRIG FE.
38570 * doc/install.texi (Testing): Add BRIG tesring requirements.
38571 * doc/invoke.texi (Overall Options): Mention BRIG.
38572 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38574 2017-01-24 Richard Biener <rguenther@suse.de>
38576 PR translation/79208
38577 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38579 2017-01-24 Martin Jambor <mjambor@suse.cz>
38582 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38583 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38584 and known_contexts.
38586 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38588 PR middle-end/79123
38589 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38590 casts from signed to unsigned really don't have a range.
38592 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38594 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38595 GMP_RNDx for compatiblity.
38597 2017-01-24 Martin Liska <mliska@suse.cz>
38600 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38601 that would prevent us to call alloca with -1 as argument.
38603 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38605 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38606 Avoid -Wformat-security warning.
38608 2017-01-23 Andrew Pinski <apinski@cavium.com>
38610 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38613 2017-01-23 Martin Sebor <msebor@redhat.com>
38615 PR middle-end/78703
38616 * gimple-ssa-sprintf.c (warn_level): New global.
38617 (format_integer): Use it here and throughout the rest of the file.
38618 Use the same switch to compute sign as base.
38619 (maybe_warn): New function.
38620 (format_directive): Factor out warnings into maybe_warn.
38621 Add debugging output. Use warn_level.
38622 (add_bytes): Use warn_level.
38623 (pass_sprintf_length::compute_format_length): Add debugging output.
38624 (try_substitute_return_value): Same.
38625 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38627 PR middle-end/78703
38628 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38629 (struct fmtresult, format_integer, format_floating): Adjust.
38630 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38631 (get_string_length, format_string,format_directive): Same.
38632 (pass_sprintf_length::compute_format_length): Same.
38633 (try_substitute_return_value): Simplify slightly.
38635 PR middle-end/78703
38636 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38637 (fmtresult::operator+=): Outlined.
38638 (struct fmtresult): Add ctors.
38639 (struct conversion_spec): Rename...
38640 (struct directive): ...to this. Add and remove data members.
38641 (directive::set_width, directive::set_precision): New functions.
38642 (format_percent): Use fmtresult ctor.
38643 (get_width_and_precision): Remove.
38644 (format_integer): Make naming changes. Avoid computing width and
38646 (format_floating): Same. Adjust indentation.
38647 (format_character, format_none): New functions.
38648 (format_string): Moved character handling to format_character.
38649 (format_directive): Remove arguments, change return type.
38650 (parse_directive): New function.
38651 (pass_sprintf_length::compute_format_length): Move directive
38652 parsing to parse_directive.
38654 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38656 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38657 (assign_assembler_name_if_needed): ... this.
38658 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38659 (assign_assembler_name_if_needed): ... this.
38660 (free_lang_data_in_cgraph): Adjust callers.
38661 * cgraphunit.c (cgraph_node::analyze): Likewise.
38662 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38665 2017-01-23 Richard Biener <rguenther@suse.de>
38667 PR tree-optimization/79088
38668 PR tree-optimization/79188
38669 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38670 resetting loop bounds after last path deletion. Reset loop
38671 bounds of the target loop, make code match the comments.
38672 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38673 Make sure loops need no fixups.
38675 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38677 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38678 exponent support with double type for first argument.
38679 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38680 type returned by __builtin_vec_extract_sig,
38681 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38682 functions from "vector int" to "vector unsigned int" or from
38683 "vector long long int" to "vector unsigned long long int".
38684 Changed type returned by __builtin_vec_extract_exp,
38685 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38686 functions from "vector int" to "vector unsigned int" or from
38687 "vector long long int" to "vector unsigned long long int".
38688 Changed return type of __builtin_vec_test_data_class,
38689 __builtin_vec_test_data_class_sp, and
38690 __builtin_vec_test_data_class_dp from "vector int" to
38691 "vector bool int" or from "vector long long int" to "vector bool
38692 long long int" and changed second argument type from "unsigned
38693 int" to "int". Added new overloaded function forms "vector float
38694 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38695 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38696 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38697 double, vector unsigned long long int)" and "vector double
38698 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38699 long int)". Changed return type of
38700 __builtin_scalar_test_data_class and
38701 __builtin_scalar_test_data_class_sp and
38702 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38703 int" and changed second argument from "unsigned int" to "int".
38704 Changed type returned by __builtin_scalar_test_neg,
38705 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38706 from "int" to "bool int". Added new overloaded function form
38707 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38708 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38709 exponent double-precision with floating point first argument.
38710 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38711 documentation of scalar_test_data_class, scalar_test_neg,
38712 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38713 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38714 vec_test_data_class built-in functions to reflect refinements in
38715 their type signatures.
38717 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38719 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38721 (aarch64_elf_asm_destructor): Likewise.
38723 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38725 PR rtl-optimization/78634
38726 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38727 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38728 * ifcvt.c (noce_try_cmove): Add missing cost check.
38730 PR rtl-optimization/71724
38731 * combine.c (if_then_else_cond): Look for situations where it is
38732 beneficial to undo the work of one of the recursive calls.
38734 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38736 PR tree-optimization/70754
38737 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38738 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38739 combined stmt before it if not NULL.
38740 (combine_chains): Process refs reversely and compute dominance point
38743 2017-01-23 Martin Liska <mliska@suse.cz>
38745 PR tree-optimization/79196
38746 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38747 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38749 (strlen_optimize_stmt): Call the renamed function.
38751 2017-01-23 Michael Matz <matz@suse.de>
38753 PR tree-optimization/78384
38754 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38756 2017-01-23 Richard Biener <rguenther@suse.de>
38758 PR tree-optimization/79186
38759 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38760 both incoming edges before moving an assert.
38762 2017-01-23 Martin Jambor <mjambor@suse.cz>
38764 * ipa-prop.c (load_from_param_1): Removed.
38765 (load_from_unmodified_param): Bits from load_from_param_1 put back
38767 (load_from_param): Removed.
38768 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38769 with stmt. Reverted back to use of load_from_unmodified_param.
38771 2017-01-23 Martin Jambor <mjambor@suse.cz>
38774 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38775 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38776 field a pointer to garbage collected vector, mark lattices and
38777 ipcp_orig_node with GTY((skip)).
38778 (ipa_get_param_count): Adjust to descriptors being a pointer.
38779 (ipa_get_param): Likewise.
38780 (ipa_get_type): Likewise.
38781 (ipa_get_param_move_cost): Likewise.
38782 (ipa_set_param_used): Likewise.
38783 (ipa_get_controlled_uses): Likewise.
38784 (ipa_set_controlled_uses): Likewise.
38785 (ipa_is_param_used): Likewise.
38786 (ipa_node_params_t): Move into garbage collector. New methods insert
38788 (ipa_node_params_sum): Annotate wth GTY(()).
38789 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38791 (ipa_load_from_parm_agg): Adjust declaration.
38792 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38793 * ipa-profile.c (ipa_profile): Likewise.
38794 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
38795 (ipa_populate_param_decls): Make descriptors parameter garbage
38797 (ipa_dump_param): Adjust to descriptors being a pointer.
38798 (ipa_alloc_node_params): Likewise.
38799 (ipa_initialize_node_params): Likewise.
38800 (load_from_param_1): Make descriptors parameter garbage collected.
38801 (load_from_unmodified_param): Likewise.
38802 (load_from_param): Likewise.
38803 (ipa_load_from_parm_agg): Likewise.
38804 (ipa_node_params::~ipa_node_params): Removed.
38805 (ipa_free_all_node_params): Remove call to delete operator.
38806 (ipa_node_params_t::insert): New.
38807 (ipa_node_params_t::remove): Likewise.
38808 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
38809 copy known_csts and known_contexts vectors.
38810 (ipa_read_node_info): Adjust to descriptors being a pointer.
38811 (ipcp_modif_dom_walker): Make m_descriptors field garbage
38813 (ipcp_transform_function): Make descriptors variable garbage
38816 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
38818 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
38819 * config/i386/avx512dqintrin.h: Ditto.
38820 * config/i386/avx512fintrin.h: Ditto.
38821 * config/i386/i386.c: Handle new builtins.
38822 * config/i386/i386-builtin.def: Add new builtins.
38823 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
38824 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
38826 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38827 Martin Liska <mliska@suse.cz>
38829 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
38830 * asan.c (asan_expand_poison_ifn): Support stores and use
38831 appropriate ASAN report function.
38832 * internal-fn.c (expand_ASAN_POISON_USE): New function.
38833 * internal-fn.def (ASAN_POISON_USE): Declare.
38834 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
38835 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
38836 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
38837 ASAN_POISON calls w/o LHS.
38838 * tree-ssa.c (execute_update_addresses_taken): Create clobber
38839 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
38840 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
38841 * gimplify.c (asan_poison_variables): Add attribute
38842 use_after_scope_memory to variables that really needs to live
38844 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
38845 having the attribute.
38847 2017-01-23 Martin Liska <mliska@suse.cz>
38849 * asan.c (create_asan_shadow_var): New function.
38850 (asan_expand_poison_ifn): Likewise.
38851 * asan.h (asan_expand_poison_ifn): New declaration.
38852 * internal-fn.c (expand_ASAN_POISON): Likewise.
38853 * internal-fn.def (ASAN_POISON): New builtin.
38854 * sanopt.c (pass_sanopt::execute): Expand
38855 asan_expand_poison_ifn.
38856 * tree-inline.c (copy_decl_for_dup_finish): Make function
38858 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
38859 * tree-ssa.c (is_asan_mark_p): New function.
38860 (execute_update_addresses_taken): Rewrite local variables
38861 (identified just by use-after-scope as addressable) into SSA.
38863 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38865 * doc/install.texi (Specific): opensource.apple.com uses https
38866 now. Remove trailing slash.
38868 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38870 * README.Portability: Remove note on an Irix compatibility issue.
38872 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
38874 * gcov.c (INCLUDE_ALGORITHM): Define.
38875 (INCLUDE_VECTOR): Define.
38876 No longer include <vector> and <algorithm> directly.
38878 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
38880 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
38882 * doc/invoke.texi (Code Gen Options): Ditto.
38884 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
38887 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
38889 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
38891 rtl-optimization/79125
38892 * cprop.c (local_cprop_pass): Handle cases where we make an
38893 unconditional trap.
38895 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
38899 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
38900 read from, for big endian.
38902 2017-01-20 Jiong Wang <jiong.wang@arm.com>
38904 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
38905 register pauth builtins for LP64 only.
38907 2017-01-20 Marek Polacek <polacek@redhat.com>
38910 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
38913 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38915 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
38917 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
38918 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
38919 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
38921 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38924 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
38925 in big-endian mode when they are not a single duplicated value.
38927 2017-01-20 Richard Biener <rguenther@suse.de>
38929 * BASE-VER: Bump to 7.0.1.
38931 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38933 * omp-low.c (omplow_simd_context): New struct. Use it...
38934 (lower_rec_simd_input_clauses): ...here and...
38935 (lower_rec_input_clauses): ...here to hold common data. Adjust all
38936 references to idx, lane, max_vf, is_simt.
38938 2017-01-20 Graham Markall <graham.markall@embecosm.com>
38940 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
38943 2017-01-20 Martin Jambor <mjambor@suse.cz>
38945 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
38946 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
38948 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
38949 (GTFILES): Rename hsa.c to hsa-common.c.
38950 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
38951 * hsa-dump.c: Likewise.
38952 * hsa-gen.c: Likewise.
38953 * hsa-regalloc.c: Likewise.
38954 * ipa-hsa.c: Likewise.
38955 * omp-expand.c: Likewise.
38956 * omp-low.c: Likewise.
38957 * toplev.c: Likewise.
38959 2017-01-20 Marek Polacek <polacek@redhat.com>
38962 * doc/invoke.texi: Document -Wduplicated-branches.
38963 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
38964 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
38965 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
38966 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
38967 return 0 only when not OEP_LEXICOGRAPHIC.
38968 (fold_build_cleanup_point_expr): Use the expression
38969 location when building CLEANUP_POINT_EXPR.
38970 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
38971 * tree.c (add_expr): Handle error_mark_node.
38973 2017-01-20 Martin Liska <mliska@suse.cz>
38976 * tree-profile.c (init_ic_make_global_vars): Do not call
38978 (gimple_init_gcov_profiler): Likewise.
38980 2017-01-20 Martin Liska <mliska@suse.cz>
38983 * cgraph.h (maybe_create_reference): Remove argument and
38985 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
38987 * ipa-cp.c (create_specialized_node): Likewise.
38988 * symtab.c (symtab_node::maybe_create_reference): Handle
38989 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
38991 2017-01-20 Martin Liska <mliska@suse.cz>
38993 * read-rtl-function.c (function_reader::create_function): Use
38994 build_decl instread of build_decl_stat.
38996 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
38998 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
38999 * config/i386/avx512dqintrin.h: Ditto.
39000 * config/i386/avx512fintrin.h: Ditto.
39001 * config/i386/i386-builtin-types.def: Add new types.
39002 * config/i386/i386.c: Handle new types.
39003 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39004 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39005 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39006 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39007 (__builtin_ia32_kshiftridi): New.
39008 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39010 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39014 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39015 define to rs6000_init_stack_protect_guard.
39016 (rs6000_init_stack_protect_guard): New function.
39018 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39019 Yunqiang Su <yunqiang.su@imgtec.com>
39021 * config.gcc (supported_defaults): Add madd4.
39022 (with_madd4): Add validation.
39023 (all_defaults): Add madd4.
39024 * config/mips/mips.opt (mmadd4): New option.
39025 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39027 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39029 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39030 (ISA_HAS_FUSED_MADD4): Likewise.
39031 * doc/invoke.texi (-mmadd4): Document the new option.
39032 * doc/install.texi (--with-madd4): Document the new option.
39034 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39036 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39037 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39038 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39039 (aarch64_init_pauth_hint_builtins): New.
39040 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39041 (aarch64_expand_builtin): Expand new builtins.
39043 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39045 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39046 * combine-stack-adj.c (no_unhandled_cfa): Handle
39047 REG_CFA_TOGGLE_RA_MANGLE.
39048 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39049 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39050 info for return address signing.
39051 (aarch64_expand_epilogue): Likewise.
39053 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39055 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39056 * config/aarch64/aarch64-protos.h
39057 (aarch64_return_address_signing_enabled): New declaration.
39058 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39060 (aarch64_expand_prologue): Sign return address before it's pushed onto
39062 (aarch64_expand_epilogue): Authenticate return address fetched from
39064 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39065 (aarch64_attributes): New function attributes for "sign-return-address".
39066 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39067 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39068 ("*do_return"): Generate combined instructions according to key index.
39069 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39070 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39072 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39073 * config/aarch64/aarch64.opt (msign-return-address=): New.
39074 * doc/extend.texi (AArch64 Function Attributes): Documents
39075 "sign-return-address=".
39076 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39078 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39080 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39081 overall option summary.
39083 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39085 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39086 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39087 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39088 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39090 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39092 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39093 -mpower9-minmax by default for -mcpu=power9.
39094 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39095 128-bit floating point.
39097 2017-01-20 Alan Modra <amodra@gmail.com>
39099 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39100 optimizing for size.
39102 2017-01-20 Alan Modra <amodra@gmail.com>
39105 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39106 for strcmp and strncmp from corresponding builtin decl.
39108 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39110 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39111 instead of i386/rtems-64.h.
39112 * config/i386/rtems-64.h: Remove.
39114 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39118 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39120 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39122 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39124 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39125 Change int to HOST_WIDE_INT.
39126 * config/aarch64/aarch64-protos.h
39127 (aarch64_simd_gen_const_vector_dup): Likewise.
39128 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39130 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39132 * langhooks-def.h (lhd_type_for_size): New decl.
39133 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39134 * langhooks.c (lhd_type_for_size): New function, taken from
39137 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39139 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39140 define_bypass for CR latency.
39141 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39142 (power9-alu2): Add define_bypass for CR latency.
39144 (power9-mul): Update insn latency.
39145 (power9-mul-compare): Update insn latency, bypass latency and remove
39148 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39150 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39152 * config/aarch64/aarch64.md
39153 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39154 aarch64_nopcrelative_literal_loads.
39155 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39157 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39159 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39160 TARGET_LOONGSON_3A.
39161 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39163 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39166 * config.gcc (supported_defaults): Add lxc1-sxc1.
39167 (with_lxc1_sxc1): Add validation.
39168 (all_defaults): Add lxc1-sxc1.
39169 * config/mips/mips.opt (mlxc1-sxc1): New option.
39170 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39172 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39173 __mips_no_lxc1_sxc1.
39174 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39175 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39176 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39178 2017-01-19 Richard Biener <rguenther@suse.de>
39180 PR tree-optimization/72488
39181 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39182 sure to restore SSA info.
39183 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39185 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39187 PR rtl-optimization/79121
39188 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39189 of the inner type when shifting an extended value.
39191 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39194 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39195 interposable aliases.
39197 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39200 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39201 Use the evmergelohi instruction.
39202 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39203 (mov_sitf_e500_subreg8_2_be): Likewise.
39204 (mov_sitf_e500_subreg12_2_le): Likewise.
39205 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39206 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39207 (mov_sitf_e500_subreg8_2_le): Likewise.
39208 (mov_sitf_e500_subreg12_2_be): Likewise.
39210 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39212 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39213 attribute from vecsimple to vecperm.
39214 (altivec_vbpermq2): Likewise.
39216 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39219 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39221 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39222 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39223 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39224 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39225 case where N arg is SIZE_MAX.
39226 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39227 (cmpstrsi): Add pattern.
39229 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39231 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39232 __builtin_vec_revb builtins.
39233 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39234 built-in functions to support generation of the ISA 3.0 XXBR<x>
39235 vector byte reverse instructions.
39236 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39237 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39238 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39239 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39240 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39241 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39242 (P9V_BUILTIN_VEC_REVB): Likewise.
39243 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39244 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39245 (p9_xxbrq_v16qi): Likewise.
39246 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39247 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39248 (p9_xxbrh_v8hi): Likewise.
39249 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39250 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39251 vec_revb built-in functions.
39253 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39255 PR rtl-optimization/78952
39256 * config/i386/i386.md (any_extract): New code iterator.
39257 (*insvqi_2): Use any_extract for source operand.
39258 (*insvqi_3): Use any_shiftrt for source operand.
39260 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39262 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39264 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39266 2017-01-18 Matthias Klose <doko@ubuntu.com>
39268 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39270 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39272 * config/rs6000/altivec.h (vec_bperm): Change #define.
39273 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39274 (altivec_vbpermq2): New define_insn.
39275 (altivec_vbpermd): Likewise.
39276 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39277 function interface.
39278 (VBPERMD): Likewise.
39279 (VBPERM): New polymorphic function interface.
39280 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39281 Add entries for P9V_BUILTIN_VEC_VBPERM.
39282 * doc/extend.texi: Add interfaces for vec_bperm.
39284 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39286 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39287 first letter of error messages.
39288 (s390_resolve_overloaded_builtin): Likewise.
39289 * config/s390/s390.c (s390_expand_builtin): Likewise.
39290 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39291 (s390_valid_target_attribute_inner_p): Likewise.
39292 * config/s390/s390.md ("tabort"): Likewise.
39294 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39296 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39297 (ISA_AVOID_DIV_HILO): New macro.
39298 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39299 (ISA_HAS_DDIV): Likewise.
39301 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39303 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39305 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39307 * doc/invoke.texi (fabi-version): Spelling fix.
39309 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39312 * doc/invoke.texi (fabi-version): Mention mangling fix for
39315 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39318 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39320 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39323 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39324 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39326 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39328 (stack_protect_test): Ditto.
39329 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39330 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39332 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39333 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39334 -mstack-protector-guard-offset=.
39335 (RS/6000 and PowerPC Options): Ditto.
39337 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39339 * config/i386/i386.h (MASK_CLASS_P): New define.
39340 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39341 there are no registers from different register sets also when
39342 mask registers are used. Update function comment.
39343 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39344 to (*k/*r) and (*k/*km) alternatives.
39346 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39348 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39349 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39350 (EH_RETURN_HANDLER_RTX): New define.
39351 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39352 Force frame pointer in EH return functions.
39353 (aarch64_expand_epilogue): Add barrier for eh_return.
39354 (aarch64_final_eh_return_addr): Remove.
39355 (aarch64_eh_return_handler_rtx): New function.
39356 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39358 (aarch64_eh_return_handler_rtx): New prototype.
39360 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39362 * config/rs6000/altivec.h (vec_rlmi): New #define.
39363 (vec_vrlnm): Likewise.
39364 (vec_rlnm): Likewise.
39365 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39366 (UNSPEC_VRLNM): Likewise.
39367 (VIlong): New mode iterator.
39368 (altivec_vrl<VI_char>mi): New define_insn.
39369 (altivec_vrl<VI_char>nm): Likewise.
39370 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39372 (VRLDNM): Likewise.
39373 (RLNM): New polymorphic function entry.
39374 (VRLWMI): New monomorphic function entry.
39375 (VRLDMI): Likewise.
39376 (RLMI): New polymorphic function entry.
39377 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39378 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39379 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39382 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39385 * dwarf2out.c (field_byte_offset): Restore the
39386 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39387 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39388 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39391 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39394 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39396 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39399 * dwarf2out.c (add_data_member_location_attribute): For constant
39400 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39401 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39402 DW_AT_byte_size attributes.
39404 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39406 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39407 after forcing to constant memory when the code model is medium.
39409 2017-01-17 Julia Koval <julia.koval@intel.com>
39412 * config/i386/avx512fintrin.h
39413 (_mm512_i32gather_ps): Change __addr type to void const*.
39414 (_mm512_mask_i32gather_ps): Ditto.
39415 (_mm512_i32gather_pd): Ditto.
39416 (_mm512_mask_i32gather_pd): Ditto.
39417 (_mm512_i64gather_ps): Ditto.
39418 (_mm512_mask_i64gather_ps): Ditto.
39419 (_mm512_i64gather_pd): Ditto.
39420 (_mm512_mask_i64gather_pd): Ditto.
39421 (_mm512_i32gather_epi32): Ditto.
39422 (_mm512_mask_i32gather_epi32): Ditto.
39423 (_mm512_i32gather_epi64): Ditto.
39424 (_mm512_mask_i32gather_epi64): Ditto.
39425 (_mm512_i64gather_epi32): Ditto.
39426 (_mm512_mask_i64gather_epi32): Ditto.
39427 (_mm512_i64gather_epi64): Ditto.
39428 (_mm512_mask_i64gather_epi64): Ditto.
39429 (_mm512_i32scatter_ps): Change __addr type to void*.
39430 (_mm512_mask_i32scatter_ps): Ditto.
39431 (_mm512_i32scatter_pd): Ditto.
39432 (_mm512_mask_i32scatter_pd): Ditto.
39433 (_mm512_i64scatter_ps): Ditto.
39434 (_mm512_mask_i64scatter_ps): Ditto.
39435 (_mm512_i64scatter_pd): Ditto.
39436 (_mm512_mask_i64scatter_pd): Ditto.
39437 (_mm512_i32scatter_epi32): Ditto.
39438 (_mm512_mask_i32scatter_epi32): Ditto.
39439 (_mm512_i32scatter_epi64): Ditto.
39440 (_mm512_mask_i32scatter_epi64): Ditto.
39441 (_mm512_i64scatter_epi32): Ditto.
39442 (_mm512_mask_i64scatter_epi32): Ditto.
39443 (_mm512_i64scatter_epi64): Ditto.
39444 (_mm512_mask_i64scatter_epi64): Ditto.
39445 * config/i386/avx512pfintrin.h
39446 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39447 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39448 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39449 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39450 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39451 (_mm512_prefetch_i32scatter_ps): Ditto.
39452 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39453 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39454 (_mm512_prefetch_i64scatter_pd): Ditto.
39455 (_mm512_prefetch_i64scatter_ps): Ditto.
39456 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39457 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39458 * config/i386/avx512vlintrin.h
39459 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39460 (_mm_mmask_i32gather_ps): Ditto.
39461 (_mm256_mmask_i32gather_pd): Ditto.
39462 (_mm_mmask_i32gather_pd): Ditto.
39463 (_mm256_mmask_i64gather_ps): Ditto.
39464 (_mm_mmask_i64gather_ps): Ditto.
39465 (_mm256_mmask_i64gather_pd): Ditto.
39466 (_mm_mmask_i64gather_pd): Ditto.
39467 (_mm256_mmask_i32gather_epi32): Ditto.
39468 (_mm_mmask_i32gather_epi32): Ditto.
39469 (_mm256_mmask_i32gather_epi64): Ditto.
39470 (_mm_mmask_i32gather_epi64): Ditto.
39471 (_mm256_mmask_i64gather_epi32): Ditto.
39472 (_mm_mmask_i64gather_epi32): Ditto.
39473 (_mm256_mmask_i64gather_epi64): Ditto.
39474 (_mm_mmask_i64gather_epi64): Ditto.
39475 (_mm256_i32scatter_ps): Change __addr type to void*.
39476 (_mm256_mask_i32scatter_ps): Ditto.
39477 (_mm_i32scatter_ps): Ditto.
39478 (_mm_mask_i32scatter_ps): Ditto.
39479 (_mm256_i32scatter_pd): Ditto.
39480 (_mm256_mask_i32scatter_pd): Ditto.
39481 (_mm_i32scatter_pd): Ditto.
39482 (_mm_mask_i32scatter_pd): Ditto.
39483 (_mm256_i64scatter_ps): Ditto.
39484 (_mm256_mask_i64scatter_ps): Ditto.
39485 (_mm_i64scatter_ps): Ditto.
39486 (_mm_mask_i64scatter_ps): Ditto.
39487 (_mm256_i64scatter_pd): Ditto.
39488 (_mm256_mask_i64scatter_pd): Ditto.
39489 (_mm_i64scatter_pd): Ditto.
39490 (_mm_mask_i64scatter_pd): Ditto.
39491 (_mm256_i32scatter_epi32): Ditto.
39492 (_mm256_mask_i32scatter_epi32): Ditto.
39493 (_mm_i32scatter_epi32): Ditto.
39494 (_mm_mask_i32scatter_epi32): Ditto.
39495 (_mm256_i32scatter_epi64): Ditto.
39496 (_mm256_mask_i32scatter_epi64): Ditto.
39497 (_mm_i32scatter_epi64): Ditto.
39498 (_mm_mask_i32scatter_epi64): Ditto.
39499 (_mm256_i64scatter_epi32): Ditto.
39500 (_mm256_mask_i64scatter_epi32): Ditto.
39501 (_mm_i64scatter_epi32): Ditto.
39502 (_mm_mask_i64scatter_epi32): Ditto.
39503 (_mm256_i64scatter_epi64): Ditto.
39504 (_mm256_mask_i64scatter_epi64): Ditto.
39505 (_mm_i64scatter_epi64): Ditto.
39506 (_mm_mask_i64scatter_epi64): Ditto.
39507 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39508 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39509 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39510 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39511 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39512 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39513 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39514 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39515 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39516 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39517 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39518 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39519 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39520 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39521 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39522 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39523 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39524 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39525 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39526 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39527 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39528 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39529 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39530 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39531 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39532 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39533 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39534 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39535 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39536 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39537 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39538 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39539 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39540 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39541 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39542 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39543 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39544 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39545 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39546 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39547 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39548 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39549 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39550 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39551 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39552 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39553 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39554 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39555 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39556 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39557 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39558 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39559 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39560 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39561 definitions accordingly.
39563 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39564 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39567 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39570 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39573 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39574 hard regs for inner regno.
39576 2017-01-17 Martin Liska <mliska@suse.cz>
39579 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39580 assumption and add comment.
39582 2017-01-17 Nathan Sidwell <nathan@acm.org>
39584 * ipa-visibility.c (localize_node): New function, broken out of ...
39585 (function_and_variable_visibility): ... here. Call it.
39587 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39589 PR middle-end/77445
39590 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39591 correctly set frequency of oudgoing edge.
39592 (duplicate_thread_path): Fix profile updating.
39594 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39597 * configure.ac: Add GCC_BASE_VER.
39598 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39599 version from BASE-VER file.
39600 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39601 (gcc.o): Depend on $(BASEVER).
39602 * common.opt (dumpfullversion): New option.
39603 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39604 * doc/invoke.texi: Document -dumpfullversion.
39605 * doc/install.texi: Document --with-gcc-major-version-only.
39606 * configure: Regenerated.
39608 2017-01-17 Richard Biener <rguenther@suse.de>
39610 PR tree-optimization/71433
39611 * tree-vrp.c (register_new_assert_for): Merge same asserts
39612 on all incoming edges.
39613 (process_assert_insertions_for): Handle insertions at the
39616 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39618 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39619 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39621 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39624 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39627 2017-01-17 Alan Modra <amodra@gmail.com>
39630 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39631 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39632 symbolic stack limit when pic.
39634 2017-01-16 Martin Sebor <msebor@redhat.com>
39636 PR tree-optimization/78608
39637 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39639 2017-01-16 Jeff Law <law@redhat.com>
39642 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39643 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39644 for several include directories that may be relative to sysroot.
39645 * config/i386/x-mingw32 (gplus_includedir): Define.
39646 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39647 (native_system_includedir): Likewise.
39648 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39649 override if TARGET_SYSTEM_ROOT is defined.
39650 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39652 PR tree-optimization/79090
39653 PR tree-optimization/33562
39654 PR tree-optimization/61912
39655 PR tree-optimization/77485
39656 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39657 and computed trims into the dump file.
39659 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39661 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39663 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39666 * gimplify.c (gimplify_init_constructor): If want_value and
39667 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39671 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39672 sequence. Formatting fixes.
39673 (doloop_optimize): Formatting fixes.
39676 * gcc.c (debug_level_greater_than_spec_func): New function.
39677 (static_spec_functions): Add debug-level-gt spec function.
39678 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39680 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39681 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39682 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39683 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39684 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39685 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39687 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39689 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39690 QImode fixups to general and mask registers only.
39692 2017-01-16 Carl Love <cel@us.ibm.com>
39694 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39695 for built-in functions
39696 vector signed char vec_nabs (vector signed char)
39697 vector signed short vec_nabs (vector signed short)
39698 vector signed int vec_nabs (vector signed int)
39699 vector signed long long vec_nabs (vector signed long long)
39700 vector float vec_nabs (vector float)
39701 vector double vec_nabs (vector double)
39702 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39704 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39705 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39706 * doc/extend.texi: Update the documentation file for the new built-in
39709 2017-01-16 Martin Sebor <msebor@redhat.com>
39711 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39714 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39716 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39717 UNSPEC_VSX__XXSPLTD to require special splat handling.
39719 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39722 * system.h: Poison strndup.
39724 2017-01-16 Alan Modra <amodra@gmail.com>
39727 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39730 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39732 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39734 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39736 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39737 call recog here. Assert that INSN_CODE (insn) is non-negative.
39739 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39742 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39744 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39745 in the currently scheduled RTL fragment.
39747 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39749 PR rtl-optimization/78751
39750 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39753 2017-01-14 Jeff Law <law@redhat.com>
39755 PR tree-optimization/79090
39756 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39757 variable length stores.
39758 (compute_trims): Delete dead assignment to *trim_tail.
39759 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39762 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39764 PR rtl-optimization/78626
39765 PR rtl-optimization/78727
39766 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39767 of a block, and split such blocks after everything else is finished.
39769 2017-01-14 Alan Modra <amodra@gmail.com>
39772 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39773 target legitimate_combined_insn.
39774 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39775 (rs6000_legitimate_combined_insn): New function.
39776 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39778 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39779 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39780 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39782 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39784 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39786 2017-01-13 Jeff Law <law@redhat.com>
39788 PR tree-optimization/33562
39789 PR tree-optimization/61912
39790 PR tree-optimization/77485
39791 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39793 (delete_dead_assignment): Likewise.
39794 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
39795 statement to delete_dead_call and delete_dead_assignment.
39797 2017-01-13 David Malcolm <dmalcolm@redhat.com>
39800 * substring-locations.c (format_warning_va): Strengthen case 1 so
39801 that both endpoints of the substring must be within the format
39802 range for just the substring to be printed.
39804 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
39806 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
39807 * config/i386/i386.c (ix86_target_string): Add missing options
39808 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
39809 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
39810 flags_other and ix86_target_other to flags2_other. Display unknown
39812 (ix86_valid_target_attribute_inner_p): Add missing options and
39813 reorder options by implied ISAs, as in ix86_target_string.
39815 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39817 * hash-table.h (hash_table::too_empty_p): New function.
39818 (hash_table::expand): Use it.
39819 (hash_table::traverse): Likewise.
39820 (hash_table::empty_slot): Use sizeof (value_type) instead of
39821 sizeof (PTR) to convert bytes to elements. Shrink the table
39822 if the current size is excessive for the current number of
39825 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39827 * ira-costs.c (record_reg_classes): Break from the inner loop
39828 early once alt_fail is known to be true. Update outer loop
39829 handling accordingly.
39831 2017-01-13 Jeff Law <law@redhat.com>
39833 * tree-ssa-dse.c (decrement_count): New function.
39834 (increment_start_addr, maybe_trim_memstar_call): Likewise.
39835 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
39836 when we know the partially dead statement is a mem* function.
39838 PR tree-optimization/61912
39839 PR tree-optimization/77485
39840 * tree-ssa-dse.c: Include expr.h.
39841 (maybe_trim_constructor_store): New function.
39842 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
39844 PR tree-optimization/33562
39845 PR tree-optimization/61912
39846 PR tree-optimization/77485
39847 * doc/invoke.texi: Document new dse-max-object-size param.
39848 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
39849 * tree-ssa-dse.c: Include params.h.
39850 (dse_store_status): New enum.
39851 (initialize_ao_ref_for_dse): New, partially extracted from
39853 (valid_ao_ref_for_dse, normalize_ref): New.
39854 (setup_live_bytes_from_ref, compute_trims): Likewise.
39855 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
39856 (maybe_trim_partially_dead_store): Likewise.
39857 (maybe_trim_complex_store): Likewise.
39858 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
39859 Track what bytes live from the original store. Return tri-state
39860 for dead, partially dead or live.
39861 (dse_dom_walker): Add constructor, destructor and new private members.
39862 (delete_dead_call, delete_dead_assignment): New extracted from
39864 (dse_optimize_stmt): Make a member of dse_dom_walker.
39865 Use initialize_ao_ref_for_dse.
39867 PR tree-optimization/33562
39868 PR tree-optimization/61912
39869 PR tree-optimization/77485
39870 * sbitmap.h (bitmap_count_bits): Prototype.
39871 (bitmap_clear_range, bitmap_set_range): Likewise.
39872 * sbitmap.c (bitmap_clear_range): New function.
39873 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
39875 2017-01-13 Martin Liska <mliska@suse.cz>
39878 * function.c (set_cfun): Add new argument force.
39879 * function.h (set_cfun): Likewise.
39880 * ipa-inline-transform.c (inline_call): Use the function when
39881 strict alising from is dropped for function we inline to.
39883 2017-01-13 Richard Biener <rguenther@suse.de>
39885 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
39886 for dumping GIMPLE INTEGER_CSTs.
39888 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
39890 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
39891 to 201112L since C++17.
39893 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
39896 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
39897 if -fsanitize=kernel-address is present.
39899 2017-01-13 Richard Biener <rguenther@suse.de>
39901 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
39902 as _Literal ( type ) number in case usual suffixes do not
39903 preserve all information.
39905 2017-01-13 Richard Biener <rguenther@suse.de>
39907 PR tree-optimization/77283
39908 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
39909 and ssa-iterators.h.
39910 (is_feasible_trace): Implement a cost model based on joiner
39913 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
39916 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
39917 char or short to __float128/_Float128 directly.
39919 2017-01-12 Martin Sebor <msebor@redhat.com>
39921 to -Wformat-overflow.
39922 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
39923 (min_bytes_remaining): Same.
39924 (get_string_length): Same.
39925 (format_string): Same.
39926 (format_directive): Same.
39928 (pass_sprintf_length::handle_gimple_call): Same.
39930 2017-01-12 Jakub Jelinek <jakub@redhat.com>
39932 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
39933 info.nowrite calls with no lhs that can't throw. Return bool
39934 whether gsi_remove has been called or not.
39935 (pass_sprintf_length::handle_gimple_call): Return bool whether
39936 try_substitute_return_value called gsi_remove. Formatting fix.
39937 (pass_sprintf_length::execute): Don't use gsi_remove if
39938 handle_gimple_call returned true.
39941 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
39942 be removed due to side-effects, don't remove following barrier nor
39943 turn the successor edge into fallthru edge.
39945 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39948 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
39949 element-reversing loads and stores as not swappable.
39951 2017-01-12 Nathan Sidwell <nathan@acm.org>
39952 Nicolai Stange <nicstange@gmail.com>
39954 * combine.c (try_combine): Don't ignore result of overlap checking
39955 loop. Combine overlap & asm check into single loop.
39957 2017-01-12 Richard Biener <rguenther@suse.de>
39959 * tree-pretty-print.c (dump_generic_node): Provide -gimple
39960 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
39962 2017-01-12 Richard Biener <rguenther@suse.de>
39964 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
39965 and TS_TARGET_OPTION directly derive from TS_BASE.
39966 * tree-core.h (tree_optimization_option): Derive from tree_base.
39967 (tree_target_option): Likewise.
39969 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
39971 * config/i386/i386.c (memory_address_length): Increase len
39972 only when rip_relative_addr_p returns false.
39974 2017-01-11 Julia Koval <julia.koval@intel.com>
39976 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
39977 (OPTION_MASK_ISA_SGX_SET): New.
39978 (ix86_handle_option): Handle OPT_msgx.
39979 * config.gcc: Added sgxintrin.h.
39980 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
39981 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
39982 * config/i386/i386.c (ix86_target_string): Add -msgx.
39984 (ix86_option_override_internal): Handle new options.
39985 (ix86_valid_target_attribute_inner_p): Add sgx.
39986 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
39987 * config/i386/i386.opt: Add msgx.
39988 * config/i386/sgxintrin.h: New file.
39989 * config/i386/x86intrin.h: Add sgxintrin.h.
39991 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39994 * fold-const.c (maybe_nonzero_address): Return 1 for function
39996 (tree_single_nonzero_warnv_p): Don't handle function local objects
40000 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40003 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40006 * opts.c: Include "spellcheck.h"
40007 (struct string_fragment): New struct.
40008 (struct edit_distance_traits<const string_fragment &>): New
40010 (get_closest_sanitizer_option): New function.
40011 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40013 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40015 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40017 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40018 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40019 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40020 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40021 for initial die_offset if dwarf_split_debug_info.
40022 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40023 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40025 (output_skeleton_debug_sections): Formatting fix. Use
40026 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40027 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40029 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40031 * config/arm/cortex-a53.md: Add bypasses for
40032 cortex_a53_r2f_cvt.
40033 (cortex_a53_r2f): Only use for transfers.
40034 (cortex_a53_f2r): Likewise.
40035 (cortex_a53_r2f_cvt): Add reservation for conversions.
40036 (cortex_a53_f2r_cvt): Likewise.
40038 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40040 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40041 to all inlined functions, change static to extern.
40043 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40046 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40048 (arm_assemble_integer): Likewise.
40050 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40052 * config.gcc: Use new awk script to check CPU, FPU and architecture
40053 parameters for --with-... options.
40054 * config/arm/parsecpu.awk: New file
40055 * config/arm/arm-cpus.in: New file.
40056 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40058 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40060 * config/arm/t-arm: Update dependency rules.
40061 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40062 of processing .def files.
40063 * config/arm/genopt.sh: Deleted.
40064 * config/arm/gentune.sh: Deleted.
40065 * config/arm/arm-cores.def: Deleted.
40066 * config/arm/arm-arches.def: Deleted.
40067 * config/arm/arm-fpus.def: Deleted.
40068 * config/arm/arm-tune.md: Regenerated.
40069 * config/arm/arm-tables.opt: Regenerated.
40070 * config/arm/arm-cpu.h: New generated file.
40071 * config/arm/arm-cpu-data.h: New generated file.
40072 * config/arm/arm-cpu-cdata.h: New generated file.
40074 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40077 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40079 (input_varpool_node): Unpack dynamically_initialized bit.
40081 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40083 PR rtl-optimization/79032
40084 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40085 the alignment of the adjusted memory reference against that of MODE,
40086 instead of the alignment of the original memory reference.
40088 2017-01-11 Martin Jambor <mjambor@suse.cz>
40090 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40092 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40093 decorated functions.
40095 2017-01-11 Richard Biener <rguenther@suse.de>
40097 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40098 set range/nonnull info for PHI results. Do not set it on
40099 stmts marked for removal.
40101 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40103 * expr.c (store_field): In the bitfield case, fetch the return value
40104 from the registers before applying a single big-endian adjustment.
40105 Always do a final load for a BLKmode value not larger than a word.
40107 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40110 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40111 that we correctly handle column numbers greater than
40112 LINE_MAP_MAX_COLUMN_NUMBER.
40114 2017-01-10 Martin Sebor <msebor@redhat.com>
40116 PR middle-end/78245
40117 * gimple-ssa-sprintf.c (get_destination_size): Call
40118 {init,fini}object_sizes.
40119 * tree-object-size.c (addr_object_size): Adjust.
40120 (pass_through_call): Adjust.
40121 (pass_object_sizes::execute): Adjust.
40122 * tree-object-size.h (fini_object_sizes): Declare.
40124 2017-01-10 Martin Sebor <msebor@redhat.com>
40126 PR tree-optimization/78775
40127 * builtins.c (get_size_range): Move...
40128 * calls.c: ...to here.
40129 (alloc_max_size): Accept zero argument.
40130 (operand_signed_p): Remove.
40131 (maybe_warn_alloc_args_overflow): Call get_size_range.
40132 * calls.h (get_size_range): Declare.
40134 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40136 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40137 from TI's devices.csv file as of September 2016.
40138 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40140 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40142 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40143 * doc/invoke.texi: Likewise.
40144 * doc/md.texi: Likewise.
40145 * doc/objc.texi: Likewise.
40147 2017-01-10 Joshua Conner <joshconner@google.com>
40149 * config/arm/fuchsia-elf.h: New file.
40150 * config/fuchsia.h: New file.
40151 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40152 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40154 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40156 2016-01-10 Richard Biener <rguenther@suse.de>
40158 PR tree-optimization/79034
40159 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40160 Propagate out degenerate PHIs in the joiner.
40162 2017-01-10 Martin Liska <mliska@suse.cz>
40164 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40165 (sort_congruence_classes_by_decl_uid): Likewise.
40166 (sort_congruence_class_groups_by_decl_uid): Likewise.
40167 (sem_item_optimizer::merge_classes): Sort class, groups in these
40168 classes and members in the groups by DECL_UID of declarations.
40169 This would make merge operations stable.
40171 2017-01-10 Martin Liska <mliska@suse.cz>
40173 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40174 usage of m_classes_vec.
40175 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40176 (sem_item_optimizer::get_group_by_hash): Likewise.
40177 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40178 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40179 (sem_item_optimizer::verify_classes): Likewise.
40180 (sem_item_optimizer::process_cong_reduction): Likewise.
40181 (sem_item_optimizer::dump_cong_classes): Likewise.
40182 (sem_item_optimizer::merge_classes): Likewise.
40183 * ipa-icf.h (congruence_class_hash): Rename from
40184 congruence_class_group_hash. Remove declaration of m_classes_vec.
40186 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40188 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40189 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40190 * config.gcc: Add avx512vpopcntdqintrin.h.
40191 * config/i386/avx512vpopcntdqintrin.h: New.
40192 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40193 * config/i386/i386-builtin-types.def: Add new types.
40194 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40195 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40196 __builtin_ia32_vpopcountq_v8di_mask): New.
40197 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40198 __AVX512VPOPCNTDQ__.
40199 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40200 (PTA_AVX512VPOPCNTDQ): Define.
40201 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40202 TARGET_AVX512VPOPCNTDQ_P): Define.
40203 * config/i386/i386.opt: Add mavx512vpopcntdq.
40204 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40205 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40207 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40209 PR middle-end/77484
40210 * predict.def (PRED_CALL): Set to 67.
40212 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40214 * expr.c (store_field): In the bitfield case, if the value comes from
40215 a function call and is of an aggregate type returned in registers, do
40216 not modify the field mode; extract the value in all cases if the mode
40217 is BLKmode and the size is not larger than a word.
40219 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40222 * config/i386/cpuid.h: Fix undefined behavior.
40224 2017-01-04 Jeff Law <law@redhat.com>
40226 PR tree-optimization/79007
40227 PR tree-optimization/67955
40228 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40229 conservative for pt.null when flag_non_call_exceptions is on.
40231 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40233 PR translation/79019
40234 PR translation/79020
40235 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40236 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40237 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40239 * config/avr/avr.opt (maccumulate-args): Likewise.
40240 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40241 * common.opt (freport-bug): Likewise.
40242 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40243 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40244 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40245 translatable string.
40246 * config/i386/i386.c (function_value_32): Likewise.
40247 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40248 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40250 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40251 * common/config/msp430/msp430-common.c (msp430_handle_option):
40253 * symtab.c (symtab_node::verify_base): Likewise.
40254 * opts.c (set_debug_level): Likewise.
40255 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40256 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40257 missing whitespace to translatable strings.
40258 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40259 * config/sh/superh.h: Likewise.
40260 * config/i386/xopintrin.h: Likewise.
40261 * config/i386/znver1.md: Likewise.
40262 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40263 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40264 * double-int.h (struct double_int): Likewise.
40265 * double-int.c (div_and_round_double): Likewise.
40266 * wide-int.cc: Likewise.
40267 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40268 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40269 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40270 (crossjumps_occurred): ... this.
40271 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40274 PR tree-optimization/78899
40275 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40276 returning bool return struct loop *, NULL for failure and the new
40278 (versionable_outer_loop_p): Don't version outer loop if it has
40279 dont_vectorized bit set.
40280 (tree_if_conversion): When versioning outer loop, ensure
40281 tree_if_conversion is performed also on the inner loop of the
40282 non-vectorizable outer loop copy.
40283 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40284 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40285 prevent vectorization of it.
40286 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40287 the outer loop vectorization of the non-scalar version is attempted
40288 before vectorization of the inner loop in scalar version. If
40289 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40290 vectorization of its inner loop.
40291 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40292 has 2 inner loops, rename also on edges from bb whose single pred
40293 is outer_loop->header. Fix typo in function comment.
40295 2017-01-09 Martin Sebor <msebor@redhat.com>
40298 * asan.c (asan_emit_stack_protection): Increase local buffer size
40299 to avoid snprintf truncation warning.
40301 2017-01-09 Andrew Pinski <apinski@cavium.com>
40303 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40304 to reference thunderx2t99 for the tuning structure
40305 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40307 (thunderx2t99_extra_costs): This.
40308 * config/aarch64/aarch64-tune.md: Regenerate.
40309 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40310 (vulcan_addrcost_table): This.
40311 (vulcan_regmove_cost): Rename to ...
40312 (thunderx2t99_regmove_cost): This.
40313 (vulcan_vector_cost): Rename to ...
40314 (thunderx2t99_vector_cost): this.
40315 (vulcan_branch_cost): Rename to ...
40316 (thunderx2t99_branch_cost): This.
40317 (vulcan_tunings): Rename to ...
40318 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40319 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40321 2017-01-09 Martin Jambor <mjambor@suse.cz>
40325 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40326 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40327 (propagate_vr_accross_jump_function): Use the above function for all
40328 value range computations for pass-through jump functions and type
40329 converasion from explicit value range values.
40330 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40331 parameters from TYPE_ARG_TYPES.
40332 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40333 (ipa_write_node_info): Stream type of the actual argument.
40334 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40336 2017-01-09 Martin Liska <mliska@suse.cz>
40339 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40340 (lookup_compiler): Do not show error message with have_E.
40342 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40344 PR tree-optimization/78938
40345 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40346 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40347 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40348 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40351 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40353 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40356 2017-01-09 Richard Biener <rguenther@suse.de>
40358 PR tree-optimization/78997
40359 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40360 name condition properly.
40362 2017-01-09 Richard Biener <rguenther@suse.de>
40365 * dwarf2out.c (is_cxx): New overload with context.
40366 (is_naming_typedef_decl): Use it.
40368 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40370 * invoke.texi (Option Summary): Correct spacing in option lists
40371 and add line breaks to fix over-long lines.
40373 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40375 PR middle-end/17660
40377 * extend.texi (Common Variable Attributes): Add xref to GCC
40378 Internals manual to explain mode attribute keywords.
40380 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40383 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40384 and Preprocessor Options.
40385 (Options for Linking): Document -pthread here....
40386 (RS/6000 and PowerPC Options): ...not here.
40387 (Solaris 2 Options): ...or here.
40388 * doc/cppopts.texi: Document -pthread.
40390 2017-01-08 Martin Sebor <msebor@redhat.com>
40392 PR middle-end/77708
40393 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40394 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40395 New member functions.
40396 (format_directive): Used them.
40398 (pass_sprintf_length::handle_gimple_call): Same.
40399 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40400 to avoid truncation for any argument.
40401 (extract_affine_mul): Same.
40402 * tree.c (get_file_function_name): Same.
40404 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40406 PR middle-end/77484
40407 * predict.def (PRED_INDIR_CALL): Set to 86.
40409 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40411 PR preprocessor/54124
40412 * doc/cppopts.texi: Reformat -d subtable to list the full name
40413 of the options. Add cross-reference to the docs for the general
40414 compiler -d options.
40415 * doc/invoke.texi (Developer Options): Add cross-reference to the
40416 preprocessor-specific -d option documentation.
40418 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40420 PR preprocessor/13498
40421 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40422 redudant material, and reflect new command-line options.
40423 (System Headers): Likewise.
40425 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40427 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40428 -isystem, and -idirafter. Copy-edit.
40429 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40430 default for -ftrack-macro-expansion. Delete obsolete and
40431 badly-formatted implementation details about -fdebug-cpp output.
40432 * doc/cppwarnopts.texi: Copy-edit.
40434 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40437 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40438 that the transition from a max line width >= 1<<10 to narrower
40439 lines works correctly.
40441 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40443 * doc/options.texi (PerFunction): New.
40444 * opt-functions.awk (switch_flags): Map both Optimization and
40445 PerFunction to CL_OPTIMIZATION.
40446 * opth-gen.awk: Test for PerFunction flag along with
40448 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40449 it only when the latter is present. Skip those that don't in
40450 the hash function generator.
40451 * common.opt (fvar-tracking): Mark as PerFunction instead of
40453 (fvar-tracking-assignments): Likewise.
40454 (fvar-tracking-assignments-toggle): Likewise.
40455 (fvar-tracking-uninit): Likewise.
40457 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40459 PR translation/79018
40460 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40463 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40466 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40467 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40468 TARGET_LONG_JUMP_TABLE_OFFSETS.
40469 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40470 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40471 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40472 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40473 * config/m68k/m68k.md (tablejump expander): Likewise.
40474 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40475 TARGET_LONG_JUMP_TABLE_OFFSETS.
40476 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40477 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40479 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40480 David Holsgrove <david.holsgrove@xilinx.com>
40482 * common/config/microblaze/microblaze-common.c
40483 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40484 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40486 * config/microblaze/microblaze.c (microblaze_must_save_register)
40487 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40489 (microblaze_eh_return): New function.
40490 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40491 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40492 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40493 * config/microblaze/microblaze.md (eh_return): New pattern.
40495 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40497 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40498 GCC_DIAGNOSTIC_STRINGIFY): Define.
40500 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40502 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40504 * config/arm/arm.md (<mcrr>): New.
40506 * config/arm/arm.c (arm_arch5te): New.
40507 (arm_option_override): Set arm_arch5te.
40508 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40510 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40511 (arm_mcrr_qualifiers): ... this. New.
40512 (MRRC_QUALIFIERS): Define to...
40513 (arm_mrrc_qualifiers): ... this. New.
40514 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40516 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40517 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40518 (MRRCI, mrrc, MRRC): New.
40519 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40520 VUNSPEC_MRRC2): New.
40522 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40524 * config/arm/arm.md (<mcr>): New.
40526 * config/arm/arm.c (arm_coproc_builtin_available): Add
40527 support for mcr, mrc, mcr2 and mrc2.
40528 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40529 (arm_mcr_qualifiers): ... this. New.
40530 (MRC_QUALIFIERS): Define to ...
40531 (arm_mrc_qualifiers): ... this. New.
40532 (MCR_QUALIFIERS): Define to ...
40533 (arm_mcr_qualifiers): ... this. New.
40534 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40536 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40537 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40538 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40539 VUNSPEC_MRC2): New.
40541 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40543 * config/arm/arm.md (*ldc): New.
40547 * config/arm/arm.c (arm_coproc_builtin_available): Add
40548 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40549 (arm_coproc_ldc_stc_legitimate_address): New.
40550 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40551 'qualifier_const_pointer'.
40552 (LDC_QUALIFIERS): Define to...
40553 (arm_ldc_qualifiers): ... this. New.
40554 (STC_QUALIFIERS): Define to...
40555 (arm_stc_qualifiers): ... this. New.
40556 * config/arm/arm-protos.h
40557 (arm_coproc_ldc_stc_legitimate_address): New.
40558 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40559 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40560 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40561 stc2, stcl, stc2l): New.
40562 * config/arm/constraints.md (Uz): New.
40563 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40564 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40565 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40566 VUNSPEC_STC2L): New.
40568 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40570 * config/arm/arm.md (<cdp>): New.
40571 * config/arm/arm.c (neon_const_bounds): Rename this ...
40572 (arm_const_bounds): ... this.
40573 (arm_coproc_builtin_available): New.
40574 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40575 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40576 (CDP_QUALIFIERS): Define to...
40577 (arm_cdp_qualifiers): ... this. New.
40579 (arm_expand_builtin_args): Add case for 6 arguments.
40580 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40581 (arm_const_bounds): ... this.
40582 (arm_coproc_builtin_available): New.
40583 * config/arm/arm_acle.h (__arm_cdp): New.
40585 * config/arm/arm_acle_builtins.def (cdp): New.
40587 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40588 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40589 'arm_const_bounds'.
40590 * config/arm/types.md (coproc): New.
40591 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40592 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40593 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40594 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40596 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40598 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40599 (UBINOP_QUALIFIERS): New.
40601 (acle_builtin_data): New. Change comment.
40602 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40603 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40604 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40605 arm_acle_builtins.def.
40606 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40607 (arm_init_acle_builtins): New.
40608 (CRC32_BUILTIN): Remove.
40609 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40610 crc32cb, crc32ch and crc32cw.
40611 (arm_init_crc32_builtins): Remove.
40612 (arm_init_builtins): Use arm_init_acle_builtins rather
40613 than arm_init_crc32_builtins.
40614 (arm_expand_acle_builtin): New.
40615 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40616 * config/arm/arm_acle_builtins.def: New.
40618 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40620 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40621 (arm_builtin_datum): ... this.
40622 (arm_init_neon_builtin): Rename to ...
40623 (arm_init_builtin): ... this. Add a new parameters PREFIX
40624 and USE_SIG_IN_NAME.
40625 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40626 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40627 'arm_builtin_datum'.
40628 (arm_init_vfp_builtins): Likewise.
40629 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40630 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40631 (arm_expand_neon_args): Rename to ...
40632 (arm_expand_builtin_args): ... this. Rename builtin_arg
40633 enum values and differentiate between ARG_BUILTIN_MEMORY
40634 and ARG_BUILTIN_NEON_MEMORY.
40635 (arm_expand_neon_builtin_1): Rename to ...
40636 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40637 values, arm_expand_builtin_args and add bool parameter NEON.
40638 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40639 (arm_expand_vfp_builtin): Likewise.
40640 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40642 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40644 PR middle-end/77484
40645 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40646 * predict.c (tree_estimate_probability_bb): Reverse direction of
40647 polymorphic call predictor.
40649 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40651 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40652 (determine_pass_name_match): ...this new function and...
40653 (should_skip_pass_p): ...this new function.
40655 2017-01-06 Nathan Sidwell <nathan@acm.org>
40657 * ipa-visibility.c (function_and_variable_visibility): Reformat
40658 comments and long lines. Remove extrneous if.
40659 * symtab.c (symtab_node::make_decl_local): Fix code format.
40660 (symtab_node::set_section_for_node): Fix comment typo.
40662 2017-01-06 Martin Liska <mliska@suse.cz>
40665 * lra-constraints.c: Rename invariant to lra_invariant.
40666 * predict.c (set_even_probabilities): Initialize e to NULL.
40668 2017-01-05 Martin Sebor <msebor@redhat.com>
40670 PR tree-optimization/78910
40671 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40672 (format_integer): Correct off-by-one error in the handling
40673 of precision with negative numbers in signed conversions..
40675 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40677 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40679 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40681 PR tree-optimization/71016
40682 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40683 factor_out_conditional_conversion. Formatting fix.
40684 (factor_out_conditional_conversion): Add cond_stmt argument.
40685 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40686 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40689 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40691 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40692 read-rtl-function.o, and selftest-rtl.o.
40693 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40694 (selftest::aarch64_test_loading_full_dump): New function.
40695 (selftest::aarch64_run_selftests): New function.
40696 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40697 selftest::aarch64_run_selftests.
40698 * config/i386/i386.c
40699 (selftest::ix86_test_loading_dump_fragment_1): New function.
40700 (selftest::ix86_test_loading_call_insn): New function.
40701 (selftest::ix86_test_loading_full_dump): New function.
40702 (selftest::ix86_test_loading_unspec): New function.
40703 (selftest::ix86_run_selftests): Call the new functions.
40704 * emit-rtl.c (maybe_set_max_label_num): New function.
40705 * emit-rtl.h (maybe_set_max_label_num): New decl.
40706 * function.c (instantiate_decls): Guard call to
40707 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40708 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40710 * gensupport.c (gen_reader::gen_reader): Pass "false"
40711 for new "compact" param of rtx_reader.
40712 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40713 rather than an empty string for NULL strings.
40714 * read-md.c: Potentially include config.h rather than bconfig.h.
40715 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40716 (have_error): New global, copied from errors.c.
40717 (md_reader::read_name): Rename to...
40718 (md_reader::read_name_1): ...this, adding "out_loc" param,
40719 and converting "missing name or number" to returning false, rather
40721 (md_reader::read_name): Reimplement in terms of read_name_1.
40722 (md_reader::read_name_or_nil): New function.
40723 (md_reader::read_string): Handle "(nil)" by returning NULL.
40724 (md_reader::md_reader): Add new param "compact".
40725 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40726 (md_reader::read_file): New method.
40727 * read-md.h (md_reader::md_reader): Add new param "compact".
40728 (md_reader::read_file): New method.
40729 (md_reader::is_compact): New accessor.
40730 (md_reader::read_name): Convert return type from void to file_location.
40731 (md_reader::read_name_or_nil): New decl.
40732 (md_reader::read_name_1): New decl.
40733 (md_reader::m_compact): New field.
40734 (noop_reader::noop_reader): Pass "false" for new "compact" param
40736 (rtx_reader::rtx_reader): Add new "compact" param.
40737 (rtx_reader::read_rtx_operand): Make virtual and convert return
40738 type from void to rtx.
40739 (rtx_reader::read_until): New decl.
40740 (rtx_reader::handle_any_trailing_information): New virtual function.
40741 (rtx_reader::postprocess): New virtual function.
40742 (rtx_reader::finalize_string): New virtual function.
40743 (rtx_reader::m_in_call_function_usage): New field.
40744 (rtx_reader::m_reuse_rtx_by_id): New field.
40745 * read-rtl-function.c: New file.
40746 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40747 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40748 (selftest::verify_three_block_rtl_cfg): New decl.
40749 * read-rtl-function.h: New file.
40750 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40751 For host, include function.h, memmodel.h, and emit-rtl.h.
40752 (one_time_initialization): New function.
40753 (struct compact_insn_name): New struct.
40754 (compact_insn_names): New array.
40755 (find_code): Handle insn codes in compact dumps.
40756 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40757 (bind_subst_iter_and_attr): Likewise.
40758 (add_condition_to_string): Likewise.
40759 (add_condition_to_rtx): Likewise.
40760 (apply_attribute_uses): Likewise.
40761 (add_current_iterators): Likewise.
40762 (apply_iterators): Likewise.
40763 (initialize_iterators): Guard usage of apply_subst_iterator with
40764 #ifdef GENERATOR_FILE.
40765 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40766 (md_reader::read_mapping): Likewise.
40767 (add_define_attr_for_define_subst): Likewise.
40768 (add_define_subst_attr): Likewise.
40769 (read_subst_mapping): Likewise.
40770 (check_code_iterator): Likewise.
40771 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40773 (one_time_initialization): New function.
40774 (rtx_reader::read_until): New method.
40775 (read_flags): New function.
40776 (parse_reg_note_name): New function.
40777 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40778 Handle reuse_rtx ids.
40779 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40780 Add parsing support for RTL dumps, mirroring the special-cases in
40781 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40782 values, and calling handle_any_trailing_information.
40783 (rtx_reader::read_rtx_operand): Convert return type from void
40784 to rtx, returning return_rtx. Handle case 'e'. Call
40785 finalize_string on XSTR and XTMPL fields.
40786 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40787 "(nil)" values were omitted. Call the postprocess vfunc on the
40789 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40790 class ctor. Initialize m_in_call_function_usage. Call
40791 one_time_initialization.
40792 * rtl-tests.c (selftest::test_uncond_jump): Call
40793 set_new_first_and_last_insn.
40794 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
40795 * selftest-rtl.c: New file.
40796 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
40797 (selftest::get_insn_by_uid): New decl.
40798 * selftest-run-tests.c (selftest::run_tests): Call
40799 read_rtl_function_c_tests.
40800 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
40801 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
40804 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
40806 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
40807 operands in a special way. Assert that pos+len <= mode precision.
40809 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40811 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
40812 3 argument Alias with unlimited for the negative form.
40813 (fno-vect-cost-model): Removed.
40815 2017-01-05 Martin Liska <mliska@suse.cz>
40817 * hsa-gen.c (gen_hsa_divmod): New function.
40818 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
40820 2017-01-05 Martin Liska <mliska@suse.cz>
40823 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
40826 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40828 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
40829 small constant length operands.
40831 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40833 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
40834 between loop iterations.
40836 2017-01-05 Martin Liska <mliska@suse.cz>
40839 * gimplify.c (gimplify_decl_expr): Compare to
40840 asan_poisoned_variables instread of checking flags.
40841 (gimplify_target_expr): Likewise.
40842 (gimplify_expr): Likewise.
40843 (gimplify_function_tree): Conditionally initialize
40844 asan_poisoned_variables.
40846 2017-01-04 Jeff Law <law@redhat.com>
40848 PR tree-optimizatin/78812
40849 * rtl.h (contains_mem_rtx_p): Prototype.
40850 * ifcvt.c (containts_mem_rtx_p): Move from here to...
40851 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
40852 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
40853 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
40854 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
40856 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40858 * input.c (assert_char_at_range): Default-initialize actual_range.
40860 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40862 * df-scan.c (df_ref_create_structure): Make regno unsigned,
40863 to match the caller.
40865 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40867 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
40868 insns after final jump in test to emit dummy move.
40870 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40872 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
40873 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
40875 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40877 * multiple_target.c (create_dispatcher_calls): Init e_next.
40878 * tree-ssa-loop-split.c (split_loop): Init border.
40879 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
40882 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
40887 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
40888 (altivec_register_operand): Do not return true if the operand
40889 contains a SUBREG mixing SImode and SFmode.
40890 (vsx_register_operand): Likewise.
40891 (vsx_reg_sfsubreg_ok): New predicate.
40892 (vfloat_operand): Do not return true if the operand contains a
40893 SUBREG mixing SImode and SFmode.
40894 (vint_operand): Likewise.
40895 (vlogical_operand): Likewise.
40896 (gpc_reg_operand): Likewise.
40897 (int_reg_operand): Likewise.
40898 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
40899 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
40900 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
40902 (rs6000_emit_move_si_sf_subreg): New helper function.
40903 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
40904 fixup SUBREGs involving SImode and SFmode.
40905 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
40906 numbers for the new peephole2 optimization.
40907 (peephole2 for SFmode unions): New peephole2 to optimize cases in
40908 the GLIBC math library that do AND/IOR/XOR operations on single
40909 precision floating point.
40910 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
40911 target macros to say whether we need to avoid SUBREGs mixing
40913 (TARGET_ALLOW_SF_SUBREG): Likewise.
40914 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
40915 (UNSPEC_SI_FROM_SF): Likewise.
40916 (iorxor): Change spacing.
40917 (and_ior_xor): New iterator for AND, IOR, and XOR.
40918 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
40919 (movdi_from_sf_zero_ext): Likewise.
40920 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
40921 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
40922 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
40923 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
40924 (fms<mode>4): Likewise.
40925 (fnma<mode>4): Likewise.
40926 (fnms<mode>4): Likewise.
40927 (nfma<mode>4): Likewise.
40928 (nfms<mode>4): Likewise.
40930 2017-01-04 Marek Polacek <polacek@redhat.com>
40933 * doc/invoke.texi: Document -Wpointer-compare.
40935 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40937 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
40940 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
40941 descriptions for -gdwarf-5 and emit them as uleb128 instead of
40944 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
40947 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
40948 documentation of the powerpc_popcntb_ok attribute.
40949 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
40950 code to issue warning messages if a requested CPU configuration is
40951 not supported by the binary (assembler and loader) toolchain.
40952 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
40953 made to define a built-in function that has been disabled.
40954 (paired_init_builtins): Add assertion to prevent ICE if attempt is
40955 made to define a built-in function that has been disabled.
40956 (altivec_init_builtins): Add comment explaining why definition
40957 of the DST built-in functions is not preceded by an assertion
40958 check. Add assertions to prevent ICE if attempts are made to
40959 define an altivec predicate or an abs* built-in function that has
40961 (htm_init_builtins): Add comment explaining why definition of the
40962 htm built-in functions is not preceded by an assertion check.
40964 2017-01-04 Jeff Law <law@redhat.com>
40966 PR tree-optimizatin/67955
40967 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
40968 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
40969 the points-to solution does not include pt_null. Use DECL_PT_UID
40972 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
40974 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
40975 Use gen_int_mode instead of gen_lopwart for const_int operands.
40977 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40979 PR tree-optimization/71563
40980 * match.pd: Simplify X << Y into X if Y is known to be 0 or
40981 out of range value - has low bits known to be zero.
40983 2017-01-04 Alan Modra <amodra@gmail.com>
40985 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
40986 * configure: Regenerate.
40987 * config.in: Regenerate.
40989 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40992 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
40993 a substring of the message, but strcmp with the whole message. Ifdef
40994 ENABLE_NLS, translate the message first using dgettext.
40996 2017-01-03 Jeff Law <law@redhat.com>
40998 PR tree-optimizatin/78856
40999 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41000 (mark_threaded_blocks): Remove code to truncate thread paths that
41001 cross multiple loop headers. Instead invalidate the cached loop
41002 iteration information and handle case of a thread path walking
41003 into an irreducible region.
41005 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41008 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41009 assertions. Add support for doing the signbit if the IEEE 128-bit
41010 floating point value is in a GPR.
41011 * config/rs6000/rs6000.md (Fsignbit): Delete.
41012 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41013 Update the length attribute if the value is in a GPR.
41014 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41015 the sign or zero extension instruction, since the value is always 0/1.
41016 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41019 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41020 extracting SImode to a GPR register so that we can generate a
41021 store, limit the vector to be in a traditional Altivec register
41022 for the vextuwrx instruction.
41024 2017-01-03 Ian Lance Taylor <iant@google.com>
41026 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41028 2017-01-03 Martin Sebor <msebor@redhat.com>
41030 PR tree-optimization/78696
41031 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41032 precision. Use MPFR for %f for greater fidelity. Correct handling
41034 (pass_sprintf_length::compute_format_length): Set width and precision
41035 specified by asrerisk to void_node for vararg functions.
41036 (try_substitute_return_value): Adjust dump output.
41038 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41040 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41042 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41044 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41045 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41046 -mlra/-mno-lra was passed to the compiler.
41048 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41050 PR rtl-optimization/65618
41051 * emit-rtl.c (try_split): Move initialization of "before" and
41052 "after" to just before the call to emit_insn_after_setloc.
41054 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41056 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41058 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41060 * dwarf2out.c (gen_enumeration_type_die): When
41061 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41063 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41065 PR tree-optimization/78965
41066 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41067 Change first argument from const call_info & to call_info &. For %n
41068 set info.nowrite to false.
41070 PR middle-end/78901
41071 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41072 possibly throwing calls.
41074 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41075 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41076 and fns handling, rather than in a separate case SSA_NAME.
41078 2017-01-02 Jeff Law <law@redhat.com>
41080 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41081 fixes for first_period and second_period variables.
41083 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41086 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41087 (*insvqi_1): New insn pattern.
41088 (*insvqi_1_mem_rex64): Ditto.
41089 (*insvqi_2): Ditto.
41090 (*insvqi_3): Rename from *insvqi.
41092 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41094 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41096 * doc/cfg.texi (Edges): Remove reference to Java.
41097 (Maintaining the CFG): Ditto.
41099 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41101 PR middle-end/77674
41102 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41103 transparent aliases.
41105 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41107 PR middle-end/77484
41108 * predict.def (PRED_CALL): Update hitrate.
41109 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41110 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41111 into direct/indirect/polymorphic variants.
41113 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41115 Update copyright years.
41117 * gcc.c (process_command): Update copyright notice dates.
41118 * gcov-dump.c (print_version): Ditto.
41119 * gcov.c (print_version): Ditto.
41120 * gcov-tool.c (print_version): Ditto.
41121 * gengtype.c (create_file): Ditto.
41122 * doc/cpp.texi: Bump @copying's copyright year.
41123 * doc/cppinternals.texi: Ditto.
41124 * doc/gcc.texi: Ditto.
41125 * doc/gccint.texi: Ditto.
41126 * doc/gcov.texi: Ditto.
41127 * doc/install.texi: Ditto.
41128 * doc/invoke.texi: Ditto.
41130 Copyright (C) 2017 Free Software Foundation, Inc.
41132 Copying and distribution of this file, with or without modification,
41133 are permitted in any medium without royalty provided the copyright
41134 notice and this notice are preserved.