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 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
6 rather than a HOST_WIDE_INT.
7 * tree-sra.c (build_ref_for_offset): Likewise.
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 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
15 * tree.c (mem_ref_offset): Likewise.
16 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
17 * builtins.c (get_object_alignment_2): Likewise.
18 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
19 * gimple-fold.c (get_base_constructor): Likewise.
20 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
21 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
23 * ipa-polymorphic-call.c
24 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
25 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
26 (get_ancestor_addr_info): Likewise.
27 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
29 * tree-data-ref.c (dr_analyze_innermost): Likewise.
30 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
31 * tree-eh.c (tree_could_trap_p): Likewise.
32 * tree-object-size.c (addr_object_size): Likewise.
33 * tree-ssa-address.c (copy_ref_info): Likewise.
34 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
35 (indirect_refs_may_alias_p): Likewise.
36 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
37 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
38 (non_rewritable_mem_ref_base): Likewise.
39 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
40 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
41 * varasm.c (decode_addr_const): Likewise.
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 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
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 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
55 as a poly_uint64 rather than an unsigned int.
56 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
58 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
59 Alan Hayward <alan.hayward@arm.com>
60 David Sherwood <david.sherwood@arm.com>
62 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
64 (decode_addr_const): Update accordingly.
66 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
67 Alan Hayward <alan.hayward@arm.com>
68 David Sherwood <david.sherwood@arm.com>
70 * tree.h (bit_field_size, bit_field_offset): New functions.
71 * hsa-gen.c (gen_hsa_addr): Use them.
72 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
73 (simplify_vector_constructor): Likewise.
74 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
75 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
76 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
77 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
80 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
81 Alan Hayward <alan.hayward@arm.com>
82 David Sherwood <david.sherwood@arm.com>
84 * expr.h (emit_group_load, emit_group_load_into_temps)
85 (emit_group_store): Take the size as a poly_int64 rather than an int.
86 * expr.c (emit_group_load_1, emit_group_load): Likewise.
87 (emit_group_load_into_temp, emit_group_store): Likewise.
89 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
90 Alan Hayward <alan.hayward@arm.com>
91 David Sherwood <david.sherwood@arm.com>
93 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
94 an unsigned int to a poly_uint64.
95 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
96 sizes as poly_uint64s rather than unsigned ints.
97 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
100 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
101 Alan Hayward <alan.hayward@arm.com>
102 David Sherwood <david.sherwood@arm.com>
104 * lra-constraints.c (emit_inc): Change inc_amount from an int
107 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
108 Alan Hayward <alan.hayward@arm.com>
109 David Sherwood <david.sherwood@arm.com>
111 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
113 (add_stack_var, stack_var_cmp, partition_stack_vars)
114 (dump_stack_var_partition): Update accordingly.
115 (alloc_stack_frame_space): Take the size as a poly_int64 rather
116 than a HOST_WIDE_INT.
117 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
118 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
119 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
120 rather than a HOST_WIDE_INT.
122 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
123 Alan Hayward <alan.hayward@arm.com>
124 David Sherwood <david.sherwood@arm.com>
126 * target.def (return_pops_args): Treat both the input and output
127 sizes as poly_int64s rather than HOST_WIDE_INTS.
128 * targhooks.h (default_return_pops_args): Update accordingly.
129 * targhooks.c (default_return_pops_args): Likewise.
130 * doc/tm.texi: Regenerate.
131 * emit-rtl.h (incoming_args): Change pops_args, size and
132 outgoing_args_size from int to poly_int64_pod.
133 * function.h (expr_status): Change x_pending_stack_adjust and
134 x_stack_pointer_delta from int to poly_int64.
135 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
136 (ARGS_SIZE_RTX): Update accordingly.
137 * calls.c (highest_outgoing_arg_in_use): Change from int to
139 (stack_usage_watermark, stored_args_watermark): New variables.
140 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
141 (emit_call_1): Change the stack_size and rounded_stack_size
142 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
144 (save_fixed_argument_area): Check stack_usage_watermark.
145 (initialize_argument_information): Change old_pending_adj from
146 a HOST_WIDE_INT * to a poly_int64_pod *.
147 (compute_argument_block_size): Return the size as a poly_int64
149 (finalize_must_preallocate): Track polynomial argument sizes.
150 (compute_argument_addresses): Likewise.
151 (internal_arg_pointer_based_exp): Track polynomial offsets.
152 (mem_overlaps_already_clobbered_arg_p): Rename to...
153 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
154 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
155 Check stored_args_used_watermark.
156 (load_register_parameters): Update accordingly.
157 (check_sibcall_argument_overlap_1): Likewise.
158 (combine_pending_stack_adjustment_and_call): Take the unadjusted
159 args size as a poly_int64 rather than an int. Return a bool
160 indicating whether the optimization was possible and return
161 the new adjustment by reference.
162 (check_sibcall_argument_overlap): Track polynomail argument sizes.
163 Update stored_args_watermark.
164 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
165 (expand_call): Likewise. Maintain stack_usage_watermark and
166 stored_args_watermark. Update calls to
167 combine_pending_stack_adjustment_and_call.
168 (emit_library_call_value_1): Handle polynomial argument sizes.
169 Call stack_region_maybe_used_p and mark_stack_region_used.
170 Maintain stack_usage_watermark.
171 (store_one_arg): Likewise. Update call to
172 mem_overlaps_already_clobbered_arg_p.
173 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
175 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
176 * config/microblaze/microblaze.c (microblaze_function_prologue):
178 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
179 TARGET_RETURN_POPS_ARGS interface.
180 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
182 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
183 * config/i386/i386.c (ix86_return_pops_args): Update for new
184 TARGET_RETURN_POPS_ARGS interface.
185 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
186 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
187 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
188 TARGET_RETURN_POPS_ARGS interface.
189 * config/vax/vax.c (vax_return_pops_args): Likewise.
190 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
191 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
192 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
193 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
194 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
195 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
196 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
197 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
198 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
199 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
200 and x_stack_pointer_delta from int to poly_int64.
201 * dojump.c (do_pending_stack_adjust): Update accordingly.
202 * explow.c (allocate_dynamic_stack_space): Handle polynomial
203 stack_pointer_deltas.
204 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
205 (pad_to_arg_alignment): Track polynomial offsets.
206 (assign_parm_find_stack_rtl): Likewise.
207 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
208 * toplev.c (output_stack_usage): Update reference to
209 current_function_pushed_stack_size.
211 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
212 Alan Hayward <alan.hayward@arm.com>
213 David Sherwood <david.sherwood@arm.com>
215 * function.c (in_arg_offset, var_offset, dynamic_offset)
216 (out_arg_offset, cfa_offset): Change from int to poly_int64.
217 (instantiate_new_reg): Return the new offset as a poly_int64_pod
218 rather than a HOST_WIDE_INT.
219 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
220 (instantiate_virtual_regs_in_insn): Likewise.
222 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
223 Alan Hayward <alan.hayward@arm.com>
224 David Sherwood <david.sherwood@arm.com>
226 * rtl.h (get_args_size, add_args_size_note): New functions.
227 (find_args_size_adjust): Return a poly_int64 rather than a
229 (fixup_args_size_notes): Likewise. Make the same change to the
230 end_args_size parameter.
231 * rtlanal.c (get_args_size, add_args_size_note): New functions.
232 * builtins.c (expand_builtin_trap): Use add_args_size_note.
233 * calls.c (emit_call_1): Likewise.
234 * explow.c (adjust_stack_1): Likewise.
235 * cfgcleanup.c (old_insns_match_p): Update use of
236 find_args_size_adjust.
237 * combine.c (distribute_notes): Track polynomial arg sizes.
238 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
239 end_true_args_size, beg_delay_args_size and end_delay_args_size
240 from HOST_WIDE_INT to poly_int64.
241 (add_cfi_args_size): Take the args_size as a poly_int64 rather
242 than a HOST_WIDE_INT.
243 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
244 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
245 polynomial arg sizes.
246 * emit-rtl.c (try_split): Use get_args_size.
247 * recog.c (peep2_attempt): Likewise.
248 * reload1.c (reload_as_needed): Likewise.
249 * expr.c (find_args_size_adjust): Return the adjustment as a
250 poly_int64 rather than a HOST_WIDE_INT.
251 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
252 to a poly_int64 and change the return type in the same way.
253 (emit_single_push_insn): Track polynomial arg sizes.
255 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
256 Alan Hayward <alan.hayward@arm.com>
257 David Sherwood <david.sherwood@arm.com>
259 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
260 from HOST_WIDE_INT to poly_int64.
261 * expr.c (push_block, emit_push_insn): Likewise.
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 * function.h (frame_space): Change start and length from HOST_WIDE_INT
269 (get_frame_size): Return the size as a poly_int64 rather than a
271 (frame_offset_overflow): Take the offset as a poly_int64 rather
272 than a HOST_WIDE_INT.
273 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
274 (assign_stack_temp): Likewise for the size.
275 * function.c (get_frame_size): Return a poly_int64 rather than
277 (frame_offset_overflow): Take the offset as a poly_int64 rather
278 than a HOST_WIDE_INT.
279 (try_fit_stack_local): Take the start, length and size as poly_int64s
280 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
281 rather than a HOST_WIDE_INT.
282 (add_frame_space): Take the start and end as poly_int64s rather than
284 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
285 (assign_stack_temp): Likewise for the size.
286 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
288 (find_temp_slot_from_address): Handle polynomial offsets.
289 (combine_temp_slots): Likewise.
290 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
292 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
293 poly_int64 rather than a HOST_WIDE_INT.
294 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
295 than a HOST_WIDE_INT.
296 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
297 polynomial frame offsets.
298 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
299 as a poly_int64 rather than an int.
300 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
301 * config/v850/v850-protos.h (compute_frame_size): Likewise.
302 * config/v850/v850.c (compute_frame_size): Likewise.
303 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
304 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
305 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
306 * config/pa/pa.c (pa_compute_frame_size): Likewise.
307 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
308 rather than a HOST_WIDE_INT.
309 * explow.c (get_dynamic_stack_base): Likewise.
310 * final.c (final_start_function): Use the constant lower bound
311 of the frame size for -Wframe-larger-than.
312 * ira.c (do_reload): Adjust for new get_frame_size return type.
313 * lra.c (lra): Likewise.
314 * reload1.c (reload): Likewise.
315 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
316 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
317 * rtlanal.c (get_initial_register_offset): Return the offset as
318 a poly_int64 rather than a HOST_WIDE_INT.
320 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
321 Alan Hayward <alan.hayward@arm.com>
322 David Sherwood <david.sherwood@arm.com>
324 * reload1.c (elim_table): Change initial_offset, offset and
325 previous_offset from HOST_WIDE_INT to poly_int64_pod.
326 (offsets_at): Change the target array's element type from
327 HOST_WIDE_INT to poly_int64_pod.
328 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
329 (elimination_costs_in_insn, update_eliminable_offsets)
330 (verify_initial_elim_offsets, set_offsets_for_label)
331 (init_eliminable_invariants): Update after above changes.
333 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
334 Alan Hayward <alan.hayward@arm.com>
335 David Sherwood <david.sherwood@arm.com>
337 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
338 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
339 (decomposition): Change start and end from HOST_WIDE_INT
341 (decompose, immune_p): Update accordingly.
342 (find_inc_amount): Return a poly_int64 rather than an int.
343 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
346 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
347 Alan Hayward <alan.hayward@arm.com>
348 David Sherwood <david.sherwood@arm.com>
350 * tree.h (get_inner_reference): Return the bitsize and bitpos
351 as poly_int64_pods rather than HOST_WIDE_INT.
352 * fold-const.h (ptr_difference_const): Return the pointer difference
353 as a poly_int64_pod rather than a HOST_WIDE_INT.
354 * expr.c (get_inner_reference): Return the bitsize and bitpos
355 as poly_int64_pods rather than HOST_WIDE_INT.
356 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
358 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
359 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
360 (optimize_bit_field_compare): Update call to get_inner_reference.
361 (decode_field_reference): Likewise.
362 (fold_unary_loc): Track polynomial offsets and sizes.
363 (split_address_to_core_and_offset): Return the bitpos as a
364 poly_int64_pod rather than a HOST_WIDE_INT.
365 (ptr_difference_const): Likewise for the pointer difference.
366 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
367 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
368 * dbxout.c (dbxout_expand_expr): Likewise.
369 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
370 (loc_list_from_tree_1, fortran_common): Likewise.
371 * gimple-laddress.c (pass_laddress::execute): Likewise.
372 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
373 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
374 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
375 * tree-affine.c (tree_to_aff_combination): Likewise.
376 (get_inner_reference_aff): Likewise.
377 * tree-data-ref.c (split_constant_offset_1): Likewise.
378 (dr_analyze_innermost): Likewise.
379 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
380 * tree-sra.c (ipa_sra_check_caller): Likewise.
381 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
382 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
383 (instrument_bool_enum_load, instrument_object_size): Likewise.
384 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
385 to get_inner_reference.
386 * hsa-gen.c (gen_hsa_addr): Likewise.
387 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
388 * tsan.c (instrument_expr): Likewise.
389 * match.pd: Update call to ptr_difference_const.
391 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
392 Alan Hayward <alan.hayward@arm.com>
393 David Sherwood <david.sherwood@arm.com>
395 * fold-const.c (fold_comparison): Track sizes and offsets as
396 poly_int64s rather than HOST_WIDE_INTs when folding address
399 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
400 Alan Hayward <alan.hayward@arm.com>
401 David Sherwood <david.sherwood@arm.com>
403 * expr.h (get_bit_range): Return the bitstart and bitend as
404 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
405 as a poly_int64 rather than a HOST_WIDE_INT.
406 * expr.c (get_bit_range): Likewise.
407 (expand_assignment): Update call accordingly.
408 * fold-const.c (optimize_bit_field_compare): Likewise.
410 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
411 Alan Hayward <alan.hayward@arm.com>
412 David Sherwood <david.sherwood@arm.com>
414 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
415 (aarch64_legitimate_address_p): Use it instead of an rtx code,
416 as an optional final parameter.
417 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
418 (aarch64_legitimate_address_p): Likewise.
419 (aarch64_print_address_internal): Take an aarch64_addr_query_type
420 instead of an rtx code.
421 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
422 (aarch64_legitimate_address_hook_p): Likewise.
423 (aarch64_print_ldpstp_address): Likewise.
424 (aarch64_print_operand_address): Likewise.
425 (aarch64_address_cost): Likewise.
426 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
427 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
428 (aarch64_mem_pair_lanes_operand): Likewise.
430 2017-12-20 Richard Biener <rguenther@suse.de>
432 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
433 update_stmt_if_modified.
435 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
437 PR tree-optimization/83491
438 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
439 before walking uses. Improve coding style and comments.
441 2017-12-20 Tom de Vries <tom@codesourcery.com>
443 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
445 2017-12-20 Jakub Jelinek <jakub@redhat.com>
448 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
450 (pass_ipa_free_fn_summary): Add small_p private data member,
451 initialize to false in the ctor.
452 (pass_ipa_free_fn_summary::clone,
453 pass_ipa_free_fn_summary::set_pass_param,
454 pass_ipa_free_fn_summary::gate): New methods.
455 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
456 | TODO_dump_symtab if small_p.
457 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
458 entry and add another instance of the pass with false parm after
460 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
461 ipa_free_fn_summary here.
463 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
465 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
467 2017-12-20 Martin Sebor <msebor@redhat.com>
470 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
471 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
472 (expand_builtin_strncmp): Same.
474 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
477 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
478 allowed debug stmts before labels.
479 (expand_gimple_basic_block): Likewise.
480 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
481 * gimple-iterator.h (gsi_after_labels): Likewise.
482 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
483 rename reused variable, and simplify using gsi_move_before.
484 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
485 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
486 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
487 (gimple_verify_flow_info, gimple_block_label): Likewise.
488 (make_blocks): Move debug markers after adjacent labels.
489 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
490 allowed debug insns outside blocks.
491 * df-scan.c (df_insn_delete): Likewise.
492 * lra-constraints.c (update_ebb_live_info): Likewise.
493 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
494 (vt_initialize, delete_vta_debug_insns): Likewise.
495 (reemit_marker_as_note): Drop BB parm. Adjust callers.
497 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
498 Alan Hayward <alan.hayward@arm.com>
499 David Sherwood <david.sherwood@arm.com>
501 * poly-int-types.h (round_down_to_byte_boundary): New macro.
502 (round_up_to_byte_boundary): Likewise.
503 * expr.h (get_bit_range): Add temporary shim.
504 * gimple-ssa-store-merging.c (store_operand_info): Change the
505 bitsize, bitpos, bitregion_start and bitregion_end fields from
506 unsigned HOST_WIDE_INT to poly_uint64.
507 (merged_store_group): Likewise load_align_base.
508 (compatible_load_p, compatible_load_p): Update accordingly.
509 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
510 (split_group, imm_store_chain_info::output_merged_store): Likewise.
511 (mem_valid_for_store_merging): Return the bitsize, bitpos,
512 bitregion_start and bitregion_end as poly_uint64s rather than
513 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
514 (handled_load): Take the bitsize, bitpos,
515 bitregion_start and bitregion_end as poly_uint64s rather than
516 unsigned HOST_WIDE_INTs.
517 (pass_store_merging::process_store): Update call to
518 mem_valid_for_store_merging.
520 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
521 Alan Hayward <alan.hayward@arm.com>
522 David Sherwood <david.sherwood@arm.com>
524 * builtins.c (get_object_alignment_2): Track polynomial offsets
525 and sizes. Update the alignment handling.
527 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
528 Alan Hayward <alan.hayward@arm.com>
529 David Sherwood <david.sherwood@arm.com>
531 * tree.h (get_inner_reference): Add a version that returns the
532 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
533 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
534 the case in which bitpos is not associated with the first byte.
536 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
537 Alan Hayward <alan.hayward@arm.com>
538 David Sherwood <david.sherwood@arm.com>
540 * tree-affine.h (get_inner_reference_aff): Return the size as a
542 * tree-affine.c (get_inner_reference_aff): Likewise.
543 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
544 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
546 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
547 Alan Hayward <alan.hayward@arm.com>
548 David Sherwood <david.sherwood@arm.com>
550 * fold-const.c (pointer_may_wrap_p): Take the offset as a
551 HOST_WIDE_INT rather than a poly_int64.
553 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
554 Alan Hayward <alan.hayward@arm.com>
555 David Sherwood <david.sherwood@arm.com>
557 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
558 HOST_WIDE_INT to poly_int64_pod.
559 (perform_symbolic_merge): Update accordingly.
560 (bswap_replace): Likewise.
562 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
563 Alan Hayward <alan.hayward@arm.com>
564 David Sherwood <david.sherwood@arm.com>
566 * tree-affine.h (aff_tree::offset): Change from widest_int
568 (wide_int_ext_for_comb): Delete.
569 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
570 constants as poly_widest_int rather than widest_int.
571 (aff_combination_constant_multiple_p): Return the multiplier
572 as a poly_widest_int.
573 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
575 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
576 version static and add an overload for poly_widest_int.
577 (aff_combination_const, aff_combination_add_cst)
578 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
579 the constants as poly_widest_int rather than widest_int.
580 (tree_to_aff_combination): Generalize INTEGER_CST case to
582 (aff_combination_to_tree): Track offsets as poly_widest_ints.
583 (aff_combination_add_product, aff_combination_mult): Handle
585 (aff_combination_constant_multiple_p): Return the multiplier
586 as a poly_widest_int.
587 * tree-predcom.c (determine_offset): Return the offset as a
589 (split_data_refs_to_components, suitable_component_p): Update
591 (valid_initializer_p): Update call to
592 aff_combination_constant_multiple_p.
593 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
594 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
595 as a poly_int64 rather than a HOST_WIDE_INT.
596 (get_address_cost): Handle polynomial offsets.
597 (iv_elimination_compare_lt): Likewise.
598 (rewrite_use_nonlinear_expr): Likewise.
600 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
601 Alan Hayward <alan.hayward@arm.com>
602 David Sherwood <david.sherwood@arm.com>
604 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
605 as a poly_int64_pod rather than a HOST_WIDE_INT.
606 (get_addr_base_and_unit_offset): Likewise.
607 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
608 (get_addr_base_and_unit_offset): Likewise.
609 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
610 to poly_int64 in example.
611 * fold-const.c (fold_binary_loc): Update call to
612 get_addr_base_and_unit_offset.
613 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
614 (maybe_canonicalize_mem_ref_addr): Likewise.
615 (gimple_fold_stmt_to_constant_1): Likewise.
616 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
618 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
619 * match.pd: Likewise.
620 * omp-low.c (lower_omp_target): Likewise.
621 * tree-sra.c (build_ref_for_offset): Likewise.
622 (build_debug_ref_for_model): Likewise.
623 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
624 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
625 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
626 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
627 (constant_pointer_difference): Likewise.
628 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
629 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
630 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
631 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
632 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
633 (set_ssa_val_to): Likewise.
634 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
635 (maybe_diag_stxncpy_trunc): Likewise.
636 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
637 * tree.c (build_simple_mem_ref_loc): Likewise.
638 (array_at_struct_end_p): Likewise.
640 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
641 Alan Hayward <alan.hayward@arm.com>
642 David Sherwood <david.sherwood@arm.com>
644 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
645 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
646 (get_ref_base_and_extent_hwi): Declare.
647 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
648 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
649 (get_ref_base_and_extent_hwi): New function.
650 * cfgexpand.c (expand_debug_expr): Update call to
651 get_ref_base_and_extent.
652 * dwarf2out.c (add_var_loc_to_decl): Likewise.
653 * gimple-fold.c (get_base_constructor): Return the offset as a
654 poly_int64_pod rather than a HOST_WIDE_INT.
655 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
656 * ipa-polymorphic-call.c
657 (ipa_polymorphic_call_context::set_by_invariant)
658 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
659 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
660 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
661 rather than get_ref_base_and_extent.
662 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
663 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
664 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
666 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
667 call to get_ref_base_and_extent.
668 * tree-sra.c (create_access, get_access_for_expr): Likewise.
669 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
670 (stmt_kills_ref_p): Likewise.
671 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
672 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
674 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
675 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
676 when calling native_encode_expr.
677 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
678 call to get_ref_base_and_extent.
679 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
680 get_ref_base_and_extent.
681 * var-tracking.c (track_expr_p): Likewise.
683 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
684 Alan Hayward <alan.hayward@arm.com>
685 David Sherwood <david.sherwood@arm.com>
687 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
688 HOST_WIDE_INT to poly_int64_pod.
689 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
690 polynomail parameter offsets.
692 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
693 Alan Hayward <alan.hayward@arm.com>
694 David Sherwood <david.sherwood@arm.com>
696 * gengtype.c (main): Handle poly_int64_pod.
697 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
698 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
699 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
700 from HOST_WIDE_INT to poly_int64_pod.
701 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
702 (copy_cfa): New function.
703 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
704 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
705 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
706 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
707 (initial_return_save): Treat offsets as poly_ints.
708 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
709 offset is nonconstant.
710 (reg_save): Take the offset as a poly_int64. Fall back to
711 DW_CFA_expression for nonconstant offsets.
712 (queue_reg_save): Take the offset as a poly_int64.
713 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
715 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
716 Alan Hayward <alan.hayward@arm.com>
717 David Sherwood <david.sherwood@arm.com>
719 * rtl.h (operand_subword, operand_subword_force): Take the offset
720 as a poly_uint64 an unsigned int.
721 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
723 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
724 Alan Hayward <alan.hayward@arm.com>
725 David Sherwood <david.sherwood@arm.com>
727 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
728 'p' format code. Use INT_LIST rather than SUBREG as the example of
729 a code with an XINT and an XEXP. Remove the implication that
730 accessing an rtx field using XINT is expected to work.
731 * rtl.def (SUBREG): Change format from "ei" to "ep".
732 * rtl.h (rtunion::rt_subreg): New field.
733 (XCSUBREG): New macro.
734 (SUBREG_BYTE): Use it.
735 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
736 Update constructor accordingly.
737 (subreg_shape::operator ==): Update accordingly.
738 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
739 than an unsigned int.
740 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
741 a poly_uint64 rather than an unsigned int.
742 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
743 than an unsigned int.
744 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
745 (subreg_size_highpart_offset): Return a poly_uint64 rather than
746 an unsigned int. Take the sizes as poly_uint64s.
747 (subreg_offset_from_lsb): Return a poly_uint64 rather than
748 an unsigned int. Take the shift as a poly_uint64 rather than
750 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
751 as a poly_uint64 rather than an unsigned int.
752 (simplify_subreg_regno): Likewise.
753 (byte_lowpart_offset): Return the memory offset as a poly_int64
755 (subreg_memory_offset): Likewise. Take the subreg offset as a
756 poly_uint64 rather than an unsigned int.
757 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
758 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
759 poly_uint64 rather than an unsigned int.
760 * rtl.c (rtx_format): Describe 'p' in comment.
761 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
762 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
763 offset as a poly_uint64 rather than an unsigned int.
764 (byte_lowpart_offset): Return the memory offset as a poly_int64
766 (subreg_memory_offset): Likewise. Take the subreg offset as a
767 poly_uint64 rather than an unsigned int.
768 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
769 mode sizes as poly_uint64s rather than unsigned ints. Return a
770 poly_uint64 rather than an unsigned int.
771 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
772 (copy_insn_1): Handle 'p'.
773 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
774 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
775 an unsigned int. Return the shift in the same way.
776 (subreg_lsb): Return the shift as a poly_uint64 rather than an
778 (subreg_size_offset_from_lsb): Take the sizes and shift as
779 poly_uint64s rather than unsigned ints. Return the offset as
781 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
782 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
784 * rtlhash.c (add_rtx): Handle 'p'.
785 * genemit.c (gen_exp): Likewise.
786 * gengenrtl.c (type_from_format, gendef): Likewise.
787 * gensupport.c (subst_pattern_match, get_alternatives_number)
788 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
789 (subst_dup): Likewise.
790 * gengtype.c (adjust_field_rtx_def): Likewise.
791 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
792 (match_pattern_2): Likewise.
793 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
794 (rtx_test::subreg_field): New function.
795 (operator ==, safe_to_hoist_p, transition_parameter_type)
796 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
797 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
798 * genpeep.c (match_rtx): Likewise.
799 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
800 (rtx_writer::print_rtx_operand): Handle 'p'.
801 (print_value): Handle SUBREG.
802 * read-rtl.c (apply_int_iterator): Likewise.
803 (rtx_reader::read_rtx_operand): Handle 'p'.
804 * alias.c (rtx_equal_for_memref_p): Likewise.
805 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
806 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
808 * calls.c (expand_call): Likewise.
809 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
810 (make_extraction, gen_lowpart_for_combine): Likewise.
811 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
813 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
814 rather than an unsigned int. Treat subreg offsets as poly_ints.
815 (exp_equiv_p): Handle 'p'.
816 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
817 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
818 * dse.c (find_shift_sequence): Likewise.
819 * dwarf2out.c (rtl_for_decl_location): Likewise.
820 * expmed.c (extract_low_bits): Likewise.
821 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
822 (expand_expr_real_2): Likewise.
823 * final.c (alter_subreg): Likewise.
824 (leaf_renumber_regs_insn): Handle 'p'.
825 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
826 Treat subreg offsets as poly_ints.
827 * fwprop.c (forward_propagate_and_simplify): Likewise.
828 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
829 * ira.c (get_subreg_tracking_sizes): Likewise.
830 * ira-conflicts.c (go_through_subreg): Likewise.
831 * ira-lives.c (process_single_reg_class_operands): Likewise.
832 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
833 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
834 as a poly_uint64 rather than an unsigned int.
835 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
836 subreg offsets as poly_ints.
837 * lra-constraints.c (operands_match_p): Handle 'p'.
838 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
839 * lra-spills.c (assign_mem_slot): Likewise.
840 * postreload.c (move2add_valid_value_p): Likewise.
841 * recog.c (general_operand, indirect_operand): Likewise.
842 * regcprop.c (copy_value, maybe_mode_change): Likewise.
843 (copyprop_hardreg_forward_1): Likewise.
844 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
845 (record_subregs_of_mode): Likewise.
846 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
847 * reload.c (operands_match_p): Handle 'p'.
848 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
849 * reload1.c (alter_reg, choose_reload_regs): Likewise.
850 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
851 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
852 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
853 (simplify_const_poly_int_tests<N>::run): Likewise.
854 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
855 a poly_uint64 rather than an unsigned int.
856 * valtrack.c (debug_lowpart_subreg): Likewise.
857 * var-tracking.c (var_lowpart): Likewise.
858 (loc_cmp): Handle 'p'.
860 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
861 Alan Hayward <alan.hayward@arm.com>
862 David Sherwood <david.sherwood@arm.com>
864 * ira.c (get_subreg_tracking_sizes): New function.
865 (init_live_subregs): Take an integer size rather than a register.
866 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
867 to init_live_subregs.
869 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
870 Alan Hayward <alan.hayward@arm.com>
871 David Sherwood <david.sherwood@arm.com>
873 * expr.c (store_constructor_field): Change bitsize from a
874 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
875 HOST_WIDE_INT to a poly_int64.
876 (store_constructor): Change size from a HOST_WIDE_INT to
878 (store_field): Likewise bitsize and bitpos.
880 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
881 Alan Hayward <alan.hayward@arm.com>
882 David Sherwood <david.sherwood@arm.com>
884 * expmed.h (store_bit_field): Change bitregion_start and
885 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
886 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
887 (store_bit_field_1, store_integral_bit_field, store_bit_field)
888 (store_fixed_bit_field, store_split_bit_field): Likewise.
889 * expr.c (store_constructor_field, store_field): Likewise.
890 (optimize_bitfield_assignment_op): Likewise. Make the same change
891 to bitsize and bitpos.
892 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
893 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
894 same change in the constructor arguments.
895 (get_best_mode): Change bitregion_start and bitregion_end from
896 unsigned HOST_WIDE_INT to poly_uint64.
897 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
898 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
900 (bit_field_mode_iterator::next_mode): Update for new types
901 of m_bitregion_start and m_bitregion_end.
902 (get_best_mode): Change bitregion_start and bitregion_end from
903 unsigned HOST_WIDE_INT to poly_uint64.
905 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
906 Alan Hayward <alan.hayward@arm.com>
907 David Sherwood <david.sherwood@arm.com>
909 * rtl.h (simplify_gen_subreg): Add a temporary overload that
910 accepts poly_uint64 offsets.
911 * expmed.h (extract_bit_field): Take bitsize and bitnum as
912 poly_uint64s rather than unsigned HOST_WIDE_INTs.
913 * expmed.c (lowpart_bit_field_p): Likewise.
914 (extract_bit_field_as_subreg): New function, split out from...
915 (extract_bit_field_1): ...here. Take bitsize and bitnum as
916 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
917 extractions, check that BITSIZE matches the size of the extracted
918 value and that BITNUM is an exact multiple of that size.
919 If all else fails, try forcing the value into memory if
920 BITNUM is variable, and adjusting the address so that the
921 offset is constant. Split the part that can only handle constant
922 bitsize and bitnum out into...
923 (extract_integral_bit_field): ...this new function.
924 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
925 rather than unsigned HOST_WIDE_INTs.
927 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
928 Alan Hayward <alan.hayward@arm.com>
929 David Sherwood <david.sherwood@arm.com>
931 * expmed.h (store_bit_field): Take bitsize and bitnum as
932 poly_uint64s rather than unsigned HOST_WIDE_INTs.
933 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
934 that returns the byte size.
935 (store_bit_field_1): Take bitsize and bitnum as
936 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
937 to simple_mem_bitfield_p. Split the part that can only handle
938 constant bitsize and bitnum out into...
939 (store_integral_bit_field): ...this new function.
940 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
941 than unsigned HOST_WIDE_INTs.
942 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
944 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
945 Alan Hayward <alan.hayward@arm.com>
946 David Sherwood <david.sherwood@arm.com>
948 * lra-int.h (lra_reg): Change offset from int to poly_int64.
949 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
951 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
952 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
953 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
954 offset as a poly_int64 rather than an int.
955 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
956 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
957 * lra-constraints.c (equiv_address_substitution): Track offsets
959 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
960 (curr_insn_transform): Handle the new form of sp_offset.
961 * lra-eliminations.c (lra_elim_table): Change previous_offset
962 and offset from HOST_WIDE_INT to poly_int64.
963 (print_elim_table, update_reg_eliminate): Update accordingly.
964 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
965 (get_elimination): Update accordingly.
966 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
967 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
968 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
969 poly_int64 offsets generally.
970 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
971 (mark_not_eliminable, init_elimination): Update accordingly.
972 (remove_reg_equal_offset_note): Return a bool and pass the new
973 offset back by pointer as a poly_int64.
974 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
975 rather than a HOST_WIDE_INT.
976 (do_remat): Track offsets poly_int64s.
977 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
979 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
980 Alan Hayward <alan.hayward@arm.com>
981 David Sherwood <david.sherwood@arm.com>
983 * rtl.h (mem_attrs): Add a default constructor. Change size and
984 offset from HOST_WIDE_INT to poly_int64.
985 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
986 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
987 (widen_memory_access): Take the sizes and offsets as poly_int64s
988 rather than HOST_WIDE_INTs.
989 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
990 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
992 (adjust_offset_for_component_ref): Change the offset from a
993 HOST_WIDE_INT to a poly_int64.
994 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
995 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
996 * dce.c (find_call_stack_args): Likewise.
997 * dse.c (record_store): Likewise.
998 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
999 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1000 * read-rtl-function.c (test_loading_mem): Likewise.
1001 * rtlanal.c (may_trap_p_1): Likewise.
1002 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1003 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1004 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1005 (mem_attrs::mem_attrs): New function.
1006 (set_mem_attributes_minus_bitpos): Change bitpos from a
1007 HOST_WIDE_INT to poly_int64.
1008 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1009 (clear_mem_offset, clear_mem_size, change_address)
1010 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1011 initialize mem_attrs.
1012 (set_mem_offset, set_mem_size, adjust_address_1)
1013 (adjust_automodify_address_1, offset_address, widen_memory_access):
1014 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1016 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1017 Alan Hayward <alan.hayward@arm.com>
1018 David Sherwood <david.sherwood@arm.com>
1020 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1021 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1022 rather than 0 to represent an unknown size. Assert that the size
1023 is known when the mode isn't BLKmode.
1024 (may_trap_p_1): Use -1 for unknown sizes.
1025 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1027 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1028 Alan Hayward <alan.hayward@arm.com>
1029 David Sherwood <david.sherwood@arm.com>
1031 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1032 to poly_int64. Update commentary for positions_needed.large.
1033 (read_info_type): Change offset and width from HOST_WIDE_INT
1035 (set_usage_bits): Likewise.
1036 (canon_address): Return the offset as a poly_int64 rather than
1037 a HOST_WIDE_INT. Use strip_offset_and_add.
1038 (set_all_positions_unneeded, any_positions_needed_p): Use
1039 positions_needed.large to track stores with non-constant widths.
1040 (all_positions_needed_p): Likewise. Take the offset and width
1041 as poly_int64s rather than ints. Assert that rhs is nonnull.
1042 (record_store): Cope with non-constant offsets and widths.
1043 Nullify the rhs of an earlier store if we can't tell which bytes
1045 (find_shift_sequence): Take the access_size and shift as poly_int64s
1047 (get_stored_val): Take the read_offset and read_width as poly_int64s
1048 rather than HOST_WIDE_INTs.
1049 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1050 non-constant offsets and widths.
1052 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1053 Alan Hayward <alan.hayward@arm.com>
1054 David Sherwood <david.sherwood@arm.com>
1056 * inchash.h (inchash::hash::add_poly_int): New function.
1057 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1058 Use poly_int64 rather than HOST_WIDE_INT.
1059 (ao_ref::max_size_known_p): New function.
1060 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1061 rather than HOST_WIDE_INT.
1062 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1063 to temporaries until its interface is adjusted to match.
1064 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1065 (aliasing_component_refs_p, decl_refs_may_alias_p)
1066 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1067 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1068 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1070 * alias.c (ao_ref_from_mem): Likewise.
1071 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1072 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1073 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1074 (maybe_trim_complex_store, maybe_trim_constructor_store)
1075 (live_bytes_read, dse_classify_store): Likewise.
1076 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1077 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1078 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1079 (vn_reference_lookup_3): Likewise.
1080 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1082 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1083 Alan Hayward <alan.hayward@arm.com>
1084 David Sherwood <david.sherwood@arm.com>
1086 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1087 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1088 instead of ranges_overlap_p.
1090 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1091 Alan Hayward <alan.hayward@arm.com>
1092 David Sherwood <david.sherwood@arm.com>
1094 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1095 sizes as poly_int64s rather than HOST_WIDE_INTs.
1097 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1098 Alan Hayward <alan.hayward@arm.com>
1099 David Sherwood <david.sherwood@arm.com>
1101 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1102 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1103 * gimple-fold.c (fold_ctor_reference): Likewise.
1105 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1106 Alan Hayward <alan.hayward@arm.com>
1107 David Sherwood <david.sherwood@arm.com>
1109 * target.def (dwarf_poly_indeterminate_value): New hook.
1110 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1111 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1112 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1113 * doc/tm.texi: Regenerate.
1114 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1115 offset as a poly_int64.
1116 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1117 offset as a poly_int64.
1118 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1119 Take the offset as a poly_int64.
1120 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1121 (frame_pointer_fb_offset): Change to a poly_int64.
1122 (int_loc_descriptor): Take the offset as a poly_int64. Use
1123 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1124 (based_loc_descr): Take the offset as a poly_int64.
1125 Use strip_offset_and_add to handle (plus X (const)).
1126 Use new_reg_loc_descr instead of an open-coded version of the
1127 previous implementation.
1128 (mem_loc_descriptor): Handle CONST_POLY_INT.
1129 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1130 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1132 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1133 Alan Hayward <alan.hayward@arm.com>
1134 David Sherwood <david.sherwood@arm.com>
1136 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1137 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1138 * inchash.h (inchash::hash::add_poly_hwi): New function.
1139 * gengtype.c (main): Register poly_int64.
1140 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1141 offset as a poly_int.
1142 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1143 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1144 offset as a poly_int64.
1145 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1146 * print-rtl.c (print_poly_int): New function.
1147 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1149 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1151 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1152 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1153 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1154 Enforce track_offset_p for parts of a PARALLEL.
1155 (vt_add_function_parameter): Use const_offset for the final
1156 offset to track. Use get_tracked_reg_offset for the parts
1159 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1160 Alan Hayward <alan.hayward@arm.com>
1161 David Sherwood <david.sherwood@arm.com>
1163 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1164 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1165 * doc/tm.texi: Regenerate.
1166 * hooks.h (hook_bool_uint_uint_true): Delete.
1167 (hook_bool_puint64_puint64_true): Declare.
1168 * hooks.c (hook_bool_uint_uint_true): Delete.
1169 (hook_bool_puint64_puint64_true): New function.
1170 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1171 instead of unsigned ints.
1172 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1173 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1175 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1176 Alan Hayward <alan.hayward@arm.com>
1177 David Sherwood <david.sherwood@arm.com>
1179 * optabs.h (expand_operand): Add an int_value field.
1180 (create_expand_operand): Add an int_value parameter and use it
1181 to initialize the new expand_operand field.
1182 (create_integer_operand): Replace with a declaration of a function
1183 that accepts poly_int64s. Move the implementation to...
1184 * optabs.c (create_integer_operand): ...here.
1185 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1186 the mode preserves the value of int_value, instead of calling
1187 const_int_operand on the rtx. Use gen_int_mode to generate
1190 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1191 Alan Hayward <alan.hayward@arm.com>
1192 David Sherwood <david.sherwood@arm.com>
1194 * dumpfile.h (dump_dec): Declare.
1195 * dumpfile.c (dump_dec): New function.
1196 * pretty-print.h (pp_wide_integer): Turn into a function and
1197 declare a poly_int version.
1198 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1200 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1201 Alan Hayward <alan.hayward@arm.com>
1202 David Sherwood <david.sherwood@arm.com>
1204 * doc/generic.texi (POLY_INT_CST): Document.
1205 * tree.def (POLY_INT_CST): New tree code.
1206 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1207 * tree-core.h (tree_poly_int_cst): New struct.
1208 (tree_node): Add a poly_int_cst field.
1209 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1210 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1211 instead of a wide_int_ref.
1212 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1214 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1215 instead of an unsigned HOST_WIDE_INT.
1216 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1217 (ptrdiff_tree_p): Declare.
1218 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1219 extern inline implementations if the target doesn't use POLY_INT_CST.
1220 (poly_int_tree_p): New function.
1221 (wi::unextended_tree): New class.
1222 (wi::int_traits <unextended_tree>): New override.
1223 (wi::extended_tree): Add a default constructor.
1224 (wi::extended_tree::get_tree): New function.
1225 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1226 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1227 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1228 (wi::tree_to_poly_wide_ref): New typedefs.
1229 (wi::ints_for): Provide overloads for extended_tree and
1231 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1232 (wi::to_wide): New functions.
1233 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1234 * tree.c (poly_int_cst_hasher): New struct.
1235 (poly_int_cst_hash_table): New variable.
1236 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1237 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1239 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1240 (init_ttree): Initialize poly_int_cst_hash_table.
1241 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1242 a poly_int64 instead of a HOST_WIDE_INT.
1243 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1244 instead of an unsigned HOST_WIDE_INT.
1245 (wide_int_to_tree): Rename to...
1246 (wide_int_to_tree_1): ...this.
1247 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1248 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1249 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1250 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1251 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1252 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1254 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1255 (streamer_read_tree_body): Likewise.
1256 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1257 (streamer_write_tree_body): Likewise.
1258 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1259 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1260 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1261 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1262 * gimple-expr.h (is_gimple_constant): Likewise.
1263 * gimplify.c (maybe_with_size_expr): Likewise.
1264 * print-tree.c (print_node): Likewise.
1265 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1266 * tree-pretty-print.c (dump_generic_node): Likewise.
1267 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1268 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1269 * tree-vrp.c (compare_values_warnv): Likewise.
1270 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1271 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1272 (force_expr_to_var_cost): Likewise.
1273 * tree-ssa-loop.c (for_each_index): Likewise.
1274 * fold-const.h (build_invariant_address, size_int_kind): Take a
1275 poly_int64 instead of a HOST_WIDE_INT.
1276 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1277 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1279 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1280 (int_const_binop_2): New function, split out from...
1281 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1282 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1283 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1284 * gimple-ssa-strength-reduction.c (slsr_process_add)
1285 (slsr_process_mul): Check for INTEGER_CSTs before using them
1287 * stor-layout.c (bits_from_bytes): New function.
1288 (bit_from_pos): Use it.
1289 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1290 by BITS_PER_UNIT to get the TYPE_SIZE.
1291 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1292 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1294 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1295 Alan Hayward <alan.hayward@arm.com>
1296 David Sherwood <david.sherwood@arm.com>
1298 * doc/rtl.texi (const_poly_int): Document. Also document the
1299 rtl sharing behavior.
1300 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1301 * rtl.h (const_poly_int_def): New struct.
1302 (rtx_def::u): Add a cpi field.
1303 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1304 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1305 (wi::rtx_to_poly_wide_ref): New typedef
1306 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1307 (poly_int_rtx_p): New functions.
1308 (trunc_int_for_mode): Declare a poly_int64 version.
1309 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1310 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1312 (strip_offset): Declare.
1313 (strip_offset_and_add): New function.
1314 * rtl.def (CONST_POLY_INT): New rtx code.
1315 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1316 (shared_const_p): Use poly_int_rtx_p.
1317 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1319 (gen_int_shift_amount): Likewise.
1320 * emit-rtl.c (const_poly_int_hasher): New class.
1321 (const_poly_int_htab): New variable.
1322 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1323 (const_poly_int_hasher::hash): New function.
1324 (const_poly_int_hasher::equal): Likewise.
1325 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1326 (immed_wide_int_const): Rename to...
1327 (immed_wide_int_const_1): ...this and make static.
1328 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1329 instead of a wide_int_ref.
1330 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1331 (gen_lowpart_common): Handle CONST_POLY_INT.
1332 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1333 * cselib.c (cselib_hash_rtx): Likewise.
1334 * dwarf2out.c (const_ok_for_output_1): Likewise.
1335 * expr.c (convert_modes): Likewise.
1336 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1337 * rtlhash.c (add_rtx): Likewise.
1338 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1339 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1340 Handle existing CONST_POLY_INT rtxes.
1341 * expmed.h (expand_shift): Take a poly_int64 instead of a
1343 * expmed.c (expand_shift): Likewise.
1344 * rtlanal.c (strip_offset): New function.
1345 (commutative_operand_precedence): Give CONST_POLY_INT the same
1346 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1348 * rtl-tests.c (const_poly_int_tests): New struct.
1349 (rtl_tests_c_tests): Use it.
1350 * simplify-rtx.c (simplify_const_unary_operation): Handle
1352 (simplify_const_binary_operation): Likewise.
1353 (simplify_binary_operation_1): Fold additions of symbolic constants
1354 and CONST_POLY_INTs.
1355 (simplify_subreg): Handle extensions and truncations of
1357 (simplify_const_poly_int_tests): New struct.
1358 (simplify_rtx_c_tests): Use it.
1359 * wide-int.h (storage_ref): Add default constructor.
1360 (wide_int_ref_storage): Likewise.
1361 (trailing_wide_ints): Use GTY((user)).
1362 (trailing_wide_ints::operator[]): Add a const version.
1363 (trailing_wide_ints::get_precision): New function.
1364 (trailing_wide_ints::extra_size): Likewise.
1366 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1367 Alan Hayward <alan.hayward@arm.com>
1368 David Sherwood <david.sherwood@arm.com>
1370 * emit-rtl.h (gen_int_shift_amount): Declare.
1371 * emit-rtl.c (gen_int_shift_amount): New function.
1372 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1374 * calls.c (shift_return_value): Likewise.
1375 * cse.c (fold_rtx): Likewise.
1376 * dse.c (find_shift_sequence): Likewise.
1377 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1378 (expand_shift, expand_smod_pow2): Likewise.
1379 * lower-subreg.c (shift_cost): Likewise.
1380 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1381 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1382 (expand_vec_perm_var): Likewise.
1383 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1384 (simplify_binary_operation_1): Likewise.
1385 * combine.c (try_combine, find_split_point, force_int_to_mode)
1386 (simplify_shift_const_1, simplify_shift_const): Likewise.
1387 (change_zero_ext): Likewise. Use simplify_gen_binary.
1389 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1391 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1393 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1395 * doc/invoke.texi (ARM Options): Document accepted extension options
1396 for -march=armv8.3-a.
1398 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1401 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1402 or --with-float=softfp, set the default CPU to arm10e.
1404 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1406 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1407 * config/visium/predicates.md (const_shift_operand): Likewise.
1408 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1409 (visium_legitimize_reload_address): Likewise.
1411 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1414 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1416 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1419 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1421 2017-12-20 Julia Koval <julia.koval@intel.com>
1423 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1424 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1425 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1426 * config.gcc: Include vpclmulqdqintrin.h.
1427 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1428 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1430 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1431 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1432 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1433 * config/i386/i386.c (isa2_opts): Add -mcx16.
1434 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1435 (ix86_option_override_internal): Move mcx16 to flags2.
1436 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1437 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1438 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1439 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1440 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1441 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1442 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1443 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1444 * doc/invoke.texi: Add -mvpclmulqdq.
1446 2017-12-20 Tom de Vries <tom@codesourcery.com>
1449 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
1450 * calls.c (rtx_for_static_chain): ... here. New function.
1451 * calls.h (rtx_for_static_chain): Declare.
1452 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1453 instead of targetm.calls.static_chain.
1454 * df-scan.c (df_get_entry_block_def_set): Same.
1456 2017-12-19 Tom de Vries <tom@codesourcery.com>
1458 PR tree-optimization/83493
1459 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1462 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1464 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1465 inform with hardcoded english plural handling.
1467 2017-12-18 Jeff Law <law@redhat.com>
1469 PR tree-optimization/83477
1470 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1471 a non-virtual PHI, always push a new range.
1473 2017-12-19 Martin Sebor <msebor@redhat.com>
1476 * builtins.c (compute_objsize): Handle non-constant offsets.
1478 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1480 PR tree-optimization/83444
1481 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1482 character load case, if get_stridx on MEM_REF's operand doesn't
1483 look usable, retry with get_addr_stridx.
1485 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1488 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1489 (variable_tracking_main_1): Keep markers even when VTA fails.
1492 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1493 even if there are markers before it.
1494 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1496 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1498 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1499 typical order conditions.
1500 * sel-sched.c (extract_new_fences_from): Likewise.
1501 * config/visium/constraints.md (J, K, L): Likewise.
1502 * config/visium/predicates.md (const_shift_operand): Likewise.
1503 * config/visium/visium.c (visium_legitimize_address,
1504 visium_legitimize_reload_address): Likewise.
1505 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1506 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1507 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1508 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1509 SET_DUMP_DETAIL): Likewise.
1510 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1511 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1512 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1513 avr_set_core_architecture, avr_set_current_function,
1514 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1515 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1516 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1517 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1518 avr_map_decompose, avr_fold_builtin): Likewise.
1519 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1520 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1521 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1522 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1523 * config/m32c/m32c.c (m32c_conditional_register_usage,
1524 m32c_address_cost): Likewise.
1525 * config/m32c/predicates.md (shiftcount_operand,
1526 longshiftcount_operand): Likewise.
1527 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1528 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1529 can_use_cdx_ldstw): Likewise.
1530 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1531 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1533 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1534 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1535 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1536 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1537 * config/vax/vax.c (adjacent_operands_p): Likewise.
1538 * config/ft32/constraints.md (L, b, KA): Likewise.
1539 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1541 * cfgexpand.c (expand_stack_alignment): Likewise.
1542 * gcse.c (insert_expr_in_table): Likewise.
1543 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1544 * cgraphunit.c (cgraph_node::expand): Likewise.
1545 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1546 * emit-rtl.c (add_insn): Likewise.
1547 * input.c (dump_location_info): Likewise.
1548 * passes.c (NEXT_PASS): Likewise.
1549 * read-rtl-function.c (parse_note_insn_name,
1550 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1552 * sched-rgn.c (sched_rgn_init): Likewise.
1553 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1554 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1555 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
1556 * explow.c (eliminate_constant_term): Likewise.
1557 * final.c (leaf_renumber_regs_insn): Likewise.
1558 * cfgrtl.c (print_rtl_with_bb): Likewise.
1559 * genhooks.c (emit_init_macros): Likewise.
1560 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1561 * tree-data-ref.c (conflict_fn): Likewise.
1562 * selftest.c (assert_streq): Likewise.
1563 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1564 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1565 fold_binary_loc, multiple_of_p): Likewise.
1566 * reload.c (push_reload, find_equiv_reg): Likewise.
1567 * et-forest.c (et_nca, et_below): Likewise.
1568 * dbxout.c (dbxout_symbol_location): Likewise.
1569 * reorg.c (relax_delay_slots): Likewise.
1570 * dojump.c (do_compare_rtx_and_jump): Likewise.
1571 * gengtype-parse.c (type): Likewise.
1572 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1573 simplify_const_relational_operation): Likewise.
1574 * reload1.c (do_output_reload): Likewise.
1575 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1576 * gengtype.c (type_for_name): Likewise.
1577 * gimple-ssa-sprintf.c (format_directive): Likewise.
1579 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1582 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1583 accessing it. Adjust comment.
1585 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1588 * predict.c (zero_one_minusone): New function.
1589 (apply_return_prediction): Avoid return prediction for functions
1590 returning only -1, 0 and 1 values, unless they only return -1 and 0
1593 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1595 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1596 fall-through warning.
1597 (arc_compute_frame_size): Remove unused variables.
1598 (arc_print_operand): Fix fprintif format.
1599 (arc_can_follow_jump): Clean fall-through warning.
1601 2017-12-19 Marek Polacek <polacek@redhat.com>
1604 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1607 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1609 * config/arc/arc.c (overriderregs): New variable.
1610 (arc_override_options): Track fixed/call saved/call options.
1611 (arc_conditional_register_usage): Check against overrideregs
1612 variable whenever we change register properties.
1614 2017-12-19 Nathan Sidwell <nathan@acm.org>
1616 * opts.c (finish_options): Don't prefix dump_base_name if it
1617 already contains directories.
1619 2017-12-19 Martin Liska <mliska@suse.cz>
1621 PR rtl-optimization/82675
1622 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1623 more element in sbitmap.
1625 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1627 * gimplify.c (gimplify_expr): Use error_operand_p.
1629 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1632 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1635 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1637 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1638 the jump is to a label.
1640 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1642 PR tree-optimization/83444
1643 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1648 * ipa-inline.c (flatten_remove_node_hook): New function.
1649 (ipa_inline): Keep only nodes with flatten attribute at the end of
1650 the array in the order from ipa_reverse_postorder, only walk that
1651 portion of array for flattening, if there is more than one such
1652 node, temporarily register a removal hook and ignore removed nodes.
1654 PR tree-optimization/80631
1655 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1656 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1657 IFN_REDUC_MAX or IFN_REDUC_MIN.
1659 2017-12-18 Jeff Law <law@redhat.com>
1661 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1662 record symbolic equivalences from backedges in the CFG.
1665 2017-11-19 Jeff Law <law@redhat.com>
1667 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1668 of degenerates resulting from ignoring an edge.
1670 2017-12-18 Martin Sebor <msebor@redhat.com>
1673 PR tree-optimization/78450
1674 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1675 (handle_builtin_strlen): Call it.
1677 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1679 PR rtl-optimization/83424
1680 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1682 2017-12-18 Marek Polacek <polacek@redhat.com>
1685 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1686 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1689 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1691 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1692 (umaddsidi4, umaddsidi_split): Likewise.
1694 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1696 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1699 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1702 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1704 2017-12-18 Richard Biener <rguenther@suse.de>
1706 PR tree-optimization/81877
1707 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1708 (outermost_indep_loop): Adjust.
1709 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1710 (can_sm_ref_p): Adjust.
1712 2017-12-18 Richard Biener <rguenther@suse.de>
1715 * tree.c (array_at_struct_end_p): Return true if the underlying
1716 object has space for at least one element in excess of what
1717 the array domain specifies.
1719 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1721 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1722 example to avoid overfull hbox.
1723 * doc/invoke.texi (Option Summary): Add missing @gol.
1724 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1727 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1728 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1730 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1731 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1732 (AArch64 Options): Document -mverbose-cost-dump.
1733 (ARM Options): Likewise, plus -mflip-thumb.
1735 2017-12-17 Martin Sebor <msebor@redhat.com>
1738 * gimple-ssa-warn-restrict.c
1739 (builtin_memref::offset_out_of_bounds): Correct the handling of
1742 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1744 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1746 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1748 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1750 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1752 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1755 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1757 * doc/invoke.texi: Fix some typos.
1759 2017-12-16 Martin Sebor <msebor@redhat.com>
1761 PR tree-optimization/78918
1762 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1763 * builtins.c (check_sizes): Rename...
1764 (check_access): ...to this. Rename function arguments for clarity.
1765 (check_memop_sizes): Adjust names.
1766 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1767 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1768 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1769 (check_strncat_sizes, expand_builtin_strncat): Same.
1770 (expand_builtin_strncpy, expand_builtin_memset): Same.
1771 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1772 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1773 (maybe_emit_sprintf_chk_warning): Same.
1774 (expand_builtin_strcpy): Adjust.
1775 (expand_builtin_stpcpy): Same.
1776 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1777 in pointer-checking forms of memcpy, memmove, and mempcpy.
1778 (gcall_to_tree_minimal, max_object_size): Define new functions.
1779 * builtins.h (max_object_size): Declare.
1780 * calls.c (alloc_max_size): Call max_object_size instead of
1781 hardcoding ssizetype limit.
1782 (get_size_range): Handle new argument.
1783 * calls.h (get_size_range): Add a new argument.
1784 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1785 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1786 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1788 (gimple_fold_builtin_memory_chk): Same.
1789 (gimple_fold_builtin_stxcpy_chk): New function.
1790 * gimple-ssa-warn-restrict.c: New source.
1791 * gimple-ssa-warn-restrict.h: New header.
1792 * gimple.c (gimple_build_call_from_tree): Propagate location.
1793 * passes.def (pass_warn_restrict): Add new pass.
1794 * tree-pass.h (make_pass_warn_restrict): Declare.
1795 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1797 (handle_builtin_strcat): Same.
1798 (strlen_optimize_stmt): Rename...
1799 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1800 stpncpy, strncpy, and their checking forms.
1802 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1804 PR rtl-optimization/82849
1805 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1806 and get_max_loop_iterations_int.
1808 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1809 Alan Hayward <alan.hayward@arm.com>
1810 David Sherwood <david.sherwood@arm.com>
1812 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1813 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1814 size as a poly_uint64.
1815 (mode_for_vector, mode_for_int_vector): Take the number of vector
1816 elements as a poly_uint64.
1817 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1818 size as a poly_uint64.
1819 (mode_for_vector, mode_for_int_vector): Take the number of vector
1820 elements as a poly_uint64.
1822 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1823 Alan Hayward <alan.hayward@arm.com>
1824 David Sherwood <david.sherwood@arm.com>
1826 * machmode.h (MACRO_MODE): New macro.
1827 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1828 * targhooks.c (default_libcall_value, default_secondary_reload)
1829 (default_memory_move_cost, default_register_move_cost)
1830 (default_class_max_nregs): Likewise.
1832 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1833 Alan Hayward <alan.hayward@arm.com>
1834 David Sherwood <david.sherwood@arm.com>
1836 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1837 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1839 * genautomata.c (main): Likewise.
1840 * genconditions.c (write_header): Likewise.
1841 * genemit.c (main): Likewise.
1842 * genextract.c (print_header): Likewise.
1843 * genopinit.c (main): Likewise.
1844 * genoutput.c (output_prologue): Likewise.
1845 * genpeep.c (main): Likewise.
1846 * genpreds.c (write_insn_preds_c): Likewise.
1847 * genrecog.c (writer_header): Likewise.
1848 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1849 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1850 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
1851 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
1852 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
1853 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
1854 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
1855 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
1856 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
1857 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
1858 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
1859 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
1860 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
1861 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
1862 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
1863 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
1864 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
1865 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
1866 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
1867 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
1868 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
1869 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
1870 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
1871 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
1872 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
1873 * config/darwin.c (IN_TARGET_CODE): Likewise.
1874 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
1875 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
1876 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
1877 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
1878 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
1879 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
1880 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
1881 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
1882 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
1883 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
1884 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
1885 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
1886 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
1887 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
1888 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
1889 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
1890 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
1891 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
1892 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
1893 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
1894 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
1895 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
1896 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
1897 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
1898 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
1899 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
1900 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
1901 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
1902 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
1903 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
1904 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
1905 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
1906 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
1907 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
1908 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
1909 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
1910 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
1911 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
1912 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
1913 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
1914 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
1915 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
1916 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
1917 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
1918 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
1919 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
1920 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
1921 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
1922 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
1923 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
1924 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
1925 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
1926 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
1927 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
1928 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
1929 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
1930 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
1931 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
1932 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
1933 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
1934 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1935 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
1936 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
1937 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
1938 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
1939 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
1940 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
1941 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
1942 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
1943 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
1944 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
1945 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1946 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
1947 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
1948 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
1949 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
1950 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
1951 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
1952 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
1953 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
1954 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
1955 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
1956 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
1957 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
1958 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
1959 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
1960 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
1961 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
1962 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
1963 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
1964 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
1965 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
1966 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
1967 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
1968 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
1969 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
1970 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
1971 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
1972 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
1973 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
1974 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
1975 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
1976 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
1977 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
1978 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
1979 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
1981 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1982 Alan Hayward <alan.hayward@arm.com>
1983 David Sherwood <david.sherwood@arm.com>
1985 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
1988 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1989 Alan Hayward <alan.hayward@arm.com>
1990 David Sherwood <david.sherwood@arm.com>
1992 * doc/generic.texi (VEC_SERIES_EXPR): Document.
1993 * doc/md.texi (vec_series@var{m}): Document.
1994 * tree.def (VEC_SERIES_EXPR): New tree code.
1995 * tree.h (build_vec_series): Declare.
1996 * tree.c (build_vec_series): New function.
1997 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
1998 * tree-pretty-print.c (dump_generic_node): Likewise.
1999 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2000 * tree-inline.c (estimate_operator_cost): Likewise.
2001 * expr.c (expand_expr_real_2): Likewise.
2002 * optabs-tree.c (optab_for_tree_code): Likewise.
2003 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2004 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2005 * expmed.c (make_tree): Handle VEC_SERIES.
2006 * optabs.def (vec_series_optab): New optab.
2007 * optabs.h (expand_vec_series_expr): Declare.
2008 * optabs.c (expand_vec_series_expr): New function.
2009 * tree-vect-generic.c (expand_vector_operations_1): Check that
2010 the operands also have vector type.
2012 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2013 Alan Hayward <alan.hawyard@arm.com>
2014 David Sherwood <david.sherwood@arm.com>
2016 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2017 (VEC_COND_EXPR): Add missing @tindex.
2018 * doc/md.texi (vec_duplicate@var{m}): Document.
2019 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2020 * tree.c (build_vector_from_val): Add stubbed-out handling of
2021 variable-length vectors, using VEC_DUPLICATE_EXPR.
2022 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2023 * cfgexpand.c (expand_debug_expr): Likewise.
2024 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2025 * tree-inline.c (estimate_operator_cost): Likewise.
2026 * tree-pretty-print.c (dump_generic_node): Likewise.
2027 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2028 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2029 (test_vec_duplicate_folding): New function.
2030 (fold_const_c_tests): Call it.
2031 * optabs.def (vec_duplicate_optab): New optab.
2032 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2033 * optabs.h (expand_vector_broadcast): Declare.
2034 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2035 vec_duplicate_optab.
2036 * expr.c (store_constructor): Try using vec_duplicate_optab for
2038 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2040 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2043 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2044 div/mod latencies a bit.
2046 2017-12-15 Jeff Law <law@redhat.com>
2048 PR tree-optimization/36550
2049 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2050 (mark_threaded_blocks): Rewrite code to avoid block copying when
2051 optimizing for size. Don't pessimize blocks which will be
2052 copied, but all the statements will be dead.
2054 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2056 PR tree-optimization/81165
2057 * tree-ssa-threadupdate.c (uses_in_bb): New.
2058 (estimate_threading_killed_stmts): New.
2059 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2060 * tree-ssa-threadedge.c
2061 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2064 2017-12-15 Jeff Law <law@redhat.com>
2066 PR tree-optimization/83410
2067 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2068 threads when parallelizing loops.
2070 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2072 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2074 * config/alpha/alpha.c (vms_attribute_table): Swap
2075 affects_type_identity and handler fields, adjust comments.
2076 * config/mips/mips.c (mips_attribute_table): Likewise.
2077 * config/visium/visium.c (visium_attribute_table): Likewise.
2078 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2079 * config/microblaze/microblaze.c (microblaze_attribute_table):
2081 * config/spu/spu.c (spu_attribute_table): Likewise.
2082 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2083 * config/arc/arc.c (arc_attribute_table): Likewise.
2084 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2085 * config/v850/v850.c (v850_handle_interrupt_attribute,
2086 v850_handle_data_area_attribute): Formatting fixes.
2087 (v850_attribute_table): Swap affects_type_identity and handler
2088 fields, adjust comments.
2089 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2090 * config/arm/arm.c (arm_attribute_table): Likewise.
2091 * config/avr/avr.c (avr_attribute_table): Likewise.
2092 * config/s390/s390.c (s390_attribute_table): Likewise.
2093 * config/sh/sh.c (sh_attribute_table): Likewise.
2094 * config/i386/i386.c (ix86_handle_cconv_attribute,
2095 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2096 (ix86_attribute_table): Swap affects_type_identity and handler
2097 fields, adjust comments.
2098 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2099 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2100 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2101 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2102 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2103 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2104 * config/rx/rx.c (rx_attribute_table): Likewise.
2105 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2106 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2107 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2108 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2109 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2110 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2111 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2112 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2113 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2114 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2115 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2117 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2119 (register_scoped_attributes, decl_attributes): Formatting fixes.
2121 PR tree-optimization/83269
2122 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2123 subtraction in arg0's type if type is signed and arg0 is unsigned.
2127 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2128 convert? on @0 instead of convert. Check type of @1, not @0.
2129 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2130 convert? on @0 instead of convert. Check type of @1, not @0.
2131 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2132 only optimize if either both @1 and @2 types are narrower
2133 precision, or both are wider or equal precision, and in the former
2134 case only if both have undefined overflow.
2136 2017-12-15 Richard Biener <rguenther@suse.de>
2139 * internal-fn.def (IFN_NOP): Add.
2140 * internal-fn.c (expand_NOP): Do nothing.
2141 * lto-streamer-in.c (input_function): Instead of removing
2142 sanitizer calls replace them with IFN_NOP calls.
2144 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2145 Alan Hayward <alan.hayward@arm.com>
2146 David Sherwood <david.sherwood@arm.com>
2148 * dse.c (store_info, read_info_type): Replace begin and end with
2150 (print_range): New function.
2151 (set_all_positions_unneeded, any_positions_needed_p)
2152 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2154 (record_store): Likewise. Optimize the case in which all positions
2156 (get_stored_val): Replace read_begin and read_end with read_offset
2158 (replace_read): Update call accordingly.
2160 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2162 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2163 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2164 (loop_cand::loop_cand): Initialize above members.
2165 (loop_cand::supported_operations): Delete.
2166 (loop_cand::can_interchange_p): Inline above function.
2167 (loop_cand::classify_simple_reduction): Record number of constant
2168 initialized simple reductions.
2169 (should_interchange_loops): New parameters. Check stmt cost of loops
2171 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2172 Update call to should_interchange_loops.
2173 (should_interchange_loop_nest): Update call to
2174 should_interchange_loops.
2176 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2179 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2180 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2181 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2183 2017-12-15 Julia Koval <julia.koval@intel.com>
2185 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2186 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
2187 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2188 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2189 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2191 2017-12-15 Julia Koval <julia.koval@intel.com>
2193 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2194 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2195 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2196 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2197 _mm_aesenc_epi128): New intrinsics.
2199 2017-12-15 Julia Koval <julia.koval@intel.com>
2201 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2202 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
2203 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2204 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2205 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2207 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2209 * gimple-ssa-strength-reduction.c (analyze_increments):
2210 Distinguish replacement costs for constant strides from those for
2213 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2215 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2217 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2219 * doc/invoke.texi: Document -Wcast-function-type.
2220 * recog.h (stored_funcptr): Change signature.
2221 * tree-dump.c (dump_node): Avoid warning.
2222 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2224 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2227 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2230 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2232 PR tree-optimization/83312
2233 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2234 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2235 "val" param, and to cope with arbitrary basic blocks.
2236 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2237 handle NULL_TREE for "val", dropping "bb" param.
2238 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2239 drop "bb" param. Handle NULL_TREE for "val".
2240 (find_case_label_for_value): Make "switch_stmt" param const.
2241 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2243 (vrp_prop::check_all_array_refs): Reimplement as...
2244 (check_array_bounds_dom_walker::before_dom_children): ...this new
2245 vfunc. Replace linear search through BB block list, excluding
2246 those with non-executable in-edges via dominator walk.
2248 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2250 * config/arm/arm.opt (mverbose-cost-dump): New option.
2251 * config/arm/arm.c (arm_rtx_costs): Use it.
2253 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2256 * reload1.c (emit_input_reload_insns): Skip debug markers.
2258 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2260 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2264 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2265 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2266 * config/sh/sh.c: Skip debug insns besides notes.
2267 * config/sh/sh.md: Likewise.
2268 * config/sh/sh_treg_combine.cc: Likewise.
2269 * config/sh/sync.md: Likewise.
2271 2017-12-14 Tom de Vries <tom@codesourcery.com>
2273 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2274 Add item for weak_undefined.
2276 2017-12-14 Richard Biener <rguenther@suse.de>
2278 PR tree-optimization/67842
2279 * sese.h (bb_in_region): Remove #if 0'ed code.
2281 2017-12-14 Richard Biener <rguenther@suse.de>
2283 PR tree-optimization/83326
2284 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2285 may_be_zero parameter and handle it by not marking the first
2286 peeled copy as not exiting the loop.
2287 (try_peel_loop): Likewise.
2288 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2289 to handle the case of constant or zero iterations and perform
2290 loop header copying on-the-fly.
2292 2017-12-14 Richard Biener <rguenther@suse.de>
2294 PR tree-optimization/83418
2295 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2296 Instead of asserting we don't get unfolded comparisons deal with
2299 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2302 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2303 basic blocks. Assert debug bind insns don't appear outside of bbs,
2304 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2307 PR tree-optimization/83198
2308 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2309 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2310 value if arg is a REAL_CST with incompatible type.
2312 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2313 Bin Cheng <bin.cheng@arm.com>
2316 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2318 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2321 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2323 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2324 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2326 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2327 Alan Hayward <alan.hayward@arm.com>
2328 David Sherwood <david.sherwood@arm.com>
2330 * poly-int.h: New file.
2331 * poly-int-types.h: Likewise.
2332 * coretypes.h: Include them.
2333 (POLY_INT_CONVERSION): Define.
2334 * target.def (estimated_poly_value): New hook.
2335 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2336 * doc/tm.texi: Regenerate.
2337 * doc/poly-int.texi: New file.
2338 * doc/gccint.texi: Include it.
2339 * doc/rtl.texi: Describe restrictions on subreg modes.
2340 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2341 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2342 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2343 * targhooks.h (default_estimated_poly_value): Declare.
2344 * targhooks.c (default_estimated_poly_value): New function.
2345 * target.h (estimated_poly_value): Likewise.
2346 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2347 (wi::unary_traits): Delete.
2348 (wi::binary_traits::signed_shift_result_type): Define for
2349 offset_int << HOST_WIDE_INT, etc.
2350 (generic_wide_int::operator <<=): Define for all types and use
2351 wi::lshift instead of <<.
2352 (wi::hwi_with_prec): Add a default constructor.
2353 (wi::ints_for): New class.
2354 (operator <<): Define for all wide-int types.
2355 (operator /): New function.
2356 (operator %): Likewise.
2357 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2358 (ASSERT_MAYBE_NE_AT): New macros.
2360 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2361 Dominik Vogt <vogt@linux.vnet.ibm.com>
2364 * emit-rtl.c (init_emit): Remove ??? comment.
2365 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2366 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2367 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2368 alignment of 3 virtual registers to BITS_PER_WORD.
2370 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2372 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2374 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2375 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2377 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2380 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2382 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2384 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2385 with the exception of debug begin stmt markers appear before
2389 * final.c (rest_of_handle_final): Call variable_tracking_main only
2390 if !flag_var_tracking.
2392 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2393 Jakub Jelinek <jakub@redhat.com>
2397 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2398 labels debug stmts that can only appear after labels.
2400 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2402 PR rtl-optimization/82398
2403 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2404 EXPR_USEFULNESS in priority comparison.
2406 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2408 PR rtl-optimization/83393
2409 * combine.c (move_deaths): If reg_stat points to a too new insn in
2410 last_death, do not use it: find the proper insn instead.
2412 2017-12-12 Jeff Law <law@redhat.com>
2414 PR tree-optimization/83298
2415 PR tree-optimization/83362
2416 PR tree-optimization/83383
2417 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2418 push_value_range a public interface. Add new argument to
2419 record_ranges_from_stmt.
2420 * gimple-ssa-evrp-analyze.c
2421 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2422 Update comments. Handle recording temporary equivalences.
2423 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2424 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2425 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2426 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2427 gimple-ssa-evrp-analyze.h.
2428 (record_temporary_equivalences_from_phis): Add new argument. When
2429 the PHI arg is an SSA_NAME, set the result's range to the range
2431 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2432 from statements too.
2433 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2434 Pass it down to children as needed.
2435 (thread_outgoing_edges): Likewise.
2436 (thread_across_edge): Likewise. Push/pop range state as needed.
2437 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2439 2017-12-12 Julia Koval <julia.koval@intel.com>
2441 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2442 (PTA_CANNONLAKE): Remove PTA_CLWB.
2444 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2447 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2448 (vcondv2div2df): Likewise.
2449 (vconduv2dfv2di): Likewise.
2451 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2453 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2454 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2455 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2456 ATTR_MATHFN_FPROUNDING_ERRNO.
2458 2017-12-12 Richard Biener <rguenther@suse.de>
2460 PR tree-optimization/83385
2461 * graphite-scop-detection.c (get_order, order): Remove.
2462 (bb_to_rpo): New global.
2464 (build_scops): Sort pbbs in RPO order.
2466 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2468 * combine.c (simplify_set): Do not transform subregs to zero_extends
2469 if the destination mode is a vector mode.
2471 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2473 PR tree-optimization/83359
2474 * tree-cfg.h (fold_loop_internal_call): Declare.
2475 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2476 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2477 (find_loop_dist_alias): New function.
2478 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2479 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2482 PR tree-optimization/80631
2483 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2484 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2485 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2486 hardcoding zero as the value if COND_EXPR is never true. For
2487 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2488 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2489 hardcoding MAX_EXPR as the reduction operation.
2490 (is_nonwrapping_integer_induction): Allow negative step.
2491 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2492 vect_create_epilog_for_reduction, if no value is suitable, don't
2493 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2495 2017-12-12 Richard Biener <rguenther@suse.de>
2497 PR tree-optimization/81889
2498 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2499 range info from the non-wrapping IV instead of just the range
2502 2017-12-12 Julia Koval <julia.koval@intel.com>
2504 * config.gcc: Add vaesintrin.h.
2505 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2507 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2508 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2510 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2511 * config/i386/immintrin.h: Include vaesintrin.h.
2512 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2513 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2514 _mm_aesdec_epi128): New intrinsics.
2516 2017-12-12 Julia Koval <julia.koval@intel.com>
2518 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2519 OPTION_MASK_ISA_VAES_UNSET): New.
2520 (ix86_handle_option): Handle -mvaes.
2521 * config/i386/cpuid.h: Define bit_VAES.
2522 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2523 * config/i386/i386-c.c (__VAES__): New.
2524 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2525 (ix86_valid_target_attribute_inner_p): Ditto.
2526 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2527 * config/i386/i386.opt: Add -mvaes.
2528 * doc/invoke.texi: Ditto.
2530 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2532 * debug.h (gcc_debug_hooks): Add inline_entry.
2533 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2534 * debug.c (do_nothing_debug_hooks): Likewise.
2535 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2536 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2537 (dwarf2_lineno_debug_hooks): Likewise.
2539 * common.opt (gstatement-frontiers): New, setting
2540 debug_nonbind_markers_p.
2541 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2542 * toplev.c (process_options): Autodetect value for debug statement
2544 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2545 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2547 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2548 markers. Integrate source bind into debug stmt expand loop.
2549 (pass_expand::execute): Check debug marker limit. Avoid deep
2550 TER and expand debug locations for debug bind insns only.
2551 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2553 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2554 * final.c (reemit_insn_block_notes): Take current block from
2555 nonbind markers. Declare note where it's first set.
2556 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2557 begin stmt markers if enabled.
2558 (notice_source_line): Handle nonbind markers. Fail if their
2559 location is unknown or that of builtins.
2560 (rest_of_handle_final): Convert begin stmt markers to notes if
2561 var-tracking didn't run.
2562 (rest_of_clean_state): Skip begin stmt markers.
2563 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2565 * function.c (allocate_struct_function): Set begin_stmt_markers.
2566 * function.h (struct function): Add debug_marker_count counter
2567 and debug_nonbind_markers flag.
2568 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2569 * gimple-low.c (lower_function_body): Adjust
2570 debug_nonbind_markers.
2571 (lower_stmt): Drop or skip gimple debug stmts.
2572 (lower_try_catch): Skip debug stmts.
2573 * gimple.c (gimple_build_debug_begin_stmt): New.
2574 (gimple_copy): Increment debug_marker_count if copying one.
2575 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2576 * gimplify.c (rexpr_location): New.
2577 (rexpr_has_location): New.
2578 (warn_switch_unreachable_r): Handle gimple debug stmts.
2579 (shortcut_cond_r): Call expr_location.
2581 (find_goto_label): New.
2582 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2584 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2586 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2587 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2588 (LANG_HOOKS_INITIALIZER): ... this.
2589 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2590 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2592 (update_ebb_live_info): Skip debug insn markers.
2593 * lra.c (debug_insn_static_data): Rename to...
2594 (debug_bind_static_data): ... this.
2595 (debug_marker_static_data): New.
2596 (lra_set_insn_recog_data): Select one of the above depending
2598 (lra_update_isn_regno_info): Don't assume debug insns have
2600 (push_insns): Skip debug insns.
2601 * lto-streamer-in.c (input_function): Drop debug stmts
2602 depending on active options. Adjust debug_nonbind_markers.
2603 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2604 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2605 begin stmt marker notes.
2606 (print_insn): Likewise.
2607 * recog.c (extract_insn): Recognize rtl for debug markers.
2608 * rtl.def (DEBUG_MARKER): New.
2609 * tree-inline.c: Include params.h.
2610 (remap_gimple_stmt): Handle nonbind markers.
2611 (maybe_move_debug_stmts_to_successors): Likewise.
2612 (copy_debug_stmt): Likewise.
2613 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2614 markers regardless of no side effects.
2615 (tsi_link_before): Don't update container's side effects when adding
2616 a begin stmt marker.
2617 (tsi_link_after): Likewise.
2618 (expr_first): Skip begin stmt markers.
2619 (expr_last): Likewise.
2620 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2621 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2622 Disregard nonbind markers.
2623 * tree.c (make_node_stat): Don't set side effects for begin stmt
2625 (build1_stat): Likewise.
2626 * tree.def (DEBUG_BEGIN_STMT): New.
2627 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2628 * var-tracking.c (delete_debug_insns): Renamed to...
2629 (delete_vta_debug_insns): ... this.
2630 (reemit_marker_as_note): New.
2631 (vt_initialize): Reemit markers.
2632 (delete_vta_debug_insns): Likewise.
2633 (vt_debug_insns_local): Reemit or delete markers.
2634 (variable_tracking_main_1): Likewise.
2635 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2636 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2637 (gimple_debug_nonbind_marker_p): New.
2638 (gimple_build_debug_bind): Adjust.
2639 (gimple_build_debug_begin_stmt): New.
2640 * doc/invoke.texi (max-debug-marker-count): New param.
2641 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2642 (debug_parameter_ref, debug_marker): New.
2643 (NOTE_INSN_BEGIN_STMT): New.
2644 (DEBUG_INSN): Describe begin stmt markers.
2646 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2647 without debug insns, we wouldn't, but clean up debug insns
2648 after a control flow insn nevertheless.
2650 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2651 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2652 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2653 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2654 (rtl_tidy_fallthru_edge): Likewise.
2655 (rtl_verify_fallthru): Likewise.
2656 (rtl_verify_bb_layout): Likewise.
2657 (skip_insns_after_block): Likewise.
2658 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2659 * dwarf2out.c: Include print-rtl.h.
2660 (dwarf2out_next_real_insn): New.
2661 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2662 Dump debug binds in asm comments.
2663 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2664 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2665 callers to use gsi_start_nondebug_bb instead.
2666 (gsi_after_labels): Skip gimple debug stmts.
2667 (gsi_start_nondebug): New.
2668 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2669 (proper_loop_form_for_interchange): Adjust.
2670 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2671 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2672 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2673 (collect_fallthrough_labels): Likewise.
2674 (should_warn_for_implicit_fallthrough): Likewise.
2675 (warn_implicit_fallthrough_r): Likewise.
2676 (expand_FALLTHROUGH_r): Likewise.
2677 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2678 (graphite_copy_stmts_from_block): Skip nonbind markers.
2679 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2680 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2681 * jump.c (clean_barriers): Skip debug insns.
2682 * omp-expand.c (expand_parallel_call): Skip debug insns.
2683 (expand_task_call): Likewise.
2684 (remove_exit_barrier): Likewise.
2685 (expand_omp_taskreg): Likewise.
2686 (expand_omp_for_init_counts): Likewise.
2687 (expand_omp_for_generic): Likewise.
2688 (expand_omp_for_static_nochunk): Likewise.
2689 (expand_omp_for_static_chunk): Likewise.
2690 (expand_omp_simd): Likewise.
2691 (expand_omp_taskloop_for_outer): Likewise.
2692 (expand_omp_taskloop_for_inner): Likewise.
2693 (expand_oacc_for): Likewise.
2694 (expand_omp_sections): Likewise.
2695 (expand_omp_single): Likewise.
2696 (expand_omp_synch): Likewise.
2697 (expand_omp_atomic_load): Likewise.
2698 (expand_omp_atomic_store): Likewise.
2699 (expand_omp_atomic_fetch_op): Likewise.
2700 (expand_omp_atomic_pipeline): Likewise.
2701 (expand_omp_atomic_mutex): Likewise.
2702 (expand_omp_target): Likewise.
2703 (grid_expand_omp_for_loop): Likewise.
2704 (grid_expand_target_grid_body): Likewise.
2705 (build_omp_regions_1): Likewise.
2706 * omp-low.c (check_combined_parallel): Skip debug stmts.
2707 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2708 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2709 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2711 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2712 (make_edges): Likewise.
2713 (cleanup_dead_labels): Likewise.
2714 (gimple_can_merge_blocks_p): Likewise.
2715 (stmt_starts_bb_p): Likewise.
2716 (gimple_block_label): Likewise.
2717 (gimple_redirect_edge_and_branch): Likewise.
2718 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2720 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2722 * tree-pretty-print (print_declaration): Omit initializer in slim
2724 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2726 (eliminate_unnecessary_stmts): Stabilize block removal order.
2727 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2728 * var-tracking.c (get_first_insn): New.
2729 (vt_emit_notes): Call it.
2730 (vt_initialize): Walk any insns before the first BB.
2731 (delete_debug_insns): Likewise.
2733 * gimple.h (enum gimple_debug_subcode): Add
2734 GIMPLE_DEBUG_BEGIN_STMT.
2735 (gimple_debug_begin_stmt_p): New.
2736 (gimple_debug_nonbind_marker_p): New.
2737 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2738 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2739 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2740 * insn-notes.def (BEGIN_STMT): New.
2741 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2742 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2743 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2744 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2745 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2746 (INSN_DEBUG_MARKER_KIND): New.
2747 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2748 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2749 (INSN_VAR_LOCATION_PTR): New.
2750 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2752 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2753 deps only if debug bind insns are enabled.
2754 (pass_expand::execute): Avoid deep TER and expand
2755 debug locations for debug bind insns only.
2756 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2757 debug stmts special handling down to debug bind stmts.
2758 * combine.c (try_combine): Narrow debug insns special handling
2759 down to debug bind insns.
2760 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2761 Narrow debug insns preexisting special handling down to debug
2763 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2764 handling down to debug bind insns.
2765 * function.c (instantiate_virtual_regs): Skip debug markers,
2766 adjust handling of debug binds.
2767 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2768 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2769 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2770 insns to debug bind insns.
2771 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2772 special handling of debug stmts to debug bind stmts.
2773 * ipa-split.c (split_function): Likewise.
2774 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2775 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2777 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2778 * regrename.c (build_def_use): Likewise.
2779 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2780 (pass_cprop_hardreg): Narrow special casing of debug insns to
2782 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2783 * reload1.c (reload): Likewise.
2784 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2785 casing of debug stmts to debug bind stmts.
2786 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2787 * ssa-iterators.h (num_imm_uses): Likewise.
2788 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2789 debug stmts to debug bind stmts.
2790 * tree-inline.c (tree_function_versioning): Narrow special casing
2791 of debug stmts to debug bind stmts.
2792 * tree-loop-distribution.c (generate_loops_for_partition):
2793 Narrow special casing of debug stmts to debug bind stmts.
2794 * tree-sra.c (analyze_access_subtree): Narrow special casing
2795 of debug stmts to debug bind stmts.
2796 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2797 stmts to debug bind stmts.
2798 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2799 casing of debug stmts to debug bind stmts.
2800 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2801 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2802 casing of debug stmts to debug bind stmts.
2803 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2805 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2806 debug stmts to debug bind stmts.
2807 (gimple_replace_ssa_lhs): Likewise.
2808 (insert_debug_temp_for_var_def): Likewise.
2809 (insert_debug_temps_for_defs): Likewise.
2810 (reset_debug_uses): Likewise.
2811 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2812 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2813 (adjust_debug_stmts): Likewise.
2814 (adjust_phi_and_debug_stmts): Likewise.
2815 (vect_do_peeling): Likewise.
2816 * tree-vect-loop.c (vect_transform_loop): Likewise.
2817 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2818 * var-tracking.c (adjust_mems): Narrow special casing of debug
2819 insns to debug bind insns.
2820 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2821 (compute_bb_dataflow, vt_find_locations): Likewise.
2822 (vt_expand_loc, emit_notes_for_changes): Likewise.
2823 (vt_init_cfa_base): Likewise.
2824 (vt_emit_notes): Likewise.
2825 (vt_initialize): Likewise.
2826 (vt_finalize): Likewise.
2828 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2829 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2830 (next_nonnote_nondebug_insn_bb): New.
2831 (prev_nonnote_nondebug_insn_bb): New.
2832 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2833 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2834 (prev_nonnote_nondebug_insn_bb): Declare.
2835 (next_nonnote_nondebug_insn_bb): Declare.
2836 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2837 * cfgrtl.c (get_last_bb_insn): Likewise.
2838 * lra.c (push_insns): Likewise.
2840 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2843 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2845 * selftest.c (selftest::test_runner::test_runner): New ctor.
2846 (selftest::test_runner::~test_runner): New dtor.
2847 * selftest.h (class selftest::test_runner): New class.
2849 2017-12-11 Carl Love <cel@us.ibm.com>
2851 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
2852 vec_extract_fp32_from_shortl]): Add #defines.
2853 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
2854 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
2855 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
2856 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
2857 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
2858 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
2859 * doc/extend.texi: Add documentation for the added builtins.
2861 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
2863 PR rtl-optimization/80693
2864 PR rtl-optimization/81019
2865 PR rtl-optimization/81020
2866 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
2867 are not mentioned in i3. Place the REG_UNUSED note on i2,
2868 possibly modified to REG_DEAD, if it did not originate in i3.
2870 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2872 * recog.c (store_data_bypass_p_1): New function.
2873 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
2874 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
2876 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
2878 PR rtl-optimization/83361
2879 * ifcvt.c (if_convert): Call fixup_partitions.
2881 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
2883 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2884 early folding of splat_u{8,16,32}.
2886 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2888 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
2889 output_operand_lossage first argument with capital letter.
2890 (aarch64_override_options): Don't start error and sorry first argument
2891 with capital letter.
2893 2017-12-11 Andi Kleen <ak@linux.intel.com>
2895 PR gcov-profile/83355
2896 * auto-profile.c (string_table::get_index_by_decl): Don't
2897 recurse when abstract origin points to itself.
2899 2017-12-11 Bin Cheng <bin.cheng@arm.com>
2901 PR tree-optimization/83320
2902 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
2903 (prune_datarefs_not_in_loop): Ditto.
2905 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
2907 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
2909 2017-12-10 Jakub Jelinek <jakub@redhat.com>
2911 PR tree-optimization/83337
2912 * gimple-loop-interchange.cc (compute_access_stride): Handle
2913 bitfield DRs properly.
2915 2017-12-09 Jakub Jelinek <jakub@redhat.com>
2917 PR tree-optimization/83338
2918 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
2919 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
2920 vector of unsigned integers to vector of signed integers.
2922 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
2924 PR rtl-optimization/83317
2925 * lra-constraints.c (process_address_1): Add insn code check.
2927 2017-12-08 Michael Matz <matz@suse.de>
2929 Fix PR tree-optimization/83323
2930 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
2931 head-controlled loops and loop BBs.
2932 * common.opt (funroll-and-jam): Remove, instead ...
2933 (floop-unroll-and-jam): ... reuse this option.
2934 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
2935 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
2936 (-floop-unroll-and-jam): ... this option.
2938 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2940 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
2941 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
2942 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
2944 2017-12-08 Richard Biener <rguenther@suse.de>
2947 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
2948 handle accesses outside of zero-sized vars.
2950 2017-12-08 Martin Jambor <mjambor@suse.cz>
2952 PR tree-optimization/83141
2953 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
2954 test for MEM_REFs implicitely changing types with padding. Remove
2956 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
2958 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2960 * config/arc/arc.c (arc_attribute_table): Add exclusions to
2962 * config/avr/avr.c (avr_attribute_table): Likewise.
2963 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2964 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2965 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2966 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
2967 of struct attribute_spec.
2968 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2970 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
2973 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
2975 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
2978 * profile-count.c (profile_count::from_gcov_type): Move from
2979 profile-count.h; handle overflow.
2980 * profile-count.h (profile_count::from_gcov_type): Move offline.
2982 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
2984 PR rtl-optimization/83304
2985 * combine.c (move_deaths): If we do not know where a register died,
2988 2017-12-08 Richard Biener <rguenther@suse.de>
2990 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2991 Provide -fopt-info-loop feedback when we interchange in a nest.
2993 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2995 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
2996 for armv6 ARM CPU IDs.
2998 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3000 * common/config/arm/arm-common.c: Include <algorithm>.
3001 (INCLUDE_VECTOR): Define.
3002 (compare_opt_names): New function.
3003 (arm_rewrite_selected_arch): Only strip out extensions that can be
3004 expressed through -mfpu. Sort the remaining extensions
3007 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3009 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3010 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3011 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3012 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3014 2017-06-08 Tristan Gingold <gindold@adacore.com>
3017 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3018 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3020 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3022 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3023 of struct attribute_spec.
3025 2017-12-08 Julia Koval <julia.koval@intel.com>
3027 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3028 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3029 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3030 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3031 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3032 _mm_maskz_dpwssds_epi32): Ditto.
3034 2017-12-08 Richard Biener <rguenther@suse.de>
3036 PR tree-optimization/81303
3037 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3038 conditions try to create a comparison vector type matching
3039 the data vector type.
3040 (vectorizable_condition): Adjust.
3041 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3042 Leave invariant conditions alone in case we can vectorize those.
3044 2017-12-08 Julia Koval <julia.koval@intel.com>
3046 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3047 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3048 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3049 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3050 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3052 2017-12-08 Julia Koval <julia.koval@intel.com>
3054 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3055 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3056 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3057 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3058 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3059 _mm_maskz_dpbusds_epi32): New intrinsics.
3061 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3063 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3065 (ld<bh>io_signed): Likewise.
3066 (st<bhw>io): Likewise.
3067 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3068 SMALL_INT12 constant integer operand.
3070 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3072 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3074 (nios2_large_constant_allowed): ...this. Adjust uses.
3075 (nios2_plus_symbolic_constant_p): Rename to...
3076 (nios2_plus_large_constant_p): ...this. Adjust uses.
3077 (nios2_legitimate_address_p): Correct CONST_INT handling.
3078 (nios2_symbolic_memory_operand_p): Rename to...
3079 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3080 (nios2_large_constant_p): Check for large constant integers too.
3081 (nios2_split_large_constant): Handle constant integers.
3082 (nios2_split_symbolic_memory_operand): Rename to...
3083 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3084 (nios2_legitimize_constant_address): Handle constant integers.
3085 (r0rel_constant_p): Handle small constant integers.
3086 (nios2_print_operand_address): Handle r0-relative integer addresses.
3087 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3088 * config/nios2/nios2.md: Adjust for renamed functions.
3090 2017-12-07 Andrew Waterman <andrew@sifive.com>
3092 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3093 (TARGET_HAVE_SRODATA_SECTION): New define.
3094 (riscv_select_section): New function.
3096 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3097 Alexander Monakov <amonakov@ispras.ru>
3098 Jakub Jelinek <jakub@redhat.com>
3101 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3103 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3105 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3106 a bool success value. Don't call output_operand_lossage here.
3107 (aarch64_print_ldpstp_address): Return a bool success value.
3108 (aarch64_print_operand_address): Call output_addr_const if
3109 aarch64_print_address_internal fails.
3110 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3111 'y'; call output_operand_lossage instead. Call output_operand_lossage
3112 if aarch64_print_ldpstp_address fails.
3114 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3116 * tree-vector-builder.h
3117 (tree_vector_builder::binary_encoded_nelts): Declare.
3118 * tree-vector-builder.c
3119 (tree_vector_builder::binary_encoded_nelts): New function.
3120 * fold-const.c (negate_expr_p): Likewise.
3121 (operand_equal_p, fold_checksum_tree): Likewise.
3122 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3123 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3124 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3125 (uniform_vector_p): Likewise.
3126 * varasm.c (const_hash_1, compare_constant): Likewise.
3127 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3128 (valid_lattice_transition): Operate directly on the VECTOR_CST
3130 * ipa-icf.c: Include tree-vector-builder.h.
3131 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3132 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3134 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3136 * tree.c (build_vector): Delete.
3137 * tree.h (build_vector): Make static and move into the self-testing
3140 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3142 * vector-builder.h (vector_builder::derived): New const overload.
3143 (vector_builder::elt): New function.
3144 * tree-vector-builder.h (tree_vector_builder::type): New function.
3145 (tree_vector_builder::apply_step): Declare.
3146 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3148 * gimple-fold.h (tree_vector_builder): Declare.
3149 (gimple_build_vector): Take a tree_vector_builder instead of a
3150 type and vector of elements.
3151 * gimple-fold.c (gimple_build_vector): Likewise.
3152 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3154 (get_initial_defs_for_reduction): Likewise.
3155 (vectorizable_induction): Likewise.
3157 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3159 * tree-vector-builder.h
3160 (tree_vector_builder::new_binary_operation): Declare.
3161 * tree-vector-builder.c
3162 (tree_vector_builder::new_binary_operation): New function.
3163 * fold-const.c (fold_relational_const): Use it.
3164 (const_binop): Likewise. Check that both input vectors have
3165 the same number of elements, thus excluding things like WIDEN_SUM.
3166 Check whether it is possible to operate directly on the encodings
3169 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3171 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3172 new_unary_operation, operating only on the encoded elements.
3173 (const_unop): Likewise.
3174 (exact_inverse): Likewise.
3175 (distributes_over_addition_p): New function.
3176 (const_binop): Use tree_vector_builder and new_unary_operation
3177 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3178 on the encoded elements unless the encoding is strided and the
3179 operation does not distribute over addition.
3180 (fold_convert_const): Use tree_vector_builder and
3181 new_unary_operation. Operate only on the encoded elements
3182 for truncating integer conversions, or for non-stepped encodings.
3184 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3186 * config/sparc/sparc.c: Include tree-vector-builder.h.
3187 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3188 * expmed.c: Include tree-vector-builder.h.
3189 (make_tree): Use tree_vector_builder instead of build_vector.
3190 * fold-const.c: Include tree-vector-builder.h.
3191 (const_binop): Use tree_vector_builder instead of build_vector.
3192 (const_unop): Likewise.
3193 (native_interpret_vector): Likewise.
3194 (fold_vec_perm): Likewise.
3195 (fold_ternary_loc): Likewise.
3196 * gimple-fold.c: Include tree-vector-builder.h.
3197 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3199 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3200 (simplify_vector_constructor): Use tree_vector_builder instead
3202 * tree-vect-generic.c: Include tree-vector-builder.h.
3203 (add_rshift): Use tree_vector_builder instead of build_vector.
3204 (expand_vector_divmod): Likewise.
3205 (optimize_vector_constructor): Likewise.
3206 * tree-vect-loop.c: Include tree-vector-builder.h.
3207 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3208 of build_vector. Explicitly use a stepped encoding for
3210 * tree-vect-slp.c: Include tree-vector-builder.h.
3211 (vect_get_constant_vectors): Use tree_vector_builder instead
3213 (vect_transform_slp_perm_load): Likewise.
3214 (vect_schedule_slp_instance): Likewise.
3215 * tree-vect-stmts.c: Include tree-vector-builder.h.
3216 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3217 (vect_gen_perm_mask_any): Likewise.
3218 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3219 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3221 (build_vector_from_val): Likewise. Explicitly use a duplicate
3224 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3226 * doc/generic.texi (VECTOR_CST): Describe new representation of
3228 * vector-builder.h: New file.
3229 * tree-vector-builder.h: Likewise.
3230 * tree-vector-builder.c: Likewise.
3231 * Makefile.in (OBJS): Add tree-vector-builder.o.
3232 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3233 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3234 (tree_vector): Change the number of elements to
3235 vector_cst_encoded_nelts.
3236 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3237 (VECTOR_CST_ELTS): Delete.
3238 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3239 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3240 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3241 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3242 (VECTOR_CST_ENCODED_ELT): Likewise.
3243 (vector_cst_encoded_nelts): New function.
3244 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3245 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3246 (vector_cst_int_elt, vector_cst_elt): Declare.
3247 * tree.c: Include tree-vector-builder.h.
3248 (tree_code_size): Abort if passed VECTOR_CST.
3249 (tree_size): Update for new VECTOR_CST layout.
3250 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3251 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3252 (build_vector): Use tree_vector_builder.
3253 (vector_cst_int_elt, vector_cst_elt): New functions.
3254 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3255 encoded elements and then create the vector in the canonical form.
3256 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3257 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3258 (tree_c_tests): Call test_vector_cst_patterns.
3259 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3261 (hash_tree): Likewise.
3262 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3263 (streamer_write_tree_header): Likewise.
3264 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3265 (streamer_alloc_tree): Likewise. Update call to make_vector.
3266 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3268 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3270 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3271 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3272 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3274 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3275 Richard Biener <rguenther@suse.de>
3277 PR tree-optimization/81303
3278 * Makefile.in (gimple-loop-interchange.o): New object file.
3279 * common.opt (floop-interchange): Reuse the option from graphite.
3280 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3281 -floop-interchange and mention it for -O3.
3282 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3283 * gimple-loop-interchange.cc: New file.
3284 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3285 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3286 * passes.def (pass_linterchange): New pass.
3287 * timevar.def (TV_LINTERCHANGE): New time var.
3288 * tree-pass.h (make_pass_linterchange): New declaration.
3289 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3290 interchange. Record IV before/after increment in new parameters.
3291 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3292 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3294 (check_reduction_path): ...New function here.
3295 * tree-vectorizer.h (check_reduction_path): New declaration.
3297 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3300 PR rtl-optimization/80818
3301 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3302 always early clobbered.
3303 * lra-lives.c (process_bb_lives): Check input hard regs for early
3304 clobbered non-operand hard reg.
3306 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3309 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3310 types_compatible_p, just that TYPE_MODE is the same.
3312 2017-12-07 Martin Sebor <msebor@redhat.com>
3315 * attribs.c (empty_attribute_table): Initialize new member of
3316 struct attribute_spec.
3317 (decl_attributes): Add argument. Handle mutually exclusive
3318 combinations of attributes.
3319 (selftests::test_attribute_exclusions): New function.
3320 (selftests::attribute_c_tests): Ditto.
3321 * attribs.h (decl_attributes): Add default argument.
3322 * selftest.h (attribute_c_tests): Declare.
3323 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3324 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3326 * doc/extend.texi (Common Function Attributes): Update const and pure.
3327 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3328 of struct attribute_spec.
3329 * config/arc/arc.c (arc_attribute_table): Same.
3330 * config/arm/arm.c (arm_attribute_table): Same.
3331 * config/avr/avr.c ( avr_attribute_table): Same.
3332 * config/bfin/bfin.c (bfin_attribute_table): Same.
3333 * config/cr16/cr16.c (cr16_attribute_table): Same.
3334 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3335 * config/h8300/h8300.c (h8300_attribute_table): Same.
3336 * config/i386/i386.c (ix86_attribute_table): Same.
3337 * config/ia64/ia64.c (ia64_attribute_table): Same.
3338 * config/m32c/m32c.c (m32c_attribute_table): Same.
3339 * config/m32r/m32r.c (m32r_attribute_table): Same.
3340 * config/m68k/m68k.c (m68k_attribute_table): Same.
3341 * config/mcore/mcore.c (mcore_attribute_table): Same.
3342 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3343 * config/mips/mips.c (mips_attribute_table): Same.
3344 * config/msp430/msp430.c (msp430_attribute_table): Same.
3345 * config/nds32/nds32.c (nds32_attribute_table): Same.
3346 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3347 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3348 * config/rl78/rl78.c (rl78__attribute_table): Same.
3349 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3350 * onfig/rx/rx.c (rx_attribute_table): Same.
3351 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3352 * config/sh/sh.c (sh_attribute_table): Same.
3353 * config/sparc/sparc.c (sparc_attribute_table): Same.
3354 * config/spu/spu.c (spu_attribute_table): Same.
3355 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3356 * config/v850/v850.c (v850_attribute_table): Same.
3357 * config/visium/visium.c (visium_attribute_table): Same.
3359 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3362 * config/arm/arm.c (INCLUDE_STRING): Define.
3363 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3364 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3367 2017-12-07 Michael Matz <matz@suse.de>
3369 Add unroll and jam pass
3371 * gimple-loop-jam.c: New file.
3372 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3373 * common.opt (funroll-and-jam): New option.
3374 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3375 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3376 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3377 * passes.def: Add pass_loop_jam.
3378 * timevar.def (TV_LOOP_JAM): Add.
3379 * tree-pass.h (make_pass_loop_jam): Declare.
3380 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3381 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3382 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3383 to flow_loop_tree_node_add.
3384 (duplicate_subloops, copy_loops_to): Append to sibling list.
3385 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3386 * doc/invoke.texi (-funroll-and-jam): Document new option.
3387 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3389 2017-12-07 Richard Biener <rguenther@suse.de>
3391 PR tree-optimization/83296
3392 PR tree-optimization/67769
3393 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3394 flow sensitive info in an unrelated BB.
3395 (value_replacement): Use reset_flow_sensitive_info.
3396 (minmax_replacement): Reset flow sensitive info on the def
3397 we move. Do not reset flow sensitive info in the whole BB
3398 we move the stmt to.
3399 (abs_replacement): Likewise.
3401 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3404 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3405 rs6000_cpu to the given -mcpu=, or to the default processor.
3407 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3409 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3410 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3411 instead of rs6000_cpu_attr.
3412 (rs6000_adjust_cost): Ditto.
3413 (is_microcoded_insn): Ditto.
3414 (rs6000_adjust_priority): Ditto.
3415 (rs6000_issue_rate): Ditto.
3416 (rs6000_use_sched_lookahead): Ditto.
3417 (rs6000_use_sched_lookahead_guard): Ditto.
3418 (rs6000_sched_reorder): Ditto.
3419 (force_new_group): Ditto.
3420 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3421 (group_ending_nop): Ditto.
3423 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3425 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3426 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3427 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3428 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3430 (rs6000_reassociation_width): Ditto.
3431 (rs6000_emit_epilogue): Ditto.
3432 (rs6000_adjust_cost): Ditto.
3433 (is_microcoded_insn): Ditto.
3434 (is_cracked_insn): Ditto.
3435 (rs6000_adjust_priority): Ditto.
3436 (rs6000_sched_reorder): Ditto.
3437 (rs6000_sched_reorder2): Ditto.
3438 (insn_must_be_first_in_group): Ditto.
3439 (insn_must_be_last_in_group): Ditto.
3440 (rs6000_register_move_cost): Ditto.
3441 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3444 2017-12-07 Julia Koval <julia.koval@intel.com>
3446 * config.gcc: Add vaesintrin.h.
3447 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3449 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3450 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3452 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3453 * config/i386/immintrin.h: Include vaesintrin.h.
3454 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3455 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3456 _mm_aesdec_epi128): New intrinsics.
3458 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3460 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3461 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3462 c-family/c-spellcheck.cc.
3463 (best_macro_match::best_macro_match): Likewise.
3465 (struct edit_distance_traits<cpp_hashnode *>): Move to
3466 c-family/c-spellcheck.h.
3467 (class best_macro_match): Likewise.
3469 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3471 PR tree-optimization/83293
3472 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3473 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3474 might insert into empty bb.
3477 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3478 simplify for plus with :c added, and pointer_plus without that.
3479 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3480 with undefined overflow and the conversion is not widening,
3481 perform negation in utype and only convert to type afterwards.
3482 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3483 simplify for plus with :c added, and pointer_plus without that.
3484 If type is integral with undefined overflow and the conversion is
3485 not widening, perform minus in utype and only convert to type
3486 afterwards. Move the last pointer_diff_expr simplify into the
3489 2017-12-06 Martin Sebor <msebor@redhat.com>
3491 PR tree-optimization/82646
3492 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3493 strncpy, not maxlen.
3495 2017-12-06 Martin Sebor <msebor@redhat.com>
3497 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3500 PR tree-optimization/83075
3501 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3502 strncat/strncpy don't change length of source string.
3504 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3507 2017-11-29 Martin Aberg <maberg@gaisler.com>
3509 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3510 to prevent b2bst errata sequence.
3511 (sqrtdf2_fix): Likewise.
3513 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3515 PR tree-optimization/81945
3516 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3517 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3518 to dest_cfun has orig_loop_num set, either remap it to the new
3519 loop number if the loop got moved too, or clear it.
3521 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3523 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3525 (thunderx2t99_nothing): New insn reservation.
3526 (thunderx2t99_mrs): New insn reservation.
3527 (thunderx2t99_multiple): New insn reservation.
3528 (thunderx2t99_alu_basi): Add bfx to reservation.
3529 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3531 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3534 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3536 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3538 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3539 tree-ssa-pre.c::remove_dead_inserted_code.
3540 * tree-ssa-dce.h: New file.
3541 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3542 (remove_dead_inserted_code): Move and rename to function
3543 tree-ssa-dce.c::simple_dce_from_worklist.
3544 (pass_pre::execute): Update use.
3546 2017-12-05 Richard Biener <rguenther@suse.de>
3548 PR tree-optimization/83277
3549 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3550 to code-gen liveout vars.
3552 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3554 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3555 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3556 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3559 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3561 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3563 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3564 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3567 2017-12-05 Richard Biener <rguenther@suse.de>
3569 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3570 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3571 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3572 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3574 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3576 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3578 2017-12-05 Martin Liska <mliska@suse.cz>
3579 Jakub Jelinek <jakub@redhat.com>
3581 * doc/invoke.texi: Document the options.
3582 * flag-types.h (enum sanitize_code): Add
3583 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3584 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3585 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3586 * opts.c: Define new sanitizer options.
3587 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3588 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3590 2017-12-05 Julia Koval <julia.koval@intel.com>
3592 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3593 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3594 (ix86_handle_option): Handle -mavx512vnni.
3595 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3596 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3597 * config/i386/i386-c (__AVX512VNNI__): New.
3598 * config/i386/i386.c (ix86_target_string): Handle new option.
3599 (ix86_valid_target_attribute_inner_p): Handle new option.
3600 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3601 * config/i386/i386.opt (mavx512vnni): New option.
3603 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3606 * athlon.md: Disable for generic.
3607 * haswell.md: Enable for generic.
3608 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3609 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3611 (ix86_adjust_cost): Move generic to haswell path.
3613 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3615 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3616 instead of MEM_P in a couple more places. Fix formatting issues.
3618 2017-12-04 Jim Wilson <jimw@sifive.com>
3620 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3621 instead of GP_REG_LAST-1.
3622 (riscv_adjust_libcall_cfi_prologue): Likewise.
3623 (riscv_adjust_libcall_cri_epilogue): Likewise.
3624 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3627 2017-12-04 Luis Machado <luis.machado@linaro.org>
3629 * ipa-pure-const.c (check_decl): Add missing newline.
3630 (state_from_flags): Likewise.
3632 2017-12-04 Jeff Law <law@redhat.com>
3634 PR tree-optimizatin/78496
3635 * gimple-ssa-evrp-analyze.h
3636 (evrp_range_analyzer::get_vr_values): Simplify.
3637 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3638 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3639 and gimple-ssa-evrp-analyze.h.
3640 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3641 (simplify_stmt_for_jump_threading): Copy a blob of code from
3642 tree-vrp.c to use ranges to simplify statements.
3643 (dom_opt_dom_walker::before_dom_children): Call
3644 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3645 (dom_opt_dom_walker::after_dom_children): Similarly for
3646 evrp_range_analyzer::leave.
3647 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3650 * gimple-ssa-evrp-analyze.c
3651 (evrp_range_analyzer::extract_range_from_stmt): Always use
3652 vr_values::update_value_range so preexisting range info is
3653 medged with new range info, even if the new range is VR_VARYING.
3655 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3657 * combine.c: Adjust comment.
3658 (use_crosses_set_p): Delete.
3659 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3660 (try_combine): Ditto.
3662 2017-12-04 Richard Biener <rguenther@suse.de>
3664 PR tree-optimization/83255
3665 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3666 Re-add zero-iteration check.
3668 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3670 PR rtl-optimization/83245
3671 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3672 hard registers as earlyclobber, also if not in an asm.
3674 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3678 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3681 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3682 rs6000_cpu based on cpu_index, not tune_index.
3684 2017-12-04 Richard Biener <rguenther@suse.de>
3686 PR tree-optimization/83238
3687 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3688 code match comment, rejecting invalid SESE regions.
3690 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3692 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3693 require base operand is a REG_POINTER prior to reload on targets
3694 with non-equivalent space registers.
3696 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3698 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3699 (update_specialized_profile): Fix updating of counts.
3700 (perhaps_add_new_callers): Likewise.
3702 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3705 * x86-tune.def: Remove obsolette FIXMEs.
3706 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3707 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3708 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3710 (X86_TUNE_PAD_RETURNS): Disable for generic.
3712 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3714 PR tree-optimization/83170
3715 PR tree-optimization/83241
3716 * gimple-ssa-store-merging.c
3717 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3718 gimple_vuse (ins_stmt) in case it has changed.
3719 (imm_store_chain_info::output_merged_store): Likewise.
3721 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3725 * tree-cfg.c (pass_warn_function_return::execute): Handle
3726 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3727 with BUILTINS_LOCATION.
3731 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3732 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3733 instead of DECL_MODE.
3735 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3736 last argument to rtx pointer.
3737 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3738 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3739 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3740 depending on the chosen ISAs.
3741 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3742 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3743 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3745 * config/i386/sse.md (mov<mode>_internal): Likewise.
3746 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3748 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3750 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3752 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3754 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3755 parameters from prototype.
3756 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3757 parameters. Don't print a comment.
3758 (emit_fusion_gpr_load): Adjust.
3759 (emit_fusion_load_store): Adjust.
3760 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3761 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3762 comment on the second line.
3764 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3767 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3768 rs6000_cpu based on cpu_index, not tune_index.
3770 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3772 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3773 which_alternative instead of which_alternative + 1.
3774 (output_asm_insn): Print an extra tab if the template is short.
3776 2017-12-01 Jim Wilson <jimw@sifive.com>
3778 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3780 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3781 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3783 * doc/tm.texi: Regenerate.
3785 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3788 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3789 whether we can allocate pseudos before trying to fix an address.
3790 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3791 memory address is indexed or indirect.
3792 (floatuns_<mode>si2_hw2): Likewise.
3794 2017-12-01 Jason Merrill <jason@redhat.com>
3796 * Makefile.in (TAGS): Add c-family/*.cc.
3798 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3800 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3801 (call_value_insn): Likewise.
3802 (sibcall_insn): Likewise.
3803 (sibcall_value_insn): Likewise.
3804 (movsi_aarch64): Likewise.
3805 (movdi_aarch64): Likewise.
3806 (add_losym_): Likewise.
3807 (ldr_got_small_): Likewise.
3808 (ldr_got_small_sidi): Likewise.
3809 (ldr_got_small_28k_): Likewise.
3810 (ldr_got_small_28k_sidi): Likewise.
3811 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3812 Move output_addr_const to symbolic case. Add error check.
3814 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3816 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3817 * config/s390/s390.md: Disable MVC merging peephole if it would
3818 disable operand forwarding.
3819 (new peephole2): Split MVCs if it would turn them into up to 2
3822 2017-12-01 Richard Biener <rguenther@suse.de>
3824 PR tree-optimization/83232
3825 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3826 detection of same access. Instead of breaking the group here
3827 do not consider the duplicate. Add comment explaining real fix.
3829 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3831 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3833 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3835 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3837 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3839 * function.h (struct function): Remove cilk_frame_decl,
3840 is_cilk_function and calls_cilk_spawn fields.
3841 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3843 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3844 cilk_elemental field.
3845 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3846 * target.def: Adjust comment.
3847 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3848 Don't test cilk_elemental.
3850 PR tree-optimization/83233
3851 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
3852 bswap_stat name for the struct.
3855 * tree.h (SWITCH_BREAK_LABEL_P): Define.
3856 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
3857 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
3858 SWITCH_BREAK_LABEL_P set on the label.
3859 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
3860 added for default case if it was missing and not all cases covered.
3861 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
3862 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
3864 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
3865 to avoid -Wimplicit-fallthrough warning.
3866 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
3867 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
3869 PR tree-optimization/83221
3870 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
3872 (init_reassoc): Formatting fix.
3875 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
3876 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
3880 * tree-cfg.c: Include asan.h.
3881 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
3883 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3885 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
3887 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3889 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
3890 vpcmpeqd instruction.
3892 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3894 * config/i386/i386.c (standard_sse_constant_opcode): Fix
3895 registers type for 128bit mode.
3897 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3899 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
3901 * gengtype-state.c (read_state_pair): Likewise.
3902 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
3903 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
3904 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
3905 * ifcvt.c (noce_try_store_flag_constants): Likewise.
3906 * tree-ssa-ccp.c (ccp_finalize): Likewise.
3907 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
3908 * builtins.c (fold_builtin_3): Likewise.
3909 * graphite-scop-detection.c
3910 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
3911 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
3914 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
3917 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
3918 parameter. Return true if ignore_decl_rtl_set_p is true and other
3919 conditions are satisfied.
3920 * asan.h (asan_protect_global): Add new parameter.
3921 * varasm.c (categorize_decl_for_section): Pass true as second parameter
3922 to asan_protect_global calls.
3924 2017-11-30 Jim Wilson <jimw@sifive.com>
3926 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
3927 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
3928 -mstrict-align, add info on default value. Delete redundant lines for
3929 -mabi. Add missing -mexplicit-relocs docs.
3931 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3933 * config/arc/arc.md (trap): New pattern.
3935 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3937 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
3938 instruction to end into a delay slot.
3939 * config/arc/arc.md (cond_delay_insn): Check if the instruction
3940 can be placed into a delay slot against reg_note.
3942 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3944 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
3945 labels number of usages.
3947 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3949 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
3951 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
3953 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3956 * internal-fn.c (expand_mul_overflow): Optimize unsigned
3957 multiplication by power of 2 constant into two shifts + comparison.
3959 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
3962 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
3964 2017-11-30 Richard Biener <rguenther@suse.de>
3966 PR tree-optimization/83202
3967 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3968 allow_peel argument and guard peeling.
3969 (canonicalize_loop_induction_variables): Likewise.
3970 (canonicalize_induction_variables): Pass false.
3971 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
3972 peeling from cunrolli.
3974 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3976 * combine.c (try_combine): Print a message to dump file whenever
3977 I0, I1, or I2 cannot be combined into I3.
3979 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3981 PR rtl-optimization/83156
3982 PR rtl-optimization/82621
3983 * combine.c (try_combine): Don't split an I2 if one of the dests is
3984 set again before I3. Allow unused dests.
3986 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3988 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
3990 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
3992 PR rtl-optimization/80818
3993 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
3994 recursively. Use insn code for clobber.
3995 (lra_set_insn_recog_data): Pass the new arg to
3996 collect_non_operand_hard_regs.
3997 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
3999 (lra_update_insn_regno_info): Pass insn to
4000 add_regs_to_insn_regno_info.
4002 2017-11-29 Jim Wilson <jimw@sifive.com>
4003 Andrew Waterman <andrew@sifive.com>
4005 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4006 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4007 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4008 use SINGLE_SHIFT_COST.
4009 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4010 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4012 2017-11-29 Julia Koval <julia.koval@intel.com>
4014 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4015 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4016 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4017 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4018 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4019 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4020 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4021 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4022 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4023 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4024 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4025 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4026 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4027 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4028 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4029 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4030 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4031 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4032 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4033 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4034 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4035 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4036 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4037 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4038 __builtin_ia32_vpshldv_v2di_mask,
4039 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4040 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4041 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4043 2017-11-29 Julia Koval <julia.koval@intel.com>
4045 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4046 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4047 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4048 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4049 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4050 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4051 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4052 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4053 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4054 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4055 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4056 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4057 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4058 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4059 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4060 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4061 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4062 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4063 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4064 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4065 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4067 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4069 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4070 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4073 2017-11-29 Martin Aberg <maberg@gaisler.com>
4075 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4076 to prevent b2bst errata sequence.
4077 (sqrtdf2_fix): Likewise.
4079 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4081 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4082 (div_sqrt_insn_p): New function.
4083 (sparc_do_work_around_errata): Insert NOP instructions to
4084 prevent sequences that could trigger the TN-0013 errata for
4085 certain LEON3 processors.
4086 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4087 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4088 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4089 (in_branch_delay): Prevent div and sqrt in delay slot if
4091 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4093 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4095 * config/sparc/sparc.c (atomic_insn_p): New function.
4096 (sparc_do_work_around_errata): Insert NOP instructions to
4097 prevent sequences that could trigger the TN-0010 errata for
4099 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4100 instruction referable in atomic_insns_p.
4102 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4104 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4105 (atomic_compare_and_swap_leon3_1): Likewise.
4108 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4110 * config/sparc/sparc.c (fpop_insn_p): New function.
4111 (sparc_do_work_around_errata): Insert NOP instructions to
4112 prevent sequences that could trigger the TN-0012 errata for
4114 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4115 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4116 (in_branch_annul_delay): Prevent floating-point instructions
4117 in delay slot of annulled integer branch.
4119 2017-11-29 Richard Biener <rguenther@suse.de>
4121 PR tree-optimization/83202
4122 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4124 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4125 nodes vectorized to the same stmts multiple times.
4126 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4127 (vect_analyze_slp_instance): Adjust.
4128 (scalar_stmts_to_slp_tree_map_t): New typedef.
4129 (vect_schedule_slp_instance): Add a map recording the SLP node
4130 representing the vectorized stmts for a set of scalar stmts.
4131 Avoid code-generating redundancies.
4132 (vect_schedule_slp): Allocate map and pass it down.
4134 2017-11-29 Nathan Sidwell <nathan@acm.org>
4137 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4140 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4142 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4143 cases for printing LDP/STP memory addresses.
4144 (aarch64_print_address_internal): Renamed from
4145 aarch64_print_operand_address, added parameter, add Pmode check.
4146 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4147 (aarch64_print_operand_address): Indirect to
4148 aarch64_print_address_internal.
4149 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4150 'y' operand output specifier.
4152 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4155 * tree.c (build_simple_mem_ref_loc): Handle
4156 get_addr_base_and_unit_offset returning a MEM_REF.
4159 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4162 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4165 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4167 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4169 (layout::move_to_column): Likewise.
4171 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4174 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4175 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4177 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4178 Martin Jambor <mjambor@suse.cz>
4181 * tree.h (expr_type_first_operand_type_p): Declare
4182 * tree.c (expr_type_first_operand_type_p): New function.
4183 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4184 (ipa_value_from_jfunc): Adjust declaration.
4185 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4186 Use it as result type for arithmetics, unless it is NULL in which case
4187 be more conservative.
4188 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4189 ipa_get_jf_pass_through_result.
4190 (propagate_vals_across_pass_through): Likewise.
4191 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4192 is to propagate_vals_across_pass_through.
4193 (propagate_constants_across_call): Pass PARM_TYPE to
4194 propagate_scalar_across_jump_function.
4195 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4196 ipa_value_from_jfunc.
4197 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4198 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4199 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4200 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4201 target_type, pass it to ipa_value_from_jfunc.
4202 (update_indirect_edges_after_inlining): Pass parameter type to
4203 try_make_edge_direct_simple_call.
4205 2017-11-28 Jeff Law <law@redhat.com>
4207 * gimple-ssa-evrp-analyze.c
4208 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4209 refine ranges if scev_initialized_p returns true.
4210 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4212 2017-11-28 Julia Koval <julia.koval@intel.com>
4214 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4215 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4216 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4217 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4218 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4219 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4220 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4221 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4222 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4223 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4224 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4225 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4226 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4227 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4228 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4229 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4230 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4231 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4232 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4233 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4234 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4235 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4237 2017-11-28 Julia Koval <julia.koval@intel.com>
4239 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4240 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4241 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4242 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4243 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4244 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4245 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4246 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4247 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4248 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4249 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4250 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4251 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4252 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4253 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4254 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4255 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4256 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4257 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4258 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4259 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4260 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4261 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4262 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4263 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4264 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4265 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4266 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4267 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4268 __builtin_ia32_vpshld_v2di_mask): New builtins.
4269 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4270 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4272 2017-11-28 Richard Biener <rguenther@suse.de>
4274 PR tree-optimization/80776
4275 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4277 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4279 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4280 If the incoming edge is an effective fallthru because the other
4281 edge only reaches a __builtin_unreachable () then record ranges
4282 derived from the controlling condition in SSA info.
4283 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4284 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4286 2017-11-28 Olivier Hainque <hainque@adacore.com>
4288 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4289 on mingw build hosts.
4291 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4294 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4296 2017-11-28 Richard Biener <rguenther@suse.de>
4299 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4300 copies generated from memcpy use a character array as reference
4303 2017-11-28 Julia Koval <julia.koval@intel.com>
4304 Sebastian Peryt <sebastian.peryt@intel.com>
4306 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4307 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4308 cilk-common.o, cilk.h, cilk-common.c): Remove.
4310 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4311 * builtins.c (is_builtin_name): Remove cilkplus condition.
4312 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4313 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4314 cilk-builtins.def, cilkplus.def): Remove.
4315 * cif-code.def (CILK_SPAWN): Remove.
4316 * cilk-builtins.def: Delete.
4317 * cilk-common.c: Ditto.
4319 * cilkplus.def: Ditto.
4320 * config/darwin.h (fcilkplus): Delete.
4321 * cppbuiltin.c: Ditto.
4322 * doc/extend.texi: Remove cilkplus doc.
4323 * doc/generic.texi: Ditto.
4324 * doc/invoke.texi: Ditto.
4325 * doc/passes.texi: Ditto.
4326 * gcc.c (fcilkplus): Remove.
4327 * gengtype.c (cilk.h): Remove.
4328 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4330 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4332 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4334 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4335 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4336 cilkplus conditions.
4337 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4338 inline_read_section): Ditto.
4339 * ipa-inline-analysis.c (cilk.h): Remove.
4340 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4341 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4342 append_linker_options): Remove condition for fcilkplus.
4343 * lto/lto-lang.c (cilk.h): Remove.
4344 (lto_init): Remove condition for fcilkplus.
4345 * omp-expand.c (expand_cilk_for_call): Delete.
4346 (expand_omp_taskreg, expand_omp_for_static_chunk,
4347 expand_omp_for): Remove cilkplus
4349 (expand_cilk_for): Delete.
4350 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4351 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4352 execute_lower_omp, diagnose_sb_0): Ditto.
4353 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4354 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4355 * tree-nested.c: Ditto.
4356 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4357 (dump_generic_node): Ditto.
4358 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4359 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4361 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4363 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4365 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4366 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4367 for vector mode and !TARGET_SIMD.
4369 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4371 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4373 * tree.h (SWITCH_LABELS): Remove.
4374 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4375 assert SWITCH_BODY is non-NULL.
4376 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4378 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4380 PR tree-optimization/80788
4381 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4382 has TREE_OVERFLOW set, call drop_tree_overflow.
4384 2017-11-28 Richard Biener <rguenther@suse.de>
4386 PR tree-optimization/83158
4387 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4389 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4392 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4393 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4395 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4397 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4398 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4400 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4403 * varasm.c (compare_constant): Take the mode of the constants into
4404 account when comparing floating point constants.
4406 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4408 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4409 from explicit instantiation of debug_helper.
4410 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4412 2017-11-27 Richard Biener <rguenther@suse.de>
4414 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4417 2017-11-27 Richard Biener <rguenther@suse.de>
4419 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4420 (type_hash_canon): Also clear int_cst_hash_table entry for
4422 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4424 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4426 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4427 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4428 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4430 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4432 * hash-map.h (gt_cleare_cache): Avoid UB.
4434 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4436 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4437 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4438 (decide_unroll_runtime_iterations): Likewise.
4439 (decide_unroll_stupid): Likewise.
4441 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4444 * config/i386/i386.md: Add a loop with incssp.
4446 2017-11-27 Martin Jambor <mjambor@suse.cz>
4448 PR tree-optimization/81248
4449 * tree-sra.c (splice_param_accesses): Remove size check.
4450 (decide_one_param_reduction): Fix size check.
4451 * gimple-pretty-print.c (dump_profile): Silence warning.
4452 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4454 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4457 * dbxout.c (lastlineno): New variable.
4458 (dbx_debug_hooks): Use dbxout_switch_text_section as
4459 switch_text_section debug hook.
4460 (dbxout_function_end): Switch to current_function_section
4461 rather than function_section. If crtl->has_bb_partition,
4462 output just one N_FUN, depending on in_cold_section_p.
4463 (dbxout_source_line): Remember last lineno in lastlineno.
4464 (dbxout_switch_text_section): New function.
4465 (dbxout_function_decl): Adjust dbxout_block caller.
4466 (dbx_block_with_cold_children): New function.
4467 (dbxout_block): Return true if any LBRAC/RBRAC have been
4468 emitted. Use dbx_block_with_cold_children at depth == 0
4469 in second partition. Add PARENT_BLOCKNUM argument, pass
4470 it optionally adjusted to children. Output LBRAC/RBRAC
4471 around recursive call only if the block is in the current
4472 partition, if not and anything was output, emit empty
4474 * final.c (final_scan_insn): Compute cold_function_name
4475 before calling switch_text_section debug hook. Call
4476 that hook even if dwarf2out_do_frame if not emitting
4480 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4481 TREE_READONLY decls.
4483 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4485 PR rtl-optimization/82488
4486 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4488 2017-11-26 Julia Koval <julia.koval@intel.com>
4490 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4492 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4495 2017-11-26 Julia Koval <julia.koval@intel.com>
4497 * config/i386/driver-i386.c (host_detect_local_cpu):
4498 Detect skylake-avx512.
4500 2017-11-26 Julia Koval <julia.koval@intel.com>
4502 * config.gcc: Add -march=cannonlake.
4503 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4504 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4505 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4506 (PTA_CANNONLAKE): New.
4507 (processor_target_table): Add cannonlake.
4508 (ix86_option_override_internal): Ditto.
4509 (fold_builtin_cpu): Ditto.
4510 (get_builtin_code_for_version): Handle cannonlake.
4511 (M_INTEL_COREI7_CANNONLAKE): New.
4512 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4513 * doc/invoke.texi: Add -march=cannonlake.
4515 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4517 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4518 (try_init_one_plugin): Alternative implementation for MinGW.
4519 * Makefile.in (plugin_implib): New.
4520 (gengtype-lex.c): Fix broken AIX workaround.
4521 * configure: Regenerate.
4522 * doc/plugins.texi: Document support for MinGW.
4524 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4526 PR rtl-optimization/81553
4527 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4528 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4529 is a shift where C1 has different mode than the whole shift, use C1's
4530 mode for MULT rather than the shift's mode.
4533 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4534 builtins not enabled in the currently selected ISA.
4536 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4538 PR tree-optimization/71026
4539 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4540 (insert_reciprocals): Change to insert reciprocals before a division
4541 by a square and to insert the square of a reciprocal.
4542 (execute_cse_reciprocals_1): Change to consider division by a square.
4543 (register_division_in): Add importance parameter.
4545 2017-11-24 Richard Biener <rguenther@suse.de>
4547 PR tree-optimization/82402
4548 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4549 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4551 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4553 * match.pd (0-ptr): New transformation.
4555 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4558 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4559 when skipping checking.
4561 2017-11-24 Richard Biener <rguenther@suse.de>
4563 PR tree-optimization/83128
4564 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4565 (vn_reference_lookup_3): Likewise.
4567 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4570 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4571 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4574 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4575 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4576 check for the latter with gimple_call_builtin_p. Do not handle
4577 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4579 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4581 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4584 2017-11-23 Julia Koval <julia.koval@intel.com>
4586 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4587 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4588 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4589 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4590 _mm512_maskz_expandloadu_epi16): New intrinsics.
4591 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4592 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4593 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4594 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4595 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4596 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4597 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4598 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4599 _mm256_maskz_expandloadu_epi8): New intrinsics.
4600 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4601 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4602 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4603 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4604 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4605 * config/i386/sse.md (VI248_VLBW): New iterator.
4606 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4608 2017-11-23 Julia Koval <julia.koval@intel.com>
4610 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4611 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4612 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4613 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4614 _mm512_mask_compressstoreu_epi16): New.
4615 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4616 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4617 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4618 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4619 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4620 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4621 _mm256_mask_compressstoreu_epi8): New.
4622 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4623 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4624 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4625 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4626 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4627 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4628 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4629 __builtin_ia32_compresshi128_mask,
4630 __builtin_ia32_compressstoreuqi512_mask,
4631 __builtin_ia32_compressstoreuhi512_mask,
4632 __builtin_ia32_compressstoreuqi256_mask,
4633 __builtin_ia32_compressstoreuqi128_mask,
4634 __builtin_ia32_compressstoreuhi256_mask,
4635 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4636 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4638 (ix86_expand_special_args_builtin): Handle new types.
4639 (s4fma_expand): Handle new builtin array.
4640 * config/i386/immintrin.h: Include new headers.
4641 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4642 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4644 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4646 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4647 fixes. Declare temp and g variables at the top in order to avoid
4648 {} in most of the cases.
4650 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4652 * match.pd (ptr-0): New transformation.
4654 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4656 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4657 (struct addr_mode_cost_table): New.
4658 (struct tune_params): Add field addr_mode_costs.
4659 * config/arm/arm.c (generic_addr_mode_costs): New.
4660 (arm_slowmul_tune): Initialise addr_mode_costs field.
4661 (arm_fastmul_tune): Likewise.
4662 (arm_strongarm_tune): Likewise.
4663 (arm_xscale_tune): Likewise.
4664 (arm_9e_tune): Likewise.
4665 (arm_marvell_pj4_tune): Likewise.
4666 (arm_v6t2_tune): Likewise.
4667 (arm_cortex_tune): Likewise.
4668 (arm_cortex_a8_tune): Likewise.
4669 (arm_cortex_a7_tune): Likewise.
4670 (arm_cortex_a15_tune): Likewise.
4671 (arm_cortex_a35_tune): Likewise.
4672 (arm_cortex_a53_tune): Likewise.
4673 (arm_cortex_a57_tune): Likewise.
4674 (arm_exynosm1_tune): Likewise.
4675 (arm_xgene1_tune): Likewise.
4676 (arm_cortex_a5_tune): Likewise.
4677 (arm_cortex_a9_tune): Likewise.
4678 (arm_cortex_a12_tune): Likewise.
4679 (arm_cortex_a73_tune): Likewise.
4680 (arm_v7m_tune): Likewise.
4681 (arm_cortex_m7_tune): Likewise.
4682 (arm_v6m_tune): Likewise.
4683 (arm_fa726te_tune): Likewise.
4684 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4686 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4688 * config/arm/arm.c (arm_mem_costs): New function.
4689 (arm_rtx_costs_internal): Use arm_mem_costs.
4691 2017-11-23 Mark Wielaard <mark@klomp.org>
4693 * dwarf2out.c (init_sections_and_labels): Use generation to create
4694 unique ranges_section_label and ranges_base_label. Return generation.
4695 (output_rnglists): Add generation argument. Use generation to create
4696 unique ranges labels.
4697 (dwarf2out_finish): Get generation from init_sections_and_labels
4698 and pass generation to output_rnglists.
4700 2017-11-23 Mike Stump <mikestump@comcast.net>
4701 Eric Botcazou <ebotcazou@adacore.com>
4703 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4704 * cfgloop.h (struct loop): Add unroll field.
4705 * function.h (struct function): Add has_unroll bitfield.
4706 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4707 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4708 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4709 (pass_rtl_unroll_loops::gate): Likewise.
4710 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4711 for which loop->unroll==1.
4712 (decide_unroll_constant_iterations): Use note for consistency and
4713 take loop->unroll into account. Return early if loop->unroll is set.
4714 Fix thinko in existing test.
4715 (decide_unroll_runtime_iterations): Use note for consistency and
4716 take loop->unroll into account.
4717 (decide_unroll_stupid): Likewise.
4718 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4719 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4720 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4722 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4723 (print_loop): Print loop->unroll if set.
4724 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4725 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4726 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4728 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4729 loop->unroll is set and smaller than the trip count. Otherwise bypass
4730 entirely the heuristics if loop->unroll is set. Remove dead note.
4731 Fix off-by-one bug in other note.
4732 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4733 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4735 (tree_unroll_loops_completely): Make static.
4736 (pass_complete_unroll::execute): Use correct type for variable.
4737 (pass_complete_unrolli::execute): Fix formatting.
4738 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4740 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4742 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4743 enable when TARGET_PREFER_AVX128 is set.
4745 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4747 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4749 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4750 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4752 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4754 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4755 frequencies. Fix estimation of aggregate parameters.
4757 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4759 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4762 2017-11-23 Tom de Vries <tom@codesourcery.com>
4764 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4766 2017-11-23 Tom de Vries <tom@codesourcery.com>
4768 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4771 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4774 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4775 sibcall_value_pcrel_fdpic): Use local variable instead of
4777 (calli_tbr_rel): Add missing operand 2.
4778 (call_valuei_tbr_rel): Add missing operand 3.
4780 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4783 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4784 bitpos/bitsize covering the whole destination, use store_expr only if
4785 the complex mode is the same. Otherwise, use expand_normal and if
4786 it returns CONCAT, subreg each part separately instead of trying to
4787 subreg the whole result.
4789 2017-11-23 Richard Biener <rguenther@suse.de>
4791 PR tree-optimization/23094
4792 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4793 come here from walking over backedges in the first iteration.
4794 (vn_reference_lookup_3): Skip clobbers that store the same value.
4796 2017-11-23 Richard Biener <rguenther@suse.de>
4798 PR tree-optimization/81403
4799 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4800 a block we need a leader relative to.
4801 (phi_translate_1): For nary processing require a leader from
4802 get_representative_for given we run expression simplification
4803 using match-and-simplify. Remove previous fix.
4805 2017-11-22 Jeff Law <law@redhat.com>
4807 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4808 Use new method allocate_value_range rather than accessing the
4809 vrp_value_range_pool data member directly.
4810 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4811 to use extract_range_from_stmt method to avoid need for
4812 extract_range_from_assignment method.
4813 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4814 method rather than setting values_propgated data member directly.
4815 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4816 and values propagated data members and extract_range_from_assignment
4817 method. Reorder private data members to conform to standards.
4818 Add new methods set_lattice_propagation_complete and
4819 allocate_value_range.
4821 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4823 PR rtl-optimization/83030
4824 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4825 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4826 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4827 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4828 CROSSING_JUMP_P flag is already set before setting it.
4829 * cfgrtl.c (fixup_partition_crossing): Likewise.
4830 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4833 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4835 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4836 Handle the case where both arguments are using gen_const_vec_series.
4838 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4841 * pretty-print.c (pp_format): Move quoting implementation to
4842 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4843 to pass address of "quote" local.
4844 (pp_begin_quote): New function.
4845 (pp_end_quote): New function.
4846 * pretty-print.h (printer_fn): Convert penultimate param from bool
4848 (pp_begin_quote): New decl.
4849 (pp_end_quote): New decl.
4850 * tree-diagnostic.c (default_tree_printer): Convert penultimate
4851 param from bool to bool *.
4852 * tree-diagnostic.h (default_tree_printer): Likewise.
4854 2017-11-22 Jeff Law <law@redhat.com>
4856 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
4857 Initialize vr_values.
4858 (evrp_range_analyzer::try_find_new_range): Call methods attached to
4859 vr_values via vr_values class instance rather than delegators.
4860 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
4861 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4862 (evrp_range_analyzer::push_value_range): Likewise.
4863 (evrp_range_analyzer::pop_value_range): Likewise.
4864 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
4865 most delegators. Those remaining are exposed as public interfaces.
4866 Make vr_values a pointer and private.
4867 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
4869 (evrp_range_analyzer::get_vr_value): New method.
4870 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
4871 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
4872 (class evrp_dom_walker): Attach evrp_folder class, initialize
4873 it in the ctor. Remove temporary delegators.
4874 (evrp_dom_walker::before_dom_children): Call methods in attached
4875 evrp_range_analyzer class via class instance pointer. Use
4876 free value_range_constant_singleton to remove need for
4877 op_with_constant_singleton_value delegator method. Do not
4878 create a vrp_prop class instance for every call! Narrow
4879 scope of a couple variables.
4880 (evrp_dom_walker::cleanup): Call methods in attached
4881 evrp_range_analyzer class via class instance pointer.
4882 * vr-values.h (class vr_values): Privatize many methods and
4885 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4887 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
4888 * cfgexpand.c (expand_debug_expr): Remove handling for them.
4889 * expr.c (expand_expr_real_2): Likewise.
4890 * fold-const.c (const_unop): Likewise.
4891 * optabs-tree.c (optab_for_tree_code): Likewise.
4892 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
4893 * tree-inline.c (estimate_operator_cost): Likewise.
4894 * tree-pretty-print.c (dump_generic_node): Likewise.
4895 (op_code_prio): Likewise.
4896 (op_symbol_code): Likewise.
4897 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
4898 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
4899 * internal-fn.c (direct_internal_fn_optab): New function.
4900 (direct_internal_fn_array, direct_internal_fn_supported_p
4901 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
4902 * fold-const-call.c (fold_const_reduction): New function.
4903 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
4905 * tree-vect-loop.c: Include internal-fn.h.
4906 (reduction_code_for_scalar_code): Rename to...
4907 (reduction_fn_for_scalar_code): ...this and return an internal
4909 (vect_model_reduction_cost): Take an internal_fn rather than
4911 (vect_create_epilog_for_reduction): Likewise. Build calls rather
4913 (vectorizable_reduction): Use internal functions rather than tree
4914 codes for the reduction operation. Update calls to the functions
4916 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
4917 Use calls to internal functions rather than REDUC tree codes.
4918 * config/aarch64/aarch64-simd.md: Update comment accordingly.
4920 2017-11-22 Olivier Hainque <hainque@adacore.com>
4922 * config/vxworks.c (vxworks_override_options): Pick default
4923 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
4924 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
4925 DWARF_GNAT_ENCODINGS_DEFAULT.
4926 * config/vxworksae.h: Likewise.
4928 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
4930 PR tree-optimization/83104
4931 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
4934 2017-11-22 Marek Polacek <polacek@redhat.com>
4935 H.J. Lu <hongjiu.lu@intel.com>
4936 Jason Merrill <jason@redhat.com>
4941 * calls.c (initialize_argument_information): Call
4942 warn_parameter_passing_abi target hook.
4943 (store_one_arg): Use 0 for empty record size. Don't push 0 size
4944 argument onto stack.
4945 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
4946 * common.opt: Update -fabi-version description.
4947 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
4948 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
4950 (ix86_is_empty_record): New function.
4951 (ix86_warn_parameter_passing_abi): New function.
4952 (TARGET_EMPTY_RECORD_P): Redefine.
4953 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
4954 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
4955 * doc/tm.texi: Regenerated.
4956 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
4957 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
4958 * dwarf2out.c (get_ultimate_context): Move to tree.c.
4959 * explow.c (hard_function_value): Call arg_int_size_in_bytes
4960 instead of int_size_in_bytes.
4961 * expr.c (copy_blkmode_to_reg): Likewise.
4962 * function.c (aggregate_value_p): Return 0 for empty types.
4963 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
4965 (locate_and_pad_parm): Call arg size_in_bytes instead
4967 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
4968 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
4969 * target.def (empty_record_p, warn_parameter_passing_abi): New target
4971 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
4972 (std_gimplify_va_arg_expr): Skip empty records. Call
4973 arg_size_in_bytes instead size_in_bytes.
4974 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
4975 * tree-core.h (tree_type_common): Add empty_flag.
4976 (tree_decl_common): Update comments.
4977 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
4979 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4980 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
4982 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4983 * tree.c (default_is_empty_type): New function.
4984 (default_is_empty_record): New function.
4985 (arg_int_size_in_bytes): New function.
4986 (arg_size_in_bytes): New function.
4987 (get_ultimate_context): New function.
4988 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
4989 TRANSLATION_UNIT_WARN_EMPTY_P.
4990 (default_is_empty_record, arg_int_size_in_bytes,
4991 arg_size_in_bytes, get_ultimate_context): Declare.
4993 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4995 * config/arm/arm.c (cmse_clear_registers): New function.
4996 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
4997 code by call to cmse_clear_registers.
4998 (cmse_nonsecure_entry_clear_before_return): Likewise.
5000 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5002 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5003 (vdot_s32, vdotq_s32): New.
5004 (vdot_lane_u32, vdotq_lane_u32): New.
5005 (vdot_lane_s32, vdotq_lane_s32): New.
5008 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5011 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5012 for unsigned values with fewer HWIs than the precision.
5013 (test_overflow): New function.
5014 (wide_int_cc_tests): Call it.
5016 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5018 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5019 has the same mode class as Pmode.
5021 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5023 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5024 padding_bits_to_clear_ptr.
5025 (cmse_nonsecure_entry_clear_before_return): Likewise.
5027 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5029 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5030 auto_sbitap instead of integer bitfield to control register needing
5033 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5035 PR tree-optimization/83044
5036 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5037 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5038 that need the upper bound. Subtract offset from
5039 get_addr_base_and_unit_offset only if positive and subtract it
5040 before division by eltsize rather than after it.
5043 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5044 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5045 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5049 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5050 expand_binop with *mul_widen_optab, make sure at least one of the
5051 operands doesn't have VOIDmode.
5054 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5056 PR rtl-optimization/82044
5057 PR tree-optimization/82042
5058 * dse.c (record_store): Check for overflow.
5059 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5060 add_wild_read instead of clear_rhs_from_active_local_stores on
5061 overflow and log it into dump_file.
5063 2017-11-22 Richard Biener <rguenther@suse.de>
5065 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5066 fake edges to exit when looking for a place to insert.
5067 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5069 (insert_into_preds_of_block): Commit edge insertion immediately,
5070 assert that doesn't require new BBs.
5071 (fini_pre): Release expressions.
5072 (pass_pre::execute): Shuffle things around a bit, if the fn
5073 is too large do not compute AVAIL either as this is really the
5076 2017-11-22 Richard Biener <rguenther@suse.de>
5078 PR tree-optimization/83089
5079 * tree-if-conv.c (pass_if_conversion::execute): If anything
5080 changed reset SCEV and free the number of iteration estimates.
5082 2017-11-21 Martin Sebor <msebor@redhat.com>
5084 PR tree-optimization/82945
5085 * calls.h (warn_nonstring_bound): Remove unused function.
5087 2017-11-21 Martin Sebor <msebor@redhat.com>
5089 PR tree-optimization/82945
5090 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5091 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5092 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5094 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5095 * calls.h (get_attr_nonstring_decl): Declare new function.
5096 * doc/extend.texi (attribute nonstring): Update.
5097 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5098 get_attr_nonstring_decl and handle it.
5099 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5100 detection of nul-termination.
5101 (strlen_to_stridx): Change to a pointer.
5102 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5103 (pass_strlen::execute): Same.
5105 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5107 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5108 for the new option -mprefer-vector-width=[none|128|256|512].
5109 * config/i386/i386.c (ix86_target_string): remove old style options
5110 -mprefer-avx256 and make -mprefer-avx128 as alias.
5111 (ix86_option_override_internal): Apply defaults for the
5112 -mprefer-vector-width=[128|256] option.
5113 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5114 Implement macros to work with -mprefer-vector-width=.
5115 * config/i386/i386.opt: Implemented option
5116 -mprefer-vector-width=[none|128|256|512].
5117 * doc/invoke.texi: Documentation for
5118 -mprefer-vector-width=[none|128|256|512].
5120 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5122 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5124 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5126 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5128 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5130 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5132 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5133 * expr.c (expand_expr_real_2): Likewise.
5134 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5135 fold_binary_loc): Likewise.
5136 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5137 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5138 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5139 MINUS_EXPR transformations.
5140 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5141 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5142 * tree-inline.c (estimate_operator_cost): Likewise.
5143 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5144 op_symbol_code): Likewise.
5145 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5146 * vr-values.c (extract_range_from_binary_expr): Likewise.
5147 * varasm.c (initializer_constant_valid_p_1): Likewise.
5148 * tree.def: New tree code POINTER_DIFF_EXPR.
5150 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5152 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5153 integer suffix to movbe mnemonic.
5154 (*bswaphi2_movbe): Ditto.
5155 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5157 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5160 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5161 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5162 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5164 2017-11-21 Martin Liska <mliska@suse.cz>
5166 * tree-inline.c (expand_call_inline): Remove not needed
5169 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5170 Jakub Jelinek <jakub@redhat.com>
5173 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5174 Remove static keyword from f variable.
5176 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5178 PR tree-optimization/83086
5179 * gimple-ssa-store-merging.c
5180 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5181 rather than n.base_addr.
5183 2017-11-21 Martin Liska <mliska@suse.cz>
5185 PR rtl-optimization/82044
5186 PR tree-optimization/82042
5187 * dse.c (check_mem_read_rtx): Check for overflow.
5189 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5191 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5194 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5196 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5197 and its corresponding call arg location note.
5199 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5200 Andrew Burgess <andrew.burgess@embecosm.com>
5202 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5204 (arc_return_slot_offset): Likewise.
5205 (arc_eh_return_address_location): New declaration.
5206 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5207 (MUST_SAVE_REGISTER): Add exception handler case.
5208 (MUST_SAVE_RETURN_ADDR): Likewise.
5209 (arc_frame_pointer_required): Likewise.
5210 (arc_frame_pointer_needed): New function.
5211 (arc_compute_frame_size): Changed.
5212 (arc_expand_prologue): Likewise.
5213 (arc_expand_epilogue): Likewise.
5214 (arc_initial_elimination_offset): Likewise.
5215 (arc_return_slot_offset): Delete.
5216 (arc_eh_return_address_location): New function.
5217 (arc_builtin_setjmp_frame_value): Likewise.
5218 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5219 (EH_RETURN_STACKADJ_RTX): Define.
5220 (EH_RETURN_HANDLER_RTX): Likewise.
5221 * config/arc/arc.md (eh_return): Delete.
5223 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5225 * print-tree.h (debug_vec_tree): Remove prototype.
5226 * gdbinit.in (pvt): Remove macro.
5228 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5230 PR tree-optimization/83047
5231 * gimple-ssa-store-merging.c
5232 (imm_store_chain_info::output_merged_store): If the loads with the
5233 same vuse are in different basic blocks, for load_gsi pick a load
5234 location that is dominated by the other loads.
5237 * config/i386/i386.c (ix86_memmodel_check): Start
5238 -Winvalid-memory-model diagnostics with lowercase letter.
5241 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5242 set in_cold_section_p to the partition containing loc_list->first.
5243 When seeing loc_list->last_before_switch node, update secname and
5244 perform range_across_switch second partition handling only after that.
5247 * run-rtl-passes.c: Include debug.h.
5248 (run_rtl_passes): Call debug_hooks->assembly_start.
5249 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5253 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5254 OPTAB_DIRECT in calls to expand_simple_binop.
5256 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5259 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5261 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5264 * spellcheck.h (best_match::blithely_get_best_candidate): New
5267 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5269 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5270 with lower case letter.
5272 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5274 * config/i386/i386.md (bswaphi2): New expander.
5275 (*bswaphi2_movbe): New insn pattern.
5276 (bswaphi -> rorhi pepehole2): New peephole pattern.
5278 2017-11-20 Jeff Law <law@redhat.com>
5280 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5281 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5282 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5283 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5284 gimple-ssa-evrp-analyze.h.
5286 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5287 set BB_VISITED here.
5288 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5290 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5292 * tree-predcom.c: Add general comment on Store-Store chains.
5293 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5295 (get_chain_last_ref_at): Rename into...
5296 (get_chain_last_write_at): ...this.
5297 (get_chain_last_write_before_load): New function.
5298 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5299 CT_STORE_STORE when write reference is added.
5300 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5301 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5302 (initialize_root_vars_store_elim_1): Ditto.
5303 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5304 definition is created.
5305 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5306 chain by replacing it with dominant stored value.
5308 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5310 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5312 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5314 * vr-values.c (extract_range_from_binary_expr): Use a full range
5317 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5319 * config/arm/arm.md (R4_REGNUM): Define constant.
5320 (nonsecure_call_internal): Remove r4 clobber.
5321 (nonsecure_call_value_internal): Likewise.
5322 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5323 clobber and resequence match_operands.
5324 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5325 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5326 (nonsecure_call_value_reg_thumb2): Likewise.
5328 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5330 PR tree-optimization/78821
5331 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5332 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5333 to the address of the base rather than the base itself.
5334 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5335 (find_bswap_or_nop_finalize): New function.
5336 (find_bswap_or_nop): Use it.
5337 (bswap_replace): Return a tree rather than bool, change first
5338 argument from gimple * to gimple_stmt_iterator, allow inserting
5339 into an empty sequence, allow ins_stmt to be NULL - then emit
5340 all stmts into gsi. Fix up MEM_REF address gimplification.
5341 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5342 (struct store_immediate_info): Add N and INS_STMT non-static
5344 (store_immediate_info::store_immediate_info): Initialize them
5345 from newly added ctor args.
5346 (merged_store_group::apply_stores): Formatting fixes. Sort by
5348 (stmts_may_clobber_ref_p): For stores call also
5349 refs_anti_dependent_p.
5350 (gather_bswap_load_refs): New function.
5351 (imm_store_chain_info::try_coalesce_bswap): New method.
5352 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5353 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5354 (imm_store_chain_info::output_merged_store): Fail if number of
5355 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5356 and NOP_EXPR rhs_code.
5357 (pass_store_merging::process_store): Compute n and ins_stmt, if
5358 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5359 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5361 (pass_store_merging::execute): Calculate dominators.
5363 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5364 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5365 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5366 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5367 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5368 class pass_optimize_bswap, bswap_replace,
5369 pass_optimize_bswap::execute): Moved to ...
5370 * gimple-ssa-store-merging.c: ... this file.
5371 Include optabs-tree.h.
5372 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5373 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5374 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5375 anonymous namespace, remove static keywords.
5376 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5377 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5379 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5382 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5384 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5386 * vec.h (debug_helper): New function.
5387 (DEFINE_DEBUG_VEC): New macro.
5388 * hash-set.h (debug_helper): New function.
5389 (DEFINE_DEBUG_HASH_SET): New macro.
5390 * cfg.c (debug_slim (edge)): New function.
5391 Call DEFINE_DEBUG_VEC for edges.
5392 Call DEFINE_DEBUG_HASH_SET for edges.
5393 * cfghooks.c (debug_slim (basic_block)): New function.
5394 Call DEFINE_DEBUG_VEC for basic blocks.
5395 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5396 * print-tree.c (debug_slim): New function to handle trees.
5397 Call DEFINE_DEBUG_VEC for trees.
5398 Call DEFINE_DEBUG_HASH_SET for trees.
5399 (debug (vec<tree, va_gc>) &): Remove.
5400 (debug (<vec<tree, va_gc>) *): Remove.
5401 * print-rtl.c (debug_slim): New function to handle const_rtx.
5402 Call DEFINE_DEBUG_VEC for rtx_def.
5403 Call DEFINE_DEBUG_VEC for rtx_insn.
5404 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5405 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5406 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5407 (debug (vec<rtx_insn *> *ptr): Remove.
5408 (debug_insn_vector): Remove.
5409 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5411 2017-11-20 Tom de Vries <tom@codesourcery.com>
5413 PR rtl-optimization/82020
5414 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5415 IF_THEN_ELSE condition.
5417 2017-11-19 Jeff Law <law@redhat.com>
5419 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5420 of degenerates resulting from ignoring an edge.
5422 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5425 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5427 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5430 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5431 for uninitialied counts.
5433 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5436 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5438 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5441 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5443 2017-11-19 Tom de Vries <tom@codesourcery.com>
5445 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5448 2017-11-19 Tom de Vries <tom@codesourcery.com>
5450 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5453 2017-11-19 Tom de Vries <tom@codesourcery.com>
5455 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5458 2017-11-19 Tom de Vries <tom@codesourcery.com>
5460 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5462 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5464 2017-11-19 Tom de Vries <tom@codesourcery.com>
5466 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5468 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5469 semicolon after MCORE_EXPORT_NAME call.
5471 2017-11-19 Tom de Vries <tom@codesourcery.com>
5474 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5475 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5477 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5479 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5480 (cgraph_edge::clone): Cleanup updating of profile.
5481 * ipa-cp.c (update_profiling_info): Likewise.
5482 * ipa-inline-transform.c (inline_transform): Likewise.
5483 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5484 * ipa-split.c (execute_split_functions): Do not split when function
5486 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5487 * profile-count.c (profile_count::dump): Add global0.
5488 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5490 (profile_count::to_sreal_scale): Likewise.
5491 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5492 (profile_count::combine_with_ipa_count): New function.
5493 * profile-count.h (profile_guessed_global0adjusted): New.
5494 (profile_count::adjusted_zero): New.
5495 (profile_count::global0adjusted): New.
5496 (profile_count::combine_with_ipa_count): New.
5497 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5498 correct profile of return block of split functions.
5499 (copy_cfg_body): Remove unused profile_count.
5500 (copy_body): Likewise.
5501 (expand_call_inline): Update.
5502 (tree_function_versioning): Update.
5504 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5506 * hash-set.h (hash_set::empty): New.
5507 * tree-ssa-threadbackward.h: Delete.
5508 * tree-ssa-threadbackward.c (class thread_jumps): New.
5509 Move max_threaded_paths into class.
5510 (fsm_find_thread_path): Remove arguments that are now in class.
5511 (profitable_jump_thread_path): Rename to...
5512 (thread_jumps::profitable_jump_thread_path): ...this.
5513 (convert_and_register_jump_thread_path): Rename to...
5514 (thread_jumps::convert_and_register_current_path): ...this.
5515 (check_subpath_and_update_thread_path): Rename to...
5516 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5517 (register_jump_thread_path_if_profitable): Rename to...
5518 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5519 (handle_phi): Rename to...
5520 (thread_jumps::handle_phi): ...this.
5521 (handle_assignment): Rename to...
5522 (thread_jumps::handle_assignment): ...this.
5523 (fsm_find_control_statement_thread_paths): Rename to...
5524 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5525 (find_jump_threads_backwards): Rename to...
5526 (thread_jumps::find_jump_threads_backwards): ...this.
5527 Initialize path local data.
5528 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5529 from within thread_jumps class.
5530 (pass_early_thread_jumps::execute): Same.
5532 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5534 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5536 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5538 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5539 to canonical location.
5541 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5544 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5546 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5548 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5550 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5551 rcrt1.o%s/grcrt1.o%s for -static-pie.
5553 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5555 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5556 ix86_shift_rotate_cost): Break out from ...
5557 (ix86_rtx_costs): ... here.
5558 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5561 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5563 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5564 when entry block was promoted unlikely.
5565 (estimate_bb_frequencies): Increase frequency scale.
5566 * profile-count.h (profile_count): Export precision info.
5568 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5570 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5571 disturbin profile of entry block.
5573 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5575 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5578 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5580 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5581 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5583 2017-11-17 Jeff Law <law@redhat.com>
5585 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5586 from evrp_dom_walker class. Various methods moved into new class.
5587 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5588 (evrp_range_analyzer::enter): New method.
5589 (evrp_range_analyzer::leave): New method.
5590 (evrp_dom_walker): Remove delegators no longer needed by this class.
5591 Replace vr_values data member with evrp_range_analyzer
5593 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5594 method extracted from evrp_dom_walker::before_dom_children.
5595 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5596 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5598 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5599 Add private copy constructor and move assignment operators.
5600 Privatize methods and class data where trivially possible.
5601 (evrp_dom_walker::cleanup): New function, extracted from
5602 execute_early_vrp. Simplify access to class data.
5604 * vr-values.h (get_output_for_vrp): Prototype.
5605 * vr-values.c (get_output_for_vrp): New function extracted from
5606 vrp_visit_assignment_or_call and extract_range_from_stmt.
5607 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5609 2017-11-17 Luis Machado <luis.machado@linaro.org>
5611 * config/aarch64/aarch64.c
5612 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5613 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5614 tune_params::AUTOPREFETCHER_WEAK.
5616 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5619 * config/arm/arm.c (arm_valid_target_attribute_rec):
5620 Parse "arch=" and "+<ext>".
5621 (arm_valid_target_attribute_tree): Re-init global options.
5622 (arm_option_override): Make non-static.
5623 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5624 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5625 (__ARM_FEATURE_CRC32): Support undef.
5626 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5627 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5629 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5631 * gdbinit.in (break-on-diagnostic): New command.
5633 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5635 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5636 return to indirect jump for EH return if control-flow
5637 protection is enabled. Change explicit 'false' argument in
5638 pro_epilogue_adjust_stack with a value of
5640 * config/i386/i386.md (simple_return_indirect_internal):
5641 Remove SImode restriction to support 64-bit.
5643 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5645 * combine.c (added_notes_insn): New.
5646 (try_combine): Handle added_notes_insn like added_links_insn.
5647 Rewrite return value code.
5648 (distribute_notes): Set added_notes_insn to the earliest insn we added
5651 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5653 PR rtl-optimization/82621
5654 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5655 dest of one of those SETs is unused.
5657 2017-11-17 Richard Biener <rguenther@suse.de>
5660 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5661 * tree-pretty-print.c (dump_generic_node): Handle
5662 annot_expr_parallel_kind.
5663 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5664 * gimplify.c (gimple_boolify): Likewise.
5666 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5668 * config.gcc (extra_headers): Add cet.h for x86 targets.
5669 * config/i386/cet.h: New file.
5670 * doc/install.texi: Add --enable-cet/--disable-cet.
5672 2017-11-17 Richard Biener <rguenther@suse.de>
5674 PR tree-optimization/83017
5675 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5676 (gen_parallel_loop): Properly count iterations.
5677 (parallelize_loops): Handle loop->can_be_parallel independent
5678 of flag_loop_parallelize_all. Make static profitability test match
5680 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5681 * invoke.texi (parloops-min-per-thread): Document.
5683 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5685 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5686 upstreaming review comments.
5688 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5690 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5691 with fast unaligned access.
5692 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5694 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5696 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5697 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5699 2017-11-17 Richard Biener <rguenther@suse.de>
5701 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5702 folding of references.
5704 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5707 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5708 of replacing call to strncmp with corresponding call to strcmp when
5711 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5713 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5714 option prefer-avx256 for skylake-avx512 configuration.
5715 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5716 (get_builtin_code_for_version): Ditto.
5718 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5719 Monk Chiang <sh.chiang04@gmail.com>
5721 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5722 (FIXED_REGISTERS): Reserve more register numbers.
5723 (CALL_USED_REGISTERS): Likewise.
5724 (REG_ALLOC_ORDER): Likewise.
5725 (REG_CLASS_CONTENTS): Likewise.
5726 (REGISTER_NAMES): Likewise.
5728 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5729 Kito Cheng <kito.cheng@gmail.com>
5731 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5733 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5735 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5737 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5738 parameter list for vec_splats.
5740 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5742 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5744 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5746 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5749 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5750 with the required version of automake 1.11.6.
5752 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5754 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5756 (DU_C2_3_power9): Correct reservation combinations.
5757 (FP_DIV_power9, VEC_DIV_power9): New.
5758 (power9-alu): Split out rotate/shift...
5759 (power9-rot): ...to here, correct dispatch resource.
5760 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5762 (power9-fp): Correct latency.
5763 (power9-sdiv): Add div/sqrt resource.
5764 (power9-ddiv): Correct latency, add div/sqrt resource.
5765 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5766 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5768 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5770 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5772 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5773 switch statement mapping KF built-ins to TF built-ins if we don't
5774 have the proper ISA 3.0 assembler support.
5776 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5778 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5779 (lower_emutls_function_body): Do not compute it.
5781 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5783 * ipa-split.c (split_bb_info): Turn time to sreal.
5784 (split_point): Likewise.
5785 (dump_split_point): Likewise.
5786 (fine_split_points): Likewise.
5787 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5789 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5791 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5794 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5796 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5798 (expensive_function_p): Remove useless assert.
5799 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5801 2017-11-16 Martin Sebor <msebor@redhat.com>
5803 PR tree-optimization/82588
5804 PR tree-optimization/82583
5805 * tree-vrp.c (check_array_ref): Handle flexible array members,
5806 string literals, and inner indices.
5807 (search_for_addr_array): Add detail to diagnostics.
5809 2017-11-16 Nathan Sidwell <nathan@acm.org>
5813 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5814 (SET_DECL_ASSEMBLER_NAME): Forward to
5815 overwrite_decl_assembler_name.
5816 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5817 (overwrite_decl_assembler_name): Declare.
5818 * tree.c (overwrite_decl_assembler_name): New.
5819 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5820 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5821 (LANG_HOOKS_INITIALIZER): Add it.
5822 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5823 * langhooks.c (lhd_set_decl_assembler_name): Use
5824 SET_DECL_ASSEMBLER_NAME.
5825 (lhd_overwrite_decl_assembler_name): Default implementation.
5827 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5828 Jackson Woodruff <jackson.woodruff@arm.com>
5830 PR tree-optimization/71026
5831 * match.pd: Canonicalize constant multiplies in division.
5833 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5835 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5836 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5837 Remove OPT_fomit_frame_pointer entry.
5838 * common/config/arc/arc-common.c: Likewise.
5839 * common/config/arm/arm-common.c: Likewise.
5840 * common/config/avr/avr-common.c: Likewise.
5841 * common/config/c6x/c6x-common.c: Likewise.
5842 * common/config/cr16/cr16-common.c: Likewise.
5843 * common/config/cris/cris-common.c: Likewise.
5844 * common/config/epiphany/epiphany-common.c: Likewise.
5845 * common/config/fr30/fr30-common.c: Likewise.
5846 * common/config/frv/frv-common.c: Likewise.
5847 * common/config/ia64/ia64-common.c: Likewise.
5848 * common/config/iq2000/iq2000-common.c: Likewise.
5849 * common/config/lm32/lm32-common.c: Likewise.
5850 * common/config/m32r/m32r-common.c: Likewise.
5851 * common/config/mcore/mcore-common.c: Likewise.
5852 * common/config/microblaze/microblaze-common.c: Likewise.
5853 * common/config/mips/mips-common.c: Likewise.
5854 * common/config/mmix/mmix-common.c: Likewise.
5855 * common/config/mn10300/mn10300-common.c: Likewise.
5856 * common/config/nios2/nios2-common.c: Likewise.
5857 * common/config/pa/pa-common.c: Likewise.
5858 * common/config/pdp11/pdp11-common.c: Likewise.
5859 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
5860 * common/config/riscv/riscv-common.c: Likewise.
5861 * common/config/rs6000/rs6000-common.c: Likewise.
5862 * common/config/rx/rx-common.c: Likewise.
5863 * common/config/s390/s390-common.c: Likewise.
5864 * common/config/sh/sh-common.c: Likewise.
5865 * common/config/sparc/sparc-common.c: Likewise.
5866 * common/config/tilegx/tilegx-common.c: Likewise.
5867 * common/config/tilepro/tilepro-common.c: Likewise.
5868 * common/config/v850/v850-common.c: Likewise.
5869 * common/config/visium/visium-common.c: Likewise.
5870 * common/config/xstormy16/xstormy16-common.c: Likewise.
5871 * common/config/xtensa/xtensa-common.c: Likewise.
5872 * invoke.texi (-fomit-frame-pointer): Update documentation.
5874 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5876 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
5878 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5880 * cfg.c (scale_bbs_frequencies_int,
5881 cale_bbs_frequencies_gcov_type): Remove.
5882 * cfg.h (scale_bbs_frequencies_int,
5883 cale_bbs_frequencies_gcov_type): Remove.
5885 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5887 * tree-ssa-loop-manip.c
5888 (scale_dominated_blocks_in_loop): Update to profile counts.
5889 (tree_transform_and_unroll_loop): Likewise.
5891 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5893 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
5894 scale_bbs_frequencies_int.
5896 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5898 * final.c (compute_alignments): Use counts rather than frequencies.
5900 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5902 * cfgloopanal.c: Include sreal.h
5903 (average_num_loop_insns): Use counts and sreal for accounting.
5905 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5907 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
5910 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5911 Kito Cheng <kito.cheng@gmail.com>
5913 * config/nds32/constraints.md: Provide more constraints.
5914 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
5915 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
5916 support constraints usage.
5918 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5920 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
5922 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5923 Kito Cheng <kito.cheng@gmail.com>
5925 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
5926 * config/nds32/nds32.opt: Refine the layout.
5927 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
5928 TARGET_EXT_STRING): Support new options.
5929 * config/nds32/nds32.h: Likewise.
5930 * config/nds32/nds32.md: Likewise.
5931 * config/nds32/nds32-predicates.c: Likewise.
5932 * config/nds32/constraints.md: Likewise.
5933 * common/config/nds32/nds32-common.c: Likewise.
5935 2017-11-16 Julia Koval <julia.koval@intel.com>
5938 * config/i386/gfniintrin.h: Add sse check.
5939 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
5941 2017-11-16 Julia Koval <julia.koval@intel.com>
5943 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
5944 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
5945 (ix86_handle_option): Handle -mavx512vbmi2.
5946 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
5947 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
5948 * config/i386/i386-c.c (__AVX512VBMI2__): New.
5949 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
5950 (ix86_valid_target_attribute_inner_p): Ditto.
5951 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
5952 * config/i386/i386.opt (mavx512vbmi2): New option.
5953 * doc/invoke.texi: Add new option.
5955 2017-11-16 Julia Koval <julia.koval@intel.com>
5957 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
5958 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
5959 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
5960 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
5961 _mm512_gf2p8mul_epi8): New intrinsics.
5962 * config/i386/i386-builtin-types.def
5963 (V64QI_FTYPE_V64QI_V64QI): New type.
5964 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
5965 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
5966 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
5967 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
5968 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
5969 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
5971 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
5973 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
5974 explicitly as a stream of bytes.
5976 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5978 * config/rs6000/altivec.h (vec_xst_be): New #define.
5979 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
5980 and externalize from *altivec_vperm_<mode>_internal.
5981 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
5983 (XL_BE_V8HI): Likewise.
5984 (XL_BE_V4SI): Likewise.
5985 (XL_BE_V4SI): Likewise.
5986 (XL_BE_V2DI): Likewise.
5987 (XL_BE_V4SF): Likewise.
5988 (XL_BE_V2DF): Likewise.
5989 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
5990 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
5991 all array entries with these keys: VSX_BUILTIN_VEC_XL,
5992 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
5993 VSX_BUILTIN_VEC_XST_BE.
5994 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
5995 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
5997 (altivec_init_builtins): Replace conditional calls to def_builtin
5998 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
5999 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6000 with unconditional calls. Remove calls to def_builtin for
6001 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6002 __builtin_vec_xst_be.
6003 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6004 to define_expand, and add alternate RTL generation for P8.
6005 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6006 vsx_ld_elemrev_v8hi.
6007 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6008 add alternate RTL generation for P8.
6009 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6010 vsx_ld_elemrev_v16qi.
6011 (vsx_st_elemrev_v8hi): Convert define_insn
6012 to define_expand, and add alternate RTL generation for P8.
6013 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6014 vsx_st_elemrev_v8hi.
6015 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6016 add alternate RTL generation for P8.
6017 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6018 vsx_st_elemrev_v16qi.
6020 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6023 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6024 TARGET_AVX512ER check.
6025 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6026 neither -mzeroupper nor -mno-zeroupper is used and
6027 TARGET_EMIT_VZEROUPPER is set.
6028 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6029 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6031 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6033 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6034 folding of vector compares.
6035 (fold_build_vec_cmp): New helper function.
6036 (fold_compare_helper): New helper function.
6037 (builtin_function_type): Add compare builtins to the list of functions
6038 having unsigned arguments. Cosmetic updates to comment indentation.
6039 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6040 the not+eq combination.
6042 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6044 PR tree-optimization/82726
6045 PR tree-optimization/70754
6046 * tree-predcom.c (order_drefs_by_pos): New function.
6047 (combine_chains): Move code setting has_max_use_after to...
6048 (try_combine_chains): ...here. New parameter. Sort combined chains
6049 according to position information.
6050 (tree_predictive_commoning_loop): Update call to above function.
6051 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6053 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6055 PR tree-optimization/82726
6057 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6059 PR tree-optimization/70754
6060 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6061 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6062 combined stmt before it if not NULL.
6063 (combine_chains): Process refs reversely and compute dominance point
6067 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6069 PR tree-optimization/79663
6070 * tree-predcom.c (combine_chains): Process refs in reverse order
6071 only for ZERO length chains, and add explaining comment.
6073 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6075 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6076 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6077 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6079 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6081 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6083 2017-11-15 Martin Liska <mliska@suse.cz>
6085 * tree-cfg.c (pass_warn_function_return::execute):
6086 Compare warn_return_type for greater than zero.
6088 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6092 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6093 to return true on Xeon and not on Xeon Phi.
6094 (ix86_check_avx256_register): Changed to ...
6095 (ix86_check_avx_upper_register): ... this. Add extra check for
6096 VALID_AVX512F_REG_OR_XI_MODE.
6097 (ix86_avx_u128_mode_needed): Changed
6098 ix86_check_avx256_register to ix86_check_avx_upper_register.
6099 (ix86_check_avx256_stores): Changed to ...
6100 (ix86_check_avx_upper_stores): ... this. Changed
6101 ix86_check_avx256_register to ix86_check_avx_upper_register.
6102 (ix86_avx_u128_mode_after): Changed
6103 avx_reg256_found to avx_upper_reg_found. Changed
6104 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6105 (ix86_avx_u128_mode_entry): Changed
6106 ix86_check_avx256_register to ix86_check_avx_upper_register.
6107 (ix86_avx_u128_mode_exit): Ditto.
6108 * config/i386/i386.h: (host_detect_local_cpu): New define.
6110 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6112 * config/arm/xgene1.md (xgene1): Split into automatons
6113 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6114 (xgene1_f_load): Adjust reservations and/or types.
6115 (xgene1_f_store): Likewise.
6116 (xgene1_load_pair): Likewise.
6117 (xgene1_store_pair): Likewise.
6118 (xgene1_fp_load1): Likewise.
6119 (xgene1_load1): Likewise.
6120 (xgene1_store1): Likewise.
6121 (xgene1_move): Likewise.
6122 (xgene1_alu): Likewise.
6123 (xgene1_simd): Likewise.
6124 (xgene1_bfm): Likewise.
6125 (xgene1_neon_load1): Likewise.
6126 (xgene1_neon_store1): Likewise.
6127 (xgene1_neon_logic): Likewise.
6128 (xgene1_neon_st1): Likewise.
6129 (xgene1_neon_ld1r): Likewise.
6130 (xgene1_alu_cond): Added.
6131 (xgene1_shift_reg): Likwise.
6132 (xgene1_bfx): Likewise.
6133 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6135 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6138 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6140 (can_widen_mult_without_libcall): New function.
6141 (expand_mul_overflow): If only checking unsigned mul overflow,
6142 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6143 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6144 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6145 (expand_DIVMOD): Formatting fix.
6146 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6147 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6148 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6150 PR tree-optimization/82977
6151 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6152 constructed temporary to strlen_to_stridx.put.
6154 2017-11-15 Martin Liska <mliska@suse.cz>
6156 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6157 * configure: Regenerate.
6159 2017-11-15 Martin Liska <mliska@suse.cz>
6162 * config/sh/sh-mem.cc: Use proper probability for
6165 2017-11-14 Jeff Law <law@redhat.com>
6167 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6168 the red zone for stack_clash_protection_final_dynamic_probe targets
6169 when the total dynamic stack size is zero bytes.
6171 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6172 blocks is post order.
6174 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6176 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6177 * final.c (rest_of_clean_state): Set it for the
6178 -fcompare-debug dump.
6179 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6180 class when TDF_COMPARE_DEBUG is set.
6182 * dwarf2out.c (gen_producer_string): Discard
6185 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6188 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6189 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6190 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6191 (__TGMATH_REAL_2_3): Remove macros.
6192 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6193 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6195 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6197 (remquo): Define using __TGMATH_REAL_3.
6199 2017-11-14 Jeff Law <law@redhat.com>
6201 * vr-values.c: New file with contents extracted from tree-vrp.c.
6202 * Makefile.in (OBJS): Add vr-values.o
6203 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6204 (set_value_range, set_and_canonicalize_value_range): Likewise.
6205 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6206 (value_range_constant_singleton, symbolic_range_p): Likewise.
6207 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6208 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6209 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6210 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6211 (find_case_label_range, find_case_label_index): Likewise.
6212 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6213 (range_int_cst_singleton_p, value_inside_range): Likewise.
6214 (get_single_symbol): Likewise.
6215 (switch_update): Move structure definition here.
6216 (to_remove_edges, to_update_switch_stmts): Provide externs.
6217 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6218 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6219 (vrp_val_is_min, set_value_range): Likewise.
6220 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6221 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6222 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6223 (range_is_nonnull, range_int_cst_p): Likewwise.
6224 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6225 (get_single_symbol, operand_less_p): Likewise
6226 (compare_values_warnv, compare_values): Likewise.
6227 (value_inside_range, value_range_constant_singleton): Likewise.
6228 (zero_nonzero_bitgs_from_vr): Likewise.
6229 (extract_range_from_binary_expr_1): Likewise.
6230 (overflow_comparison_p): Likewise.
6231 (to_remove_edges, to_update_switch_stmts): Likewise.
6232 (find_case_label-index, find_case_label_range): Likewise.
6233 (switch_update, set_value_range_to_nonnegative): Remove.
6234 (set_value_range_to_truthvalue): Likewise.
6235 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6236 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6237 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6238 (find_case_label_ranges, test_for_singularity): Likewise.
6239 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6240 (x_vr_values): Move to its remaining use site.
6242 2017-11-10 Jeff Law <law@redhat.com>
6244 * vr-values.h (VR_INITIALIZER): Move #define here.
6245 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6246 * Makefile.in (OBJS): Add tree-evrp.o
6247 * tree-vrp.h (assert_info): Move structure definition here.
6248 (set_value_range_to_varying): Prototype.
6249 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6250 (infer_value_range, register_edge_assert_for): Likewise.
6251 (stmt_interesting_for_vrp): Likewise.
6252 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6253 (set_value_range_to_varying): No longer static.
6254 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6255 (infer_value_range, register_edge_assert_for): Likewise.
6257 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6259 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6260 generate the XXBRD instruction.
6262 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6263 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6264 as being compatible if -mabi=ieeelongdouble.
6265 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6266 to setup float128 built-ins with hardware support.
6267 (BU_FLOAT128_HW_2): Likewise.
6268 (BU_FLOAT128_HW_3): Likewise.
6269 (BU_FLOAT128_HW_VSX_1): Likewise.
6270 (BU_FLOAT128_HW_VSX_2): Likewise.
6271 (scalar_extract_expq): Change float128 built-in functions to
6272 accommodate having both KFmode and TFmode functions. Use the
6273 KFmode variant as the default.
6274 (scalar_extract_sigq): Likewise.
6275 (scalar_test_neg_qp): Likewise.
6276 (scalar_insert_exp_q): Likewise.
6277 (scalar_insert_exp_qp): Likewise.
6278 (scalar_test_data_class_qp): Likewise.
6279 (sqrtf128_round_to_odd): Delete processing the round to odd
6280 built-in functions as special built-in functions, and define them
6281 as float128 built-ins. Use the KFmode variant as the default.
6282 (truncf128_round_to_odd): Likewise.
6283 (addf128_round_to_odd): Likewise.
6284 (subf128_round_to_odd): Likewise.
6285 (mulf128_round_to_odd): Likewise.
6286 (divf128_round_to_odd): Likewise.
6287 (fmaf128_round_to_odd): Likewise.
6288 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6289 support for KFmode and TFmode xststdcqp calls.
6290 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6291 point, switch the built-in handlers for the get/set float128
6292 exponent, get float128 mantissa, float128 test built-ins, and the
6293 float128 round to odd built-in functions. Eliminate creating the
6294 float128 round to odd built-in functions as special built-ins.
6295 (rs6000_init_builtins): Eliminate special creation of the float128
6296 round to odd built-in functions.
6297 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6298 function insns to support both TFmode and KFmode variants.
6299 (xsxsigqp_<mode>): Likewise.
6300 (xsiexpqpf_<mode>): Likewise.
6301 (xsiexpqp_<mode>): Likewise.
6302 (xststdcqp_<mode>): Likewise.
6303 (xststdcnegqp_<mode>): Likewise.
6304 (xststdcqp_<mode>): Likewise.
6306 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6308 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6309 unused path_in_freq_ptr parameter.
6310 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6312 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6314 * ipa-inline.c (edge_badness): Dump sreal frequency.
6315 (compute_inlined_call_time): Match natural implementaiton ...
6316 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6317 forgotten division by CGRAPH_FREQ_BASE.
6319 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6321 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6322 Solaris 11. Update comment.
6323 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6325 * config/sol2.h (STARTFILE_SPEC): Likewise.
6326 * configure: Regenerate.
6328 2017-11-14 Carl Love <cel@us.ibm.com>
6330 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6331 le_ and be_ prefixes to swap* variables. Remove
6332 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6335 2017-11-14 Jason Merrill <jason@redhat.com>
6337 Support GTY((cache)) on hash_map.
6338 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6339 (ggc_cache_remove): Override it instead of ggc_mx.
6340 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6341 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6342 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6343 (simple_cache_map_traits): Override maybe_mx.
6344 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6345 (hash_map): Friend gt_cleare_cache.
6346 (gt_cleare_cache): New.
6347 * tree.h (tree_cache_traits): New hash_map traits class.
6348 (tree_cache_map): New typedef.
6350 2017-11-14 Richard Biener <rguenther@suse.de>
6352 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6353 paramter and handling.
6354 (cleanup_control_flow_bb): Likewise.
6355 (cleanup_control_flow_pre): New helper performing a DFS walk
6356 to call cleanup_control_flow_bb in PRE order.
6357 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6358 via cleanup_control_flow_pre.
6360 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6362 * config/aarch64/aarch64-simd.md
6363 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6364 (*aarch64_simd_bsl<mode>_alt): Likewise.
6365 (aarch64_simd_bsldi_internal): New.
6366 (aarch64_simd_bsldi_alt): Likewise.
6368 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6370 * tracer.c (better_p): Do not compare frequencies.
6371 * reg-stack.c (better_edge): Likewise.
6372 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6375 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6377 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6378 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6379 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6380 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6381 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6382 * predict.c (maybe_hot_frequency_p): Inline to ...
6383 (maybe_hot_count_p): ... here; rewrite to counts.
6384 (counts_to_freqs): Rename to ...
6385 (update_max_bb_count): ... this one.
6386 (expensive_function_p): Use counts.
6387 (estimate_bb_frequencies): Update.
6388 (rebuild_frequencies): Update.
6389 * predict.h (counts_to_freqs): Rename to ...
6390 (update_max_bb_count): ... this one.
6391 * profile.c (compute_branch_probabilities): Add debug info
6392 * tree-inline.c (expand_call_inline): Update debug info.
6393 (optimize_inline_calls): Use update_max_bb_count..
6394 (tree_function_versioning): Use update_max_bb_count..
6395 * value-prof.c (gimple_value_profile_transformations):
6396 Do not use update_max_bb_count.
6398 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6400 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6401 always use frequencies.
6403 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6405 * bb-reorder.c: Remove frequencies from comments.
6406 (better_edge_p): Use profile counts.
6407 (find_traces): Dump profile counts.
6408 (rotate_loop): Use profile counts.
6409 (find_traces_1_round): Likewise.
6410 (connect_better_edge_p): Use counts instead of probabilities for
6412 (copy_bb_p): Drop early check for non-0 frequency.
6413 (sanitize_hot_paths): Update comments.
6415 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6417 * ipa-split.c (struct split_point): Add count.
6418 (consider_split): Do not compute incoming frequency; compute incoming
6419 count and store it to split_point.
6420 (split_function): Set count of the call to split part correctly.
6422 2017-11-13 Carl Love <cel@us.ibm.com>
6424 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6426 2017-11-13 Tom Tromey <tom@tromey.com>
6428 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6430 2017-11-13 Carl Love <cel@us.ibm.com>
6432 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6433 Add support for builtins:
6434 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6435 vector {un,}signed {char,int,short},
6436 vector {un,}signed {char,int,short}) arguments.
6437 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6438 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6439 Add BU_P9V_AV_2 expansions for the builtins.
6440 * config/rs6000/altivec.h (vec_first_match_index,
6441 vec_first_mismatch_index, vec_first_match_or_eos_index,
6442 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6443 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6444 new extern declaration.
6445 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6446 * config/rs6000/vsx.md (first_match_index_<mode>,
6447 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6448 first_mismatch_or_eos_index_<mode>): Add define expand.
6449 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6450 * doc/extend.texi: Update the built-in documenation file for the new
6453 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6455 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6456 into the min/max operations for _Float<N> and _Float<N>X types.
6458 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6461 * dwarf2out.c (do_eh_frame): New static variable.
6462 (dwarf2out_begin_prologue): Set it.
6463 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6465 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6467 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6469 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6471 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6474 * debug.h (dwarf2out_do_eh_frame): Declare.
6475 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6476 (dwarf2out_do_frame): Use it.
6477 (dwarf2out_do_cfi_asm): Likewise.
6478 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6479 (dwarf2out_assembly_start): Likewise.
6480 (dwarf2out_begin_prologue): Fix comment.
6481 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6482 if the target needs either debug or unwind DWARF2 info.
6483 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6484 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6485 -ftrapping-math, -ftrapv and -fwrapv.
6487 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6489 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6490 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6491 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6492 (estimate_edge_size_and_time): Likewise.
6493 (ipa_merge_fn_summary_after_inlining): Likewise.
6494 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6495 (compute_uninlined_call_time): Use sreal_frequency.
6496 (compute_inlined_call_time): Likewise.
6497 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6498 * profile-count.c: Include sreal.h.
6499 (profile_count::to_sreal_scale): New.
6500 * profile-count.h: Forward declare sreal.
6501 (profile_count::to_sreal_scale): Declare.
6503 2017-11-13 Nathan Sidwell <nathan@acm.org>
6505 * diagnostic.c (maybe_line_and_column): New.
6506 (diagnostic_get_location_text): Use it.
6507 (diagnostic_report_current_module): Likewise.
6508 (test_diagnostic_get_location_text): Add tests.
6510 2017-11-13 Luis Machado <luis.machado@linaro.org>
6512 * doc/md.texi (Specifying processor pipeline description): Fix
6513 incorrect latency for the div instruction example.
6515 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6517 PR tree-optimization/78821
6518 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6519 that bit_not_p is the same.
6520 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6521 (split_group): Count precisely bit_not_p bits in each statement.
6522 (invert_op): New function.
6523 (imm_store_chain_info::output_merged_store): Use invert_op to
6524 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6525 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6527 2017-11-13 Martin Liska <mliska@suse.cz>
6529 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6530 (struct source_info): Likewise.
6531 (add_branch_counts): Likewise.
6532 (add_line_counts): Likewise.
6533 (function_summary): Likewise.
6534 (output_intermediate_line): Likewise.
6535 (generate_results): Likewise.
6537 2017-11-13 Martin Liska <mliska@suse.cz>
6539 * gcov.c (struct block_info): Remove typedef for block_t.
6540 (struct line_info): Likewise.
6541 (line_info::has_block): Likewise.
6542 (EXIT_BLOCK): Likewise.
6543 (unblock): Likewise.
6544 (circuit): Likewise.
6545 (get_cycles_count): Likewise.
6546 (process_file): Likewise.
6547 (read_graph_file): Likewise.
6548 (solve_flow_graph): Likewise.
6549 (find_exception_blocks): Likewise.
6550 (add_line_counts): Likewise.
6551 (accumulate_line_info): Likewise.
6552 (output_line_details): Likewise.
6554 2017-11-13 Martin Liska <mliska@suse.cz>
6556 * gcov.c (struct arc_info): Remove typedef for arc_t.
6557 (struct line_info): Likewise.
6558 (add_branch_counts): Likewise.
6559 (output_branch_count): Likewise.
6560 (function_info::~function_info): Likewise.
6561 (circuit): Likewise.
6562 (output_intermediate_line): Likewise.
6563 (read_graph_file): Likewise.
6564 (solve_flow_graph): Likewise.
6565 (find_exception_blocks): Likewise.
6566 (add_line_counts): Likewise.
6567 (accumulate_line_info): Likewise.
6568 (output_line_details): Likewise.
6569 (output_function_details): Likewise.
6571 2017-11-13 Martin Liska <mliska@suse.cz>
6573 * gcov.c (struct function_info): Remove typedef for function_t.
6574 (struct source_info): Likewise.
6575 (source_info::get_functions_at_location): Likewise.
6576 (solve_flow_graph): Likewise.
6577 (find_exception_blocks): Likewise.
6578 (add_line_counts): Likewise.
6579 (output_intermediate_file): Likewise.
6580 (process_file): Likewise.
6581 (generate_results): Likewise.
6582 (release_structures): Likewise.
6583 (read_graph_file): Likewise.
6584 (read_count_file): Likewise.
6585 (accumulate_line_counts): Likewise.
6586 (output_lines): Likewise.
6588 2017-11-13 Martin Liska <mliska@suse.cz>
6590 * gcov.c (function_info::function_info): Remove num_counts
6591 and add vector<gcov_type>.
6592 (function_info::~function_info): Use the vector.
6593 (process_file): Likewise.
6594 (read_graph_file): Likewise.
6595 (read_count_file): Likewise.
6596 (solve_flow_graph): Likewise.
6598 2017-11-13 Martin Liska <mliska@suse.cz>
6600 * gcov.c (function_info::is_artificial): New function.
6601 (process_file): Erase all artificial early.
6602 (generate_results): Skip as all artificial are already
6605 2017-11-13 Martin Liska <mliska@suse.cz>
6607 * gcov.c (read_graph_file): Store to global vector of functions.
6608 (read_count_file): Iterate the vector.
6609 (process_file): Likewise.
6610 (generate_results): Likewise.
6611 (release_structures): Likewise.
6613 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6615 PR tree-optimization/82954
6616 * gimple-ssa-store-merging.c
6617 (imm_store_chain_info::coalesce_immediate_stores): If
6618 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6620 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6622 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6623 Upddate call to ENDIAN_LANE_N.
6624 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6625 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6626 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6627 and use aarch64_endian_lane_rtx.
6628 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6630 2017-11-12 Tom de Vries <tom@codesourcery.com>
6632 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6634 2017-11-12 Tom de Vries <tom@codesourcery.com>
6636 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6638 2017-11-12 Tom de Vries <tom@codesourcery.com>
6640 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6642 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6643 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6644 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6646 2017-11-11 Martin Sebor <msebor@redhat.com>
6649 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6652 * prefic.c (translate_name): Replace strncpy with memcpy to
6653 avoid -Wstringop-truncation.
6655 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6657 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6659 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6661 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6663 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6665 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6668 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6670 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6671 enable generating XXBRH if the value is in a vector register.
6672 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6673 value is in a vector register.
6674 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6675 register to register bswap64's instead of doing the GPR sequence
6676 used on previous machines.
6677 (bswapdi2_xxbrd): New insn.
6678 (bswapdi2_reg): Disallow on ISA 3.0.
6679 (register to register bswap64 splitter): Do not split the insn on
6680 ISA 3.0 systems that use XXBRD.
6682 2017-11-10 Martin Sebor <msebor@redhat.com>
6685 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6687 (find_subframework_file): Same.
6689 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6691 * auto-profile.c (afdo_indirect_call): Drop frequency.
6692 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6693 (cgraph_node::create_edge): Drop frequency argument.
6694 (cgraph_node::create_indirect_edge): Drop frequency argument.
6695 (cgraph_edge::make_speculative): Drop frequency arguments.
6696 (cgraph_edge::resolve_speculation): Do not update frequencies
6697 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6698 (cgraph_node::dump): Check consistency in IPA mode.
6699 (cgraph_edge::maybe_hot_p): Use IPA counter.
6700 (cgraph_edge::verify_count_and_frequency): Rename to ...
6701 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6702 (cgraph_node::verify_node): Update.
6703 * cgraph.h (struct cgraph_edge): Drop frequency.
6704 (cgraph_edge::frequency): New function.
6705 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6707 (cgraph_edge::rebuild_edges): Likewise.
6708 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6709 (duplicate_thunk_for_node): Do not pass frequency.
6710 (cgraph_node::create_clone): Scale only counts.
6711 (cgraph_node::create_virtual_clone): Do not pass frequency.
6712 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6713 (cgraph_node::create_version_clone): Do not pass frequency.
6714 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6715 (cgraph_node::expand_thunk): Do not pass frequency.
6716 (cgraph_node::create_wrapper): Do not pass frequency.
6717 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6719 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6720 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6721 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6722 (gather_caller_stats): Use frequency function.
6723 (ipcp_cloning_candidate_p): Use frequency function.
6724 (ipcp_propagate_stage): Use frequency function.
6725 (get_info_about_necessary_edges): Use frequency function.
6726 (update_profiling_info): Update only IPA profile.
6727 (update_specialized_profile): Use frequency functoin.
6728 (perhaps_add_new_callers): Update only IPA profile.
6729 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6730 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6731 (dump_ipa_call_summary): Use frequency function.
6732 (estimate_edge_size_and_time): Use frequency function.
6733 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6734 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6735 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6736 (update_noncloned_counts): ... ths one; scale counts only.
6737 (clone_inlined_nodes): Do not scale frequency.
6738 (inline_call): Do not pass frequency.
6739 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6740 (compute_inlined_call_time): Use IPA profile.
6741 (want_inline_small_function_p): Use IPA profile.
6742 (want_inline_self_recursive_call_p): Use IPA profile.
6743 (edge_badness): Use IPA profile.
6744 (lookup_recursive_calls): Use IPA profile.
6745 (recursive_inlining): Do not pass frequency.
6746 (resolve_noninline_speculation): Do not update frequency.
6747 (inline_small_functions): Collect max of IPA profile.
6748 (dump_overall_stats): Dump IPA porfile.
6749 (dump_inline_stats): Dump IPA porfile.
6750 (ipa_inline): Collect IPA stats.
6751 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6752 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6753 (ipa_propagate_frequency): Use frequency function.
6754 (ipa_profile): Cleanup.
6755 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6756 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6757 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6758 (input_node): Do not stream frequency.
6759 (input_edge): Do not stream frequency.
6760 (merge_profile_summaries): Scale only IPA profiles.
6761 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6762 * predict.c (drop_profile): Do not recompute frequency.
6763 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6764 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6765 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6766 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6767 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6768 * tree-inline.c (copy_bb): Do not scale frequency.
6769 (expand_call_inline): Do not scale frequency.
6770 (tree_function_versioning): Do not scale frequency.
6771 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6773 2017-11-10 Julia Koval <julia.koval@intel.com>
6775 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6776 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6777 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6778 (_mm256_maskz_gf2p8affine_epi64_epi8)
6779 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6780 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6781 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6782 (__builtin_ia32_vgf2p8affineqb_v32qi)
6783 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6784 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6786 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6790 (arm_option_override): Refactor.
6791 (arm_option_reconfigure_globals): New.
6792 (arm_options_perform_arch_sanity_checks): New.
6793 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6795 (arm_options_perform_arch_sanity_checks): Likewise
6797 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6799 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6800 (power9-qpmul): New.
6801 * rs6000/rs6000.md ("type" attr): Add qmul.
6802 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6803 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6804 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6806 2017-11-10 Martin Sebor <msebor@redhat.com>
6809 * builtins.c (compute_objsize): Handle arrays that
6810 compute_builtin_object_size likes to fail for. Make extern.
6811 * builtins.h (compute_objsize): Declare.
6812 (check_strncpy_sizes): New function.
6813 (expand_builtin_strncpy): Call check_strncpy_sizes.
6814 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6815 -Wstringop-truncation.
6816 (gimple_fold_builtin_strncat): Same.
6817 * gimple.c (gimple_build_call_from_tree): Set call location.
6818 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6819 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6820 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6821 (handle_builtin_strlen): Use strlen_to_stridx.
6822 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6824 Use strlen_to_stridx.
6825 (pass_strlen::execute): Release strlen_to_stridx.
6826 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6827 (-Wstringop-truncation): Document new option.
6829 2017-11-10 Martin Liska <mliska@suse.cz>
6831 PR gcov-profile/82702
6832 * gcov.c (main): Handle intermediate files in a different
6834 (get_gcov_intermediate_filename): New function.
6835 (output_gcov_file): Remove support of intermediate files.
6836 (generate_results): Allocate intermediate file.
6837 (release_structures): Clean-up properly fn_end.
6838 (output_intermediate_file): Start iterating with line 1.
6840 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6842 PR tree-optimization/82929
6843 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6844 ops_swapped_p non-static data member.
6845 (store_immediate_info::store_immediate_info): Clear it.
6846 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6847 ops set ops_swapped_p.
6848 (count_multiple_uses): Handle ops_swapped_p.
6850 2017-11-10 Martin Liska <mliska@suse.cz>
6852 * coverage.c (coverage_init): Stream information about
6853 support of has_unexecuted_blocks.
6854 * doc/gcov.texi: Document that.
6855 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
6856 * gcov.c (read_graph_file): Likewise.
6857 (output_line_beginning): Fix a small issue with
6860 2017-11-10 Bin Cheng <bin.cheng@arm.com>
6862 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
6863 reference of trivial component.
6865 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6868 * gimple-ssa-store-merging.c
6869 (pass_store_merging::terminate_all_aliasing_chains): For
6870 gimple_store_p stmts also call refs_output_dependent_p.
6872 PR rtl-optimization/82913
6873 * compare-elim.c (try_merge_compare): Punt if def_insn is not
6876 2017-11-09 Jeff Law <law@redhat.com>
6878 * vr-values.h: New file with vr_values class.
6879 * tree-vrp.c: Include vr-values.h
6880 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
6881 data objects into the vr_values class.
6882 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
6883 (get_value_range): Make it a member function within vr_values class.
6884 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
6885 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
6886 (op_with_constant_singleton_value_range): Likewise.
6887 (extract_range_for_var_from_comparison_expr): Likewise.
6888 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
6889 (extract_range_from_binary_expr): Likewise.
6890 (extract_range_from_unary_expr): Likewise.
6891 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
6892 (check_for_binary_op_overflow, extract_range_basic): Likewise.
6893 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
6894 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
6895 (compare_name_with_value, compare_names): Likewise.
6896 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
6897 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
6898 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
6899 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
6900 (extract_range_from_phi_node): Likewise.
6901 (simplify_truth_ops_using_ranges): Likewise.
6902 (simplify_div_or_mod_using_ranges): Likewise.
6903 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
6904 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
6905 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
6906 (simplify_float_conversion_using_ranges): Likewise.
6907 (simplify_internal_call_using_ranges): Likewise.
6908 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
6909 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
6910 poitner via x_vr_values for calls into gimple folder.
6911 (vrp_initialize_lattice): Make this the vr_values ctor.
6912 (vrp_free_lattice): Make this the vr_values dtor.
6913 (set_vr_value): New function.
6914 (class vrp_prop): Add vr_values data member. Add various member
6915 functions as well as member functions that delegate to vr_values.
6916 (check_array_ref): Make a member function within vrp_prop class.
6917 (search_for_addr_array, vrp_initialize): Likewise.
6918 (vrp_finalize): Likewise. Revamp to avoid direct access to
6919 vr_value, values_propagated, etc.
6920 (check_array_bounds): Extract vrp_prop class instance pointer from
6921 walk info structure. Use it to call member functions.
6922 (check_all_array_refs): Make a member function within vrp_prop class.
6923 Smuggle class instance pointer via walk info structure.
6924 (x_vr_values): New local static.
6925 (vrp_valueize): Use x_vr_values to get class instance.
6926 (vr_valueize_1): Likewise.
6927 (class vrp_folder): Add vr_values data member. Add various member
6928 functions as well as member functions that delegate to vr_values.
6929 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
6930 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
6931 class instance from vr_values. Use it to call member functions.
6932 (vrp_dom_walker): Add vr_values data member.
6933 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
6934 instance via x_vr_values.
6935 (identify_jump_threads): Accept vr_values as argument. Store
6936 it into the walker structure.
6937 (evrp_dom_walker): Add vr_values class data member. Add various
6939 (evrp_dom_walker::try_find_new_range): Use vr_values data
6940 member to access the memory allocator.
6941 (evrp_dom_walker::before_dom_children): Store vr_values class
6942 instance into the vrp_folder class.
6943 (evrp_dom_walker::push_value_range): Rework to avoid direct
6944 access to num_vr_values and vr_value.
6945 (evrp_dom_walker::pop_value_range): Likewise.
6946 (execute_early_vrp): Remove call to vrp_initialize_lattice.
6947 Use vr_values to get to dump_all_value_ranges member function.
6948 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
6949 and simplify_cond_using_ranges_2 via vrp_prop class instance.
6950 Pass vr_values class instance down to identify_jump_threads.
6951 Remove call to vrp_free_lattice.
6952 (debug_all_value_ranges): Remove.
6954 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
6955 (vrp_folder): Likewise.
6957 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
6958 Get it from the existing bitmap instead.
6959 (vrp_intersect_ranges_1): Likewise.
6961 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6963 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6965 (store_immediate_info::store_immediate_info): Add bitnotp argument,
6966 set bit_not_p to it.
6967 (imm_store_chain_info::coalesce_immediate_stores): Break group
6968 if bit_not_p is different.
6969 (count_multiple_uses, split_group,
6970 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
6971 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
6972 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
6973 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
6974 to store_immediate_info ctor.
6976 2017-11-09 Jim Wilson <jimw@sifive.com>
6978 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
6979 (scan_prog_file): Likewise.
6981 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6983 * bb-reorder.c (max_entry_frequency): Remove.
6984 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
6985 connect_traces, push_to_next_round_p): Remove prototypes.
6986 (find_traces_1_round): Use counts only.
6987 (push_to_next_round_p): Likewise.
6988 (find_traces): Likewise.
6989 (rotate_loop): Likewise.
6990 (find_traces_1_round): Likewise.
6991 (connect_traces): Likewise.
6992 (edge_order): Likewise.
6994 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
6996 * config/arm/arm.c (output_return_instruction): Add comments to
6997 indicate requirement for cmse_nonsecure_entry return to account
6998 for the size of clearing instruction output here.
6999 (thumb_exit): Likewise.
7000 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7001 return in hardfloat mode.
7003 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7005 * config/rs6000/rs6000.c (machine_function): Add a bool,
7006 "toc_is_wrapped_separately".
7007 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7008 if it wasn't explicitly set or unset, we are optimizing for speed, and
7009 doing separate shrink-wrapping.
7010 (rs6000_get_separate_components): Enable the TOC component if
7011 saving the TOC register in the prologue.
7012 (rs6000_components_for_bb): Handle the TOC component.
7013 (rs6000_emit_prologue_components): Store the TOC register where needed.
7014 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7015 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7017 2017-11-09 Martin Jambor <mjambor@suse.cz>
7019 * ipa-param-manipulation.c: New file.
7020 * ipa-param-manipulation.h: Likewise.
7021 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7022 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7023 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7024 (ipa_parm_adjustment): Likewise.
7025 (ipa_parm_adjustment_vec): Likewise.
7026 (ipa_get_vector_of_formal_parms): Moved declaration to
7027 ipa-param-manipulation.h.
7028 (ipa_get_vector_of_formal_parm_types): Likewise.
7029 (ipa_modify_formal_parameters): Likewise.
7030 (ipa_modify_call_arguments): Likewise.
7031 (ipa_combine_adjustments): Likewise.
7032 (ipa_dump_param_adjustments): Likewise.
7033 (ipa_modify_expr): Likewise.
7034 (ipa_get_adjustment_candidate): Likewise.
7035 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7036 ipa-param-manipulation.c.
7037 (ipa_get_vector_of_formal_parm_types): Likewise.
7038 (ipa_modify_formal_parameters): Likewise.
7039 (ipa_modify_call_arguments): Likewise.
7040 (ipa_modify_expr): Likewise.
7041 (get_ssa_base_param): Likewise.
7042 (ipa_get_adjustment_candidate): Likewise.
7043 (index_in_adjustments_multiple_times_p): Likewise.
7044 (ipa_combine_adjustments): Likewise.
7045 (ipa_dump_param_adjustments): Likewise.
7046 * tree-sra.c: Also include ipa-param-manipulation.h
7047 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7050 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7051 Alan Hayward <alan.hayward@arm.com>
7052 David Sherwood <david.sherwood@arm.com>
7054 * doc/sourcebuild.texi (vect_masked_store): Document.
7056 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7057 Alan Hayward <alan.hayward@arm.com>
7058 David Sherwood <david.sherwood@arm.com>
7060 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7062 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7063 Alan Hayward <alan.hayward@arm.com>
7064 David Sherwood <david.sherwood@arm.com>
7066 * doc/sourcebuild.texi (vect_variable_length): Document.
7068 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7069 Alan Hayward <alan.hayward@arm.com>
7070 David Sherwood <david.sherwood@arm.com>
7072 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7074 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7075 Alan Hayward <alan.hayward@arm.com>
7076 David Sherwood <david.sherwood@arm.com>
7078 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7080 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7081 Alan Hayward <alan.hayward@arm.com>
7082 David Sherwood <david.sherwood@arm.com>
7084 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7085 previously undocumented selectors.
7086 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7088 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7090 * doc/rtl.texi (const_vector): Say that elements can be
7091 const_wide_ints too.
7092 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7093 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7094 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7095 * optabs.c (expand_vector_broadcast): Likewise.
7097 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7098 Alan Hayward <alan.hayward@arm.com>
7099 David Sherwood <david.sherwood@arm.com>
7101 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7102 scaled index even if the unscaled address was invalid.
7103 Don't increase the complexity of using a scale in that case.
7105 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7106 Alan Hayward <alan.hayward@arm.com>
7107 David Sherwood <david.sherwood@arm.com>
7109 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7110 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7111 UNITS_PER_WORD bytes.
7112 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7113 based on REGMODE_NATURAL_SIZE of the inner mode.
7114 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7115 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7116 * expmed.c (lowpart_bit_field_p): Divide the value up into
7117 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7118 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7119 whether something is likely to occupy more than one register.
7121 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7124 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7125 profile_count::adjust_for_ipa_scaling.
7126 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7127 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7129 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7131 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7133 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7134 (split_group): Add total_orig and total_new arguments, estimate the
7135 number of statements related to the store group without store merging
7136 and with store merging.
7137 (imm_store_chain_info::output_merged_store): Adjust split_group
7138 callers, punt if estimated number of statements with store merging
7139 is not smaller than estimated number of statements without it.
7141 (handled_load): Remove has_single_use checks.
7142 (pass_store_merging::process_store): Likewise.
7144 2017-11-09 Richard Biener <rguenther@suse.de>
7146 PR tree-optimization/82902
7147 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7149 2017-11-09 Martin Liska <mliska@suse.cz>
7152 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7155 2017-11-09 Martin Liska <mliska@suse.cz>
7157 PR tree-optimization/82669
7158 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7160 2017-11-09 Martin Liska <mliska@suse.cz>
7162 PR gcov-profile/48463
7163 * coverage.c (coverage_begin_function): Output also end locus
7164 of a function and information whether the function is
7166 * gcov-dump.c (tag_function): Parse and print the information.
7167 * gcov.c (INCLUDE_MAP): Add include.
7168 (INCLUDE_SET): Likewise.
7169 (struct line_info): Move earlier in the source file because
7170 of vector<line_info> in function_info structure.
7171 (line_info::line_info): Likewise.
7172 (line_info::has_block): Likewise.
7173 (struct source_info): Add new member index.
7174 (source_info::get_functions_at_location): New function.
7175 (function_info::group_line_p): New function.
7176 (output_intermediate_line): New function.
7177 (output_intermediate_file): Use the mentioned function.
7178 (struct function_start): New.
7179 (struct function_start_pair_hash): Likewise.
7180 (process_file): Add code that identifies group functions.
7181 Assign lines either to global or function scope.
7182 (generate_results): Skip artificial functions.
7183 (find_source): Assign index for each source file.
7184 (read_graph_file): Read new flag artificial and end_line.
7185 (add_line_counts): Assign it either to global of function scope.
7186 (accumulate_line_counts): Isolate core of the function to
7187 accumulate_line_info and call it for both function and global
7189 (accumulate_line_info): New function.
7190 (output_line_beginning): Fix GNU coding style.
7191 (print_source_line): New function.
7192 (output_line_details): Likewise.
7193 (output_function_details): Likewise.
7194 (output_lines): Iterate both source (global) scope and function
7196 (struct function_line_start_cmp): New class.
7197 * doc/gcov.texi: Reflect changes in documentation.
7199 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7202 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7203 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7204 and similarly for not instead of neg.
7206 2017-11-08 Andi Kleen <ak@linux.intel.com>
7208 * config/i386/i386.opt: Add -mforce-indirect-call.
7209 * config/i386/predicates.md: Check for flag_force_indirect_call.
7210 * doc/invoke.texi: Document -mforce-indirect-call
7212 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7214 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7216 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7217 config/riscv/riscv.c (predict.h): New include.
7218 (riscv_slow_unaligned_access_p): No longer static.
7219 (riscv_block_move_straight): Add require.
7220 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7222 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7225 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7226 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7227 nonimmediate_operand predicate for operand 1 instead of
7230 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7232 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7234 * config/aarch64/constraints.md (Uml): New constraint.
7235 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7238 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7240 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7241 of two vec_duplicates into a vec_concat.
7243 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7245 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7246 Simplify vec_merge of vec_duplicate and vec_concat.
7247 * config/aarch64/constraints.md (Utq): New constraint.
7248 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7251 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7253 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7254 Simplify vec_merge of vec_duplicate and const_vector.
7255 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7257 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7258 mode iterator. Update predicate on operand 1 to
7259 handle non-const_vec constants. Delete constraints.
7260 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7262 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7264 PR tree-optimization/78821
7265 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7267 (store_operand_info::store_operand_info): Initialize it to false.
7268 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7269 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7270 store in the group, and if chain_info is non-NULL, to ignore altogether
7272 (compatible_load_p): Fail if bit_not_p does not match.
7273 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7274 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7275 (pass_store_merging::process_store): Adjust
7276 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7277 call terminate_all_aliasing_chains newly when adding a store into
7278 a chain with non-NULL chain_info.
7280 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7282 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7284 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7286 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7288 (aarch64_layout_frame): Initialise emit_frame_chain.
7289 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7290 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7292 2017-11-08 Martin Liska <mliska@suse.cz>
7294 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7295 of gimple_call_internal_p.
7297 2017-11-07 Tom de Vries <tom@codesourcery.com>
7299 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7301 2017-11-07 Tom de Vries <tom@codesourcery.com>
7303 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7306 2017-11-08 Martin Liska <mliska@suse.cz>
7309 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7311 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7313 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7315 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7318 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7319 (*cmp<mode>_ccz_1): New insn with $k alternative.
7322 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7323 RTX_COMM_COMPARE as commutative as well.
7324 (ix86_binary_operator_ok): Formatting fix.
7325 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7326 *<code><mode>3<mask_name><round_saeonly_name>,
7327 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7328 *<s>mul<mode>3_highpart<mask_name>,
7329 *vec_widen_umult_even_v16si<mask_name>,
7330 *vec_widen_umult_even_v8si<mask_name>,
7331 *vec_widen_umult_even_v4si<mask_name>,
7332 *vec_widen_smult_even_v16si<mask_name>,
7333 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7334 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7335 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7336 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7337 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7338 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7339 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7340 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7341 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7342 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7343 ix86_binary_operator_ok. Formatting fixes.
7344 (*<plusminus_insn><mode>3<mask_name><round_name>,
7345 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7348 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7350 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7351 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7352 eq and ne if TARGET_ISEL.
7353 (cmp): New code_iterator.
7354 (UNS, UNSU_, UNSIK): New code_attrs.
7355 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7356 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7358 ("eq<mode>3"): ... this.
7359 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7361 ("ne<mode>3"): ... this.
7363 2017-11-07 Julia Koval <julia.koval@intel.com>
7366 * common/config/i386/i386-common.c
7367 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7368 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7369 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7370 * config/i386/i386.opt: Ditto.
7371 * config/i386/i386.c (ix86_target_string): Ditto.
7372 (ix86_option_override_internal): Ditto.
7373 (ix86_init_mpx_builtins): Move MPX to args2.
7374 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7375 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7376 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7377 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7378 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7379 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7380 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7382 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7385 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7386 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7387 (zero-extendsidi peephole2): Remove peephole.
7389 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7392 * stor-layout.c: Include attribs.h.
7393 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7394 explicit lookup of "aligned" attribute.
7396 2017-11-07 Andrew Waterman <andrew@sifive.com>
7398 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7399 (riscv_expand_block_move): Likewise.
7400 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7402 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7403 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7404 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7405 (riscv_adjust_block_mem): Likewise.
7406 (riscv_block_move_loop): Likewise.
7407 (riscv_expand_block_move): Likewise.
7408 * config/riscv/riscv.md (movmemsi): New pattern.
7410 2017-11-07 Michael Clark <michaeljclark@mac.com>
7412 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7413 (MUSL_DYNAMIC_LINKER): Likewise.
7415 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7417 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7418 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7421 2017-11-07 Richard Biener <rguenther@suse.de>
7423 * match.pd: Fix build.
7425 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7426 Jackson Woodruff <jackson.woodruff@arm.com>
7428 PR tree-optimization/71026
7429 * match.pd: Canonicalize negate in division.
7431 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7434 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7436 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7438 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7439 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7441 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7443 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7444 non-scalar integral types.
7445 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7446 (-(A-B), -(~A)): New transformations.
7448 2017-11-07 Tom de Vries <tom@codesourcery.com>
7450 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7451 semicolon after "do {} while (0)".
7452 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7453 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7454 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7455 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7456 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7458 2017-11-07 Tom de Vries <tom@codesourcery.com>
7460 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7461 after "do {} while (0)".
7462 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7463 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7464 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7465 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7466 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7468 2017-11-07 Tom de Vries <tom@codesourcery.com>
7471 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7473 (arm_rtx_costs_internal): Add missing semicolon after
7474 HANDLE_NARROW_SHIFT_ARITH call.
7476 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7478 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7479 disable isel if it was not set explicitly.
7481 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7483 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7484 (add_type_attribute) likewise.
7486 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7488 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7489 of struct ix86_frame.
7490 (ix86_initial_elimination_offset): Likewise.
7491 (ix86_expand_split_stack_prologue): Likewise.
7493 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7495 * tree-vrp.h (enum value_range_type): Update stale comment.
7497 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7498 Alan Hayward <alan.hayward@arm.com>
7499 David Sherwood <david.sherwood@arm.com>
7501 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7502 (aarch64_expand_vec_perm_const): Take the number of units too.
7503 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7504 (aarch64_expand_vec_perm_const): Likewise.
7505 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7506 (vec_perm<mode>): Update accordingly.
7508 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7509 Alan Hayward <alan.hayward@arm.com>
7510 David Sherwood <david.sherwood@arm.com>
7512 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7513 Take the number of units too.
7514 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7515 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7516 but check for a vector mode before rather than after the call.
7517 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7518 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7519 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7520 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7521 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7522 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7523 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7524 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7525 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7526 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7527 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7528 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7529 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7530 (aarch64_sqdmull2_n<mode>): Update accordingly.
7532 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7533 Alan Hayward <alan.hayward@arm.com>
7534 David Sherwood <david.sherwood@arm.com>
7536 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7537 the number of units too.
7538 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7539 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7540 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7541 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7542 (vec_store_lanesxi<mode>): Update accordingly.
7544 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7545 Alan Hayward <alan.hayward@arm.com>
7546 David Sherwood <david.sherwood@arm.com>
7548 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7549 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7550 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7551 of units rather than the mode.
7552 * config/aarch64/iterators.md (nunits): New mode attribute.
7553 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7554 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7555 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7556 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7557 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7558 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7559 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7560 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7561 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7562 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7563 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7564 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7565 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7566 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7567 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7568 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7569 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7570 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7571 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7572 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7573 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7574 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7575 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7576 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7577 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7578 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7579 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7580 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7581 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7582 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7583 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7584 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7585 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7586 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7587 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7588 (aarch64_simd_vec_setv2di): Likewise.
7590 2017-11-06 Carl Love <cel@us.ibm.com>
7592 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7594 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7595 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7597 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7599 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7600 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7601 Add power 8 macro expansions.
7602 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7603 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7604 power 8 instructions. (VSX_XXBR): Add iterator.
7606 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7608 * config/arm/arm.md (predicable_short_it): Change default to "no",
7609 improve documentation, remove uses that are identical to the default.
7610 (enabled_for_depr_it): Rename to enabled_for_short_it.
7611 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7612 Remove default uses.
7613 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7614 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7615 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7616 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7618 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7621 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7622 float128 helper macros, which are no longer used after deleting
7623 the old 'q' built-in functions, and moving the round to odd
7624 built-in functions to being special built-in functions.
7625 (BU_FLOAT128_2): Likewise.
7626 (BU_FLOAT128_1_HW): Likewise.
7627 (BU_FLOAT128_2_HW): Likewise.
7628 (BU_FLOAT128_3_HW): Likewise.
7629 (FABSQ): Delete old 'q' built-in functions.
7630 (COPYSIGNQ): Likewise.
7631 (SQRTF128_ODD): Move round to odd built-in functions to be
7632 special built-in functions, so that we can handle
7633 -mabi=ieeelongdouble.
7634 (TRUNCF128_ODD): Likewise.
7635 (ADDF128_ODD): Likewise.
7636 (SUBF128_ODD): Likewise.
7637 (MULF128_ODD): Likewise.
7638 (DIVF128_ODD): Likewise.
7639 (FMAF128_ODD): Likewise.
7640 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7641 built-in names to 'f128'.
7642 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7643 old 'q' built-in functions, as the machine independent code for
7644 'f128' built-in functions handles this.
7645 (rs6000_expand_builtin): Add expansion for float128 round to odd
7646 functions, keying off on -mabi=ieeelongdouble of whether to use
7647 the KFmode or TFmode variant.
7648 (rs6000_init_builtins): Initialize the _Float128 round to odd
7650 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7651 _Float128 'q' built-in functions are now mapped into the new
7652 'f128' built-in functions.
7654 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7656 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7657 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7659 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7662 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7663 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7664 * toplev.c: Include "ipa-fnsummary.h".
7665 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7667 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7669 PR tree-optimization/82838
7670 * gimple-ssa-store-merging.c
7671 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7672 on a separate gimple_seq which is then appended to seq.
7674 2017-11-06 Jeff Law <law@redhat.com>
7677 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7678 (get_probe_interval): New functions.
7679 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7680 (ix86_adjust_stack_and_probe): Likewise.
7681 (output_adjust_stack_and_probe): Likewise.
7682 (ix86_emit_probe_stack_range): Likewise.
7683 (ix86_expand_prologue): Likewise.
7685 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7687 PR tree-optimization/82816
7688 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7689 if the modes of the two types are the same.
7690 (convert_plusminus_to_widen): Likewise.
7692 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7694 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7696 (usadv16qi): New define_expand.
7697 (usadv8hi): New define_expand.
7699 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7702 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7704 (inline_transform): Likewise.
7705 * predict.c (counts_to_freqs): Remove useless conditional.
7706 * profile-count.h (profile_count::apply_scale): Move sanity check.
7707 * tree-inline.c (copy_bb): Always scale.
7708 (copy_cfg_body): Likewise.
7710 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7713 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7715 (*cmp_ite0): Add enabled_for_depr_it attribute.
7716 (*cmp_ite1): Likewise.
7718 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7720 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7723 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7725 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7726 return the value by pointer.
7727 (extract_range_from_multiplicative_op_1): Update accordingly.
7728 Return as soon as an operation fails.
7730 2017-11-05 Tom de Vries <tom@codesourcery.com>
7733 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7734 (DEF_SANITIZER_BUILTIN): ... here.
7735 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7736 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7738 2017-11-05 Tom de Vries <tom@codesourcery.com>
7741 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7743 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7744 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7745 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7747 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7748 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7749 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7751 2017-11-05 Tom de Vries <tom@codesourcery.com>
7754 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7757 2017-11-04 Michael Clark <michaeljclark@mac.com>
7759 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7760 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7762 (*addsi3_extended): Likewise.
7763 (*addsi3_extended2): Likewise.
7764 (<optab>si3): Likewise.
7765 (<optab>di3): Likewise.
7766 (<optab><mode>3): Likewise.
7767 (<*optabe>si3_internal): Likewise.
7768 (zero_extendqi<SUPERQI:mode>2): Likewise.
7769 (*add<mode>hi3): Likewise.
7770 (*xor<mode>hi3): Likewise.
7771 (<optab>di3): Likewise.
7772 (*<optab>si3_extend): Likewise.
7773 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7774 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7775 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7777 2017-11-04 Andrew Waterman <andrew@sifive.com>
7779 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7780 TARGET_STRICT_ALIGN based upon -mtune argument.
7782 2017-11-04 Andrew Waterman <andrew@sifive.com>
7784 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7786 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7788 * config/i386/i386.c (choose_basereg): Use optional scratch
7789 register and add assertion.
7790 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7791 needed, and don't allocate stack.
7792 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7793 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7794 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7796 2017-11-03 Jeff Law <law@redhat.com>
7798 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7799 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7800 to probe at the start of a noreturn function.
7802 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7804 PR tree-optimization/78821
7805 * gimple-ssa-store-merging.c: Update the file comment.
7806 (MAX_STORE_ALIAS_CHECKS): Define.
7807 (struct store_operand_info): New type.
7808 (store_operand_info::store_operand_info): New constructor.
7809 (struct store_immediate_info): Add rhs_code and ops data members.
7810 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7811 and op1r arguments to the ctor, initialize corresponding data members.
7812 (struct merged_store_group): Add load_align_base and load_align
7814 (merged_store_group::merged_store_group): Initialize them.
7815 (merged_store_group::do_merge): Update them.
7816 (merged_store_group::apply_stores): Pick the constant for
7817 encode_tree_to_bitpos from one of the two operands, or skip
7818 encode_tree_to_bitpos if neither operand is a constant.
7819 (class pass_store_merging): Add process_store method decl. Remove
7820 bool argument from terminate_all_aliasing_chains method decl.
7821 (pass_store_merging::terminate_all_aliasing_chains): Remove
7822 var_offset_p argument and corresponding handling.
7823 (stmts_may_clobber_ref_p): New function.
7824 (compatible_load_p): New function.
7825 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7826 if there is overlap and rhs_code is not INTEGER_CST. For
7827 non-overlapping stores terminate group if rhs is not mergeable.
7828 (get_alias_type_for_stmts): Change first argument from
7829 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7830 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7831 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7833 (get_location_for_stmts): Change first argument from
7834 auto_vec<gimple *> & to vec<gimple *> &.
7835 (struct split_store): Remove orig_stmts data member, add orig_stores.
7836 (split_store::split_store): Create orig_stores rather than orig_stmts.
7837 (find_constituent_stmts): Renamed to ...
7838 (find_constituent_stores): ... this. Change second argument from
7839 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7840 to info structures rather than the statements.
7841 (split_group): Rename ALLOW_UNALIGNED argument to
7842 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7843 it. Adjust find_constituent_stores caller.
7844 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7845 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7846 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7847 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7848 (mem_valid_for_store_merging): New function.
7849 (handled_load): New function.
7850 (pass_store_merging::process_store): New method.
7851 (pass_store_merging::execute): Use process_store method. Adjust
7852 terminate_all_aliasing_chains caller.
7854 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7856 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
7857 Return true for more constants, symbols and label references.
7858 (aarch64_valid_floating_const): Remove unused function.
7860 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7863 * config/aarch64/aarch64.c (aarch64_layout_frame):
7864 Undo forcing of LR at bottom of frame.
7866 2017-11-03 Jeff Law <law@redhat.com>
7869 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
7870 for int_registers_saved.
7872 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
7873 extracted from tree-ssa-dom.c.
7874 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
7875 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
7876 (record_equivalences_from_incoming_edge): Add additional argument
7877 to single_pred_edge_ignoring_loop_edges call.
7878 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
7879 (uncprop_dom_walker::before_dom_children): Add additional argument
7880 to single_pred_edge_ignoring_loop_edges call.
7881 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
7882 single_pred_edge_ignoring_loop_edges rather than open coding.
7883 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
7885 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
7887 * match.pd (-(-A)): Rewrite.
7889 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7891 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
7892 (rs6000_emit_int_cmove): New declaration.
7893 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
7894 (rs6000_emit_sISEL): Delete.
7895 (rs6000_emit_int_cmove): Make non-static.
7896 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
7897 instead of rs6000_emit_sISEL.
7899 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
7901 * asan.c (create_cond_insert_point): Maintain profile.
7902 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
7904 * basic-block.h (struct basic_block_def): Remove frequency.
7905 (EDGE_FREQUENCY): Use to_frequency
7906 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
7908 (find_traces): Update to use to_frequency.
7909 (find_traces_1_round): Likewise; use only IPA counts.
7910 (bb_to_key): Likewise.
7911 (connect_traces): Use IPA counts only.
7912 (copy_bb_p): Update to use to_frequency.
7913 (fix_up_crossing_landing_pad): Likewise.
7914 (sanitize_hot_paths): Likewise.
7915 * bt-load.c (basic_block_freq): Likewise.
7916 * cfg.c (init_flow): Set count_max to uninitialized.
7917 (check_bb_profile): Remove frequencies; check counts.
7918 (dump_bb_info): Do not dump frequencies.
7919 (update_bb_profile_for_threading): Update counts only.
7920 (scale_bbs_frequencies_int): Likewise.
7921 (MAX_SAFE_MULTIPLIER): Remove.
7922 (scale_bbs_frequencies_gcov_type): Update counts only.
7923 (scale_bbs_frequencies_profile_count): Update counts only.
7924 (scale_bbs_frequencies): Update counts only.
7925 * cfg.h (struct control_flow_graph): Add count-max.
7926 (update_bb_profile_for_threading): Update prototype.
7927 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
7928 (find_many_sub_basic_blocks): Likewise.
7929 * cfgcleanup.c (try_forward_edges): Likewise.
7930 (try_crossjump_to_edge): Likewise.
7931 * cfgexpand.c (expand_gimple_cond): Likewise.
7932 (expand_gimple_tailcall): Likewise.
7933 (construct_init_block): Likewise.
7934 (construct_exit_block): Likewise.
7935 * cfghooks.c (verify_flow_info): Check consistency of counts.
7936 (dump_bb_for_graph): Do not dump frequencies.
7937 (split_block_1): Do not update frequencies.
7938 (split_edge): Do not update frequencies.
7939 (make_forwarder_block): Do not update frequencies.
7940 (duplicate_block): Do not update frequencies.
7941 (account_profile_record): Do not update frequencies.
7942 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
7943 for global heuristics.
7944 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
7945 (expected_loop_iterations_unbounded): Use counts only.
7946 * cfgloopmanip.c (scale_loop_profile): Simplify.
7947 (create_empty_loop_on_edge): Simplify
7949 (duplicate_loop_to_header_edge): Simplify
7950 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
7951 (update_br_prob_note): Take care of removing note when profile
7953 (relink_block_chain): Do not dump frequency.
7954 (rtl_account_profile_record): Use to_frequency.
7955 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
7956 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
7957 (cgraph_update_edges_for_call_stmt_node): Likewise.
7958 (cgraph_edge::verify_count_and_frequency): Update.
7959 (cgraph_node::verify_node): Temporarily disable frequency verification.
7960 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
7961 to_cgraph_frequency.
7962 (cgraph_edge::rebuild_edges): Convert to ipa counts.
7963 * cgraphunit.c (init_lowered_empty_function): Do not initialize
7965 (cgraph_node::expand_thunk): Update profile.
7966 * except.c (dw2_build_landing_pads): Do not update frequency.
7967 * final.c (compute_alignments): Use to_frequency.
7968 (dump_basic_block_info): Do not dump frequency.
7969 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
7970 (dump_gimple_bb_header): Do not dump frequency.
7971 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
7973 * gimple-streamer-in.c (input_bb): Do not stream frequency.
7974 * gimple-streamer-out.c (output_bb): Do not stream frequency.
7975 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
7976 (init_before_recovery): Do not update frequency.
7977 (sched_create_recovery_edges): Do not update frequency.
7978 * hsa-gen.c (convert_switch_statements): Do not update frequency.
7979 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
7980 (ipa_cp_c_finalize): Set max_count to uninitialized.
7981 * ipa-fnsummary.c (get_minimal_bb): Use counts.
7982 (param_change_prob): Use counts.
7983 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
7985 * ipa-split.c (consider_split): Use to_frequency.
7986 (split_function): Use to_frequency.
7987 * ira-build.c (loop_compare_func): Likewise.
7988 (mark_loops_for_removal): Likewise.
7989 (mark_all_loops_for_removal): Likewise.
7990 * loop-doloop.c (doloop_modify): Do not update frequency.
7991 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
7993 * lto-streamer-in.c (input_function): Update count_max.
7994 * omp-expand.c (expand_omp_taskreg): Update count_max.
7995 * omp-simd-clone.c (simd_clone_adjust): Update profile.
7996 * predict.c (maybe_hot_frequency_p): Use to_frequency.
7997 (maybe_hot_count_p): Use ipa counts only.
7998 (maybe_hot_bb_p): Simplify.
7999 (maybe_hot_edge_p): Simplify.
8000 (probably_never_executed): Do not take frequency argument.
8001 (probably_never_executed_bb_p): Do not pass frequency.
8002 (probably_never_executed_edge_p): Likewise.
8003 (combine_predictions_for_bb): Check that profile is nonzero.
8004 (propagate_freq): Do not set frequency.
8005 (drop_profile): Simplify.
8006 (counts_to_freqs): Simplify.
8007 (expensive_function_p): Use to_frequency.
8008 (propagate_unlikely_bbs_forward): Simplify.
8009 (determine_unlikely_bbs): Simplify.
8010 (estimate_bb_frequencies): Add hack to silence graphite issues.
8011 (compute_function_frequency): Use ipa counts.
8012 (pass_profile::execute): Update.
8013 (rebuild_frequencies): Use counts only.
8014 (force_edge_cold): Use counts only.
8015 * profile-count.c (profile_count::dump): Dump new count types.
8016 (profile_count::differs_from_p): Check compatiblity.
8017 (profile_count::to_frequency): New function.
8018 (profile_count::to_cgraph_frequency): New function.
8019 * profile-count.h (struct function): Declare.
8020 (enum profile_quality): Add profile_guessed_local and
8021 profile_guessed_global0.
8022 (class profile_proability): Decrease number of bits to 29;
8023 update from_reg_br_prob_note and to_reg_br_prob_note.
8024 (class profile_count: Update comment; decrease number of bits
8025 to 61. Check compatibility.
8026 (profile_count::compatible_p): New private member function.
8027 (profile_count::ipa_p): New member function.
8028 (profile_count::operator<): Handle global zero correctly.
8029 (profile_count::operator>): Handle global zero correctly.
8030 (profile_count::operator<=): Handle global zero correctly.
8031 (profile_count::operator>=): Handle global zero correctly.
8032 (profile_count::nonzero_p): New member function.
8033 (profile_count::force_nonzero): New member function.
8034 (profile_count::max): New member function.
8035 (profile_count::apply_scale): Handle IPA scalling.
8036 (profile_count::guessed_local): New member function.
8037 (profile_count::global0): New member function.
8038 (profile_count::ipa): New member function.
8039 (profile_count::to_frequency): Declare.
8040 (profile_count::to_cgraph_frequency): Declare.
8041 * profile.c (OVERLAP_BASE): Delete.
8042 (compute_frequency_overlap): Delete.
8043 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8044 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8045 * sched-ebb.c (rank): Use counts only.
8046 * shrink-wrap.c (handle_simple_exit): Use counts only.
8047 (try_shrink_wrapping): Use counts only.
8048 (place_prologue_for_one_component): Use counts only.
8049 * tracer.c (find_best_predecessor): Use to_frequency.
8050 (find_trace): Use to_frequency.
8051 (tail_duplicate): Use to_frequency.
8052 * trans-mem.c (expand_transaction): Do not update frequency.
8053 * tree-call-cdce.c: Do not update frequency.
8054 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8055 (gimple_merge_blocks): Likewise.
8056 (gimple_split_edge): Likewise.
8057 (gimple_duplicate_sese_region): Likewise.
8058 (gimple_duplicate_sese_tail): Likewise.
8059 (move_sese_region_to_fn): Likewise.
8060 (gimple_account_profile_record): Likewise.
8061 (insert_cond_bb): Likewise.
8062 * tree-complex.c (expand_complex_div_wide): Likewise.
8063 * tree-eh.c (lower_resx): Update profile.
8064 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8066 (initialize_cfun): Do not initialize frequencies
8067 (freqs_to_counts): Delete.
8068 (copy_cfg_body): Ignore count parameter.
8069 (copy_body): Update.
8070 (expand_call_inline): Update count_max.
8071 (optimize_inline_calls): Update count_max.
8072 (tree_function_versioning): Update count_max.
8073 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8074 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8076 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8077 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8078 (try_peel_loop): Likewise.
8079 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8081 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8082 (tree_transform_and_unroll_loop): Do not use frequencies
8083 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8084 Use reliable prediction only.
8085 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8086 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8087 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8088 probability scaling.
8089 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8090 not update frequency
8091 (any_remaining_duplicated_blocks): Likewise.
8092 (update_profile): Likewise.
8093 (estimated_freqs_path): Delete.
8094 (freqs_to_counts_path): Delete.
8095 (clear_counts_path): Delete.
8096 (ssa_fix_duplicate_block_edges): Likewise.
8097 (duplicate_thread_path): Likewise.
8098 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8099 * tree-tailcall.c (decrease_profile): Do not update frequency.
8100 (eliminate_tail_call): Likewise.
8101 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8102 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8103 (optimize_mask_stores): Likewise.
8104 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8105 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8106 (ubsan_expand_ptr_ifn): Update profile.
8107 * value-prof.c (gimple_ic): Simplify.
8108 * value-prof.h (gimple_ic): Update prototype.
8109 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8110 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8112 (want_inline_self_recursive_call_p): Likewise.
8113 (resolve_noninline_speculation): Only cummulate defined counts.
8114 (inline_small_functions): Use nonzero_p.
8115 (ipa_inline): Do not access freed node.
8117 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8119 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8120 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8122 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8124 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8125 non-legitimate address.
8127 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8129 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8130 (*lt0_<mode>di, *lt0_<mode>si): New.
8132 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8134 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8135 TARGET_PAIRED_FLOAT.
8137 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8138 Jim Wilson <jim.wilson@linaro.org>
8140 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8141 * config/aarch64/aarch64-tune.md: Regenerated.
8142 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8143 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8145 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8147 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8148 (arc_expand_prologue): Restore blink for millicode.
8149 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8151 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8154 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8155 gen_vec_duplicate after forcing the scalar into a register.
8157 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8159 * combine (try_combine): Print the insns input to try_combine to the
8162 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8165 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8166 Remove second argument from aarch64_process_target_attr call.
8167 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8169 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8171 (aarch64_handle_attr_arch): Remove second argument.
8172 (aarch64_handle_attr_cpu): Ditto.
8173 (aarch64_handle_attr_tune): Ditto.
8174 (aarch64_handle_attr_isa_flags): Ditto.
8175 (aarch64_process_one_target_attr): Ditto.
8176 (aarch64_process_target_attr): Ditto.
8177 (aarch64_option_valid_attribute_p): Remove second argument.
8178 on aarch64_process_target_attr call.
8180 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8182 * diagnostic.c: Include "selftest-diagnostic.h".
8183 (selftest::assert_location_text): New function.
8184 (selftest::test_diagnostic_get_location_text): New function.
8185 (selftest::diagnostic_c_tests): Call it.
8187 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8189 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8190 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8191 (class selftest::test_diagnostic_context): Move to...
8192 * selftest-diagnostic.c: New file.
8193 * selftest-diagnostic.h: New file.
8195 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8197 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8198 offset range for FT32B.
8199 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8200 * config/ft32/ft32.md: Add TARGET_NOPM.
8201 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8202 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8204 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8206 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8208 2017-11-02 Jeff Law <law@redhat.com>
8210 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8211 virtual keyword on FINAL OVERRIDE members.
8213 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8214 virtuals together. Add virtual destructor.
8215 (substitute_and_fold_engine): Similarly.
8217 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8219 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8221 2017-11-02 Richard Biener <rguenther@suse.de>
8223 PR tree-optimization/82795
8224 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8226 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8228 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8230 (gcc_GAS_CHECK_FEATURE): Remove.
8231 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8232 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8233 * configure: Regenerate.
8235 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8237 * config/arc/arc.c (hwloop_optimize): Account for empty
8240 2017-11-02 Richard Biener <rguenther@suse.de>
8243 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8244 Truncate ARRAY_REF index and element size.
8246 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8248 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8250 2017-11-01 Jeff Law <law@redhat.com>
8252 * tree-ssa-ccp.c (ccp_folder): New class derived from
8253 substitute_and_fold_engine.
8254 (ccp_folder::get_value): New member function.
8255 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8256 (ccp_fold_stmt): Remove prototype.
8257 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8258 * tree-ssa-copy.c (copy_folder): New class derived from
8259 substitute_and_fold_engine.
8260 (copy_folder::get_value): Renamed from get_value.
8261 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8262 * tree-vrp.c (vrp_folder): New class derived from
8263 substitute_and_fold_engine.
8264 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8265 (vrp_folder::get_value): New member function.
8266 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8267 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8268 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8269 provide a class interface to folder/substitute routines.
8270 (ssa_prop_fold_stmt_fn): Remove typedef.
8271 (ssa_prop_get_value_fn): Likewise.
8272 (subsitute_and_fold): Remove prototype.
8273 (replace_uses_in): Likewise.
8274 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8275 Renamed from replace_uses_in. Call the virtual member function
8276 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8277 (substitute_and_fold_dom_walker): Remove initialization of
8278 data member entries for calbacks. Add substitute_and_fold_engine
8279 member and initialize it.
8280 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8281 member functions for get_value, replace_phi_args_in c
8282 replace_uses_in, and fold_stmt calls.
8283 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8284 substitute_and_fold. Remove assert. Update ctor call.
8286 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8287 (ssa_prop_visit_phi_fn): Likewise.
8288 (class ssa_propagation_engine): New class to provide an interface
8290 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8292 (ssa_prop_visit_phi): Likewise.
8293 (ssa_propagation_engine::simulate_stmt): Moved into class.
8294 Call visit_phi/visit_stmt from the class rather than via
8295 file scoped static variables.
8296 (ssa_propagation_engine::simulate_block): Moved into class.
8297 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8298 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8299 set file scoped statics for the visit_stmt/visit_phi callbacks.
8300 * tree-complex.c (complex_propagate): New class derived from
8301 ssa_propagation_engine.
8302 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8303 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8304 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8306 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8307 ssa_propagation_engine.
8308 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8309 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8310 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8311 * tree-ssa-copy.c (copy_prop): New class derived from
8312 ssa_propagation_engine.
8313 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8314 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8315 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8316 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8317 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8318 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8319 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8321 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8323 PR rtl-optimization/82778
8324 PR rtl-optimization/82597
8325 * compare-elim.c (struct comparison): Add in_a_setter field.
8326 (find_comparison_dom_walker::before_dom_children): Remove killed
8327 bitmap and df_simulate_find_defs call, instead walk the defs.
8328 Compute last_setter and initialize in_a_setter. Merge definitions
8329 with first initialization for a few variables.
8330 (try_validate_parallel): Use insn_invalid_p instead of
8331 recog_memoized. Return insn rather than just the pattern.
8332 (try_merge_compare): Fix up comment. Don't uselessly test if
8333 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8335 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8338 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8339 Alan Hayward <alan.hayward@arm.com>
8340 David Sherwood <david.sherwood@arm.com>
8342 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8343 aarch64_hard_regno_nregs to get the number of registers
8346 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8347 Alan Hayward <alan.hayward@arm.com>
8348 David Sherwood <david.sherwood@arm.com>
8350 * config/aarch64/constraints.md (Upl): Rename to...
8352 * config/aarch64/aarch64.md
8353 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8356 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8357 Alan Hayward <alan.hayward@arm.com>
8358 David Sherwood <david.sherwood@arm.com>
8360 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8361 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8364 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8365 Alan Hayward <alan.hayward@arm.com>
8366 David Sherwood <david.sherwood@arm.com>
8368 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8369 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8370 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8372 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8373 of a single element.
8374 * config/aarch64/iterators.md: Add a comment above the permute
8375 unspecs to say that they are generated directly by
8376 aarch64_expand_vec_perm_const.
8377 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8379 2017-11-01 Nathan Sidwell <nathan@acm.org>
8381 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8383 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8385 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8386 and medany code models, and describe what they do.
8388 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8390 Revert accidental duplicate:
8392 * combine.c (can_change_dest_mode): Reject changes in
8393 REGMODE_NATURAL_SIZE.
8395 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8397 PR rtl-optimization/64682
8398 PR rtl-optimization/69567
8399 PR rtl-optimization/69737
8400 PR rtl-optimization/82683
8401 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8402 register mentioned in the note, drop the note, unless it came from I3,
8403 in which case it should go to I3 again.
8405 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8407 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8408 and return false if not.
8409 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8411 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8413 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8414 range is known to be empty.
8416 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8417 Alan Hayward <alan.hayward@arm.com>
8418 David Sherwood <david.sherwood@arm.com>
8420 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8421 and CONST_VECTOR_NUNITS instead of computing the number of units from
8422 the byte sizes of the vector and element.
8423 (simplify_binary_operation_1): Likewise.
8424 (simplify_const_binary_operation): Likewise.
8425 (simplify_ternary_operation): Likewise.
8427 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8428 Alan Hayward <alan.hayward@arm.com>
8429 David Sherwood <david.sherwood@arm.com>
8431 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8432 (int_mem_offset): ...this new function.
8433 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8434 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8435 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8438 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8439 Alan Hayward <alan.hayward@arm.com>
8440 David Sherwood <david.sherwood@arm.com>
8442 * lower-subreg.c (interesting_mode_p): New function.
8443 (compute_costs, find_decomposable_subregs, decompose_register)
8444 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8445 (resolve_clobber, dump_choices): Use it.
8447 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8448 Alan Hayward <alan.hayward@arm.com>
8449 David Sherwood <david.sherwood@arm.com>
8451 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8453 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8454 Alan Hayward <alan.hayward@arm.com>
8455 David Sherwood <david.sherwood@arm.com>
8457 * alias.c (find_base_value, find_base_term): Only process integer
8458 truncations. Check the precision rather than the size.
8460 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8461 Alan Hayward <alan.hayward@arm.com>
8462 David Sherwood <david.sherwood@arm.com>
8464 * machmode.h (is_narrower_int_mode): New function
8465 * optabs.c (expand_float, expand_fix): Use it.
8466 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8468 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8469 Alan Hayward <alan.hayward@arm.com>
8470 David Sherwood <david.sherwood@arm.com>
8472 * rtl.h (narrower_subreg_mode): New function.
8473 * ira-color.c (update_costs_from_allocno): Use it.
8475 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8476 Alan Hayward <alan.hayward@arm.com>
8477 David Sherwood <david.sherwood@arm.com>
8479 * optabs-query.h (convert_optab_p): New function, split out from...
8480 (convert_optab_handler): ...here.
8481 (widening_optab_handler): Delete.
8482 (find_widening_optab_handler): Remove permit_non_widening parameter.
8483 (find_widening_optab_handler_and_mode): Likewise. Provide an
8484 override that operates on mode class wrappers.
8485 * optabs-query.c (widening_optab_handler): Delete.
8486 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8487 parameter. Assert that the two modes are the same class and that
8488 the "from" mode is narrower than the "to" mode. Use
8489 convert_optab_handler instead of widening_optab_handler.
8490 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8491 instead of widening_optab_handler.
8492 * expr.c (expand_expr_real_2): Update calls to
8493 find_widening_optab_handler.
8494 * optabs.c (expand_widen_pattern_expr): Likewise.
8495 (expand_binop_directly): Take the insn_code as a parameter.
8496 (expand_binop): Only call find_widening_optab_handler for
8497 conversion optabs; use optab_handler otherwise. Update calls
8498 to find_widening_optab_handler and expand_binop_directly.
8499 Use convert_optab_handler instead of widening_optab_handler.
8500 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8501 find_widening_optab_handler and use scalar_mode rather than
8503 (convert_plusminus_to_widen): Likewise.
8505 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8506 Alan Hayward <alan.hayward@arm.com>
8507 David Sherwood <david.sherwood@arm.com>
8509 * machmode.h (fixed_size_mode): New class.
8510 * rtl.h (get_pool_mode): Return fixed_size_mode.
8511 * gengtype.c (main): Add fixed_size_mode.
8512 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8513 (get_raw_arg_mode): Likewise.
8514 * doc/tm.texi: Regenerate.
8515 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8516 * targhooks.c (default_get_reg_raw_mode): Likewise.
8517 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8518 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8519 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8520 (msp430_get_raw_result_mode): Likewise.
8521 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8522 * dbxout.c (dbxout_parms): Require fixed-size modes.
8523 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8524 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8525 * omp-low.c (lower_oacc_reductions): Likewise.
8526 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8527 (simplify_subreg): Update accordingly.
8528 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8529 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8530 that aren't fixed-size.
8531 (get_pool_mode): Return a fixed_size_mode.
8532 (output_constant_pool_2): Take a fixed_size_mode.
8534 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8535 Alan Hayward <alan.hayward@arm.com>
8536 David Sherwood <david.sherwood@arm.com>
8538 * doc/rtl.texi (vec_series): Document.
8539 (const): Say that the operand can be a vec_series.
8540 * rtl.def (VEC_SERIES): New rtx code.
8541 * rtl.h (const_vec_series_p_1): Declare.
8542 (const_vec_series_p): New function.
8543 * emit-rtl.h (gen_const_vec_series): Declare.
8544 (gen_vec_series): Likewise.
8545 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8546 (gen_vec_series): Likewise.
8547 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8548 * simplify-rtx.c (simplify_unary_operation): Handle negations
8550 (simplify_binary_operation_series): New function.
8551 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8552 (test_vector_ops_series): New function.
8553 (test_vector_ops): Call it.
8554 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8555 gen_const_vec_series.
8556 (altivec_lvsr): Likewise.
8557 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8559 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8560 Alan Hayward <alan.hayward@arm.com>
8561 David Sherwood <david.sherwood@arm.com>
8563 * doc/rtl.texi (const): Update description of address constants.
8564 Say that vector constants are allowed too.
8565 * common.md (E, F): Use CONSTANT_P instead of checking for
8567 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8568 checking for CONST_VECTOR.
8569 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8571 * expr.c (expand_expr_real_2): Check for vector modes instead
8572 of checking for CONST_VECTOR.
8573 * rtl.h (const_vec_p): New function.
8574 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8575 (unwrap_const_vec_duplicate): Handle them here too.
8577 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8578 David Malcolm <dmalcolm@redhat.com>
8579 Alan Hayward <alan.hayward@arm.com>
8580 David Sherwood <david.sherwood@arm.com>
8582 * rtl.h (vec_duplicate_p): New function.
8583 * selftest-rtl.c (assert_rtx_eq_at): New function.
8584 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8585 (assert_rtx_eq_at): Declare.
8586 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8587 * selftest-run-tests.c (selftest::run_tests): Call it.
8588 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8589 (simplify_unary_operation_1): Recursively handle vector duplicates.
8590 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8592 (simplify_subreg): Handle subregs of vector duplicates.
8593 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8594 (selftest::simplify_rtx_c_tests): New functions.
8596 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8597 Alan Hayward <alan.hayward@arm.com>
8598 David Sherwood <david.sherwood@arm.com>
8600 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8601 (gen_vec_duplicate): Likewise.
8602 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8604 (gen_const_vector): ...here.
8605 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8606 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8607 whose elements are all equal.
8608 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8609 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8610 (simplify_relational_operation): Likewise.
8611 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8613 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8614 (aarch64_expand_vector_init): Likewise.
8615 * config/arm/arm.c (neon_vdup_constant): Likewise.
8616 (neon_expand_vector_init): Likewise.
8617 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8618 (arm_block_set_unaligned_vect): Likewise.
8619 (arm_block_set_aligned_vect): Likewise.
8620 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8621 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8622 (expand_vec_perm_even_odd_pack): Likewise.
8623 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8624 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8625 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8626 gen_const_vec_duplicate.
8627 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8628 * config/mips/mips.c (mips_gen_const_int_vector): Use
8629 gen_const_vec_duplicate.
8630 (mips_expand_vector_init): Use CONST0_RTX.
8631 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8632 (define_split): Use gen_const_vec_duplicate.
8633 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8634 (define_split): Use gen_const_vec_duplicate.
8635 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8636 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8637 * config/spu/spu.c (spu_const): Likewise.
8639 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8640 Alan Hayward <alan.hayward@arm.com>
8641 David Sherwood <david.sherwood@arm.com>
8643 * combine.c (can_change_dest_mode): Reject changes in
8644 REGMODE_NATURAL_SIZE.
8646 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8648 * configure.ac (--enable-libssp): New.
8649 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8650 trying to determine target-specific default. Adjust indentation.
8651 * configure: Regenerated.
8652 * doc/install.texi (Configuration): Expand --disable-libssp
8655 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8657 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8660 2017-10-31 Martin Jambor <mjambor@suse.cz>
8663 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8665 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8667 * auto-profile.c (autofdo_source_profile::read): Use
8668 UNKNOWN_LOCATION rather than 0.
8669 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8670 (warning_at): ...this overload.
8671 (warning_at_rich_loc_n): Rename to...
8672 (warning_n): ...this overload.
8673 (error_at_rich_loc): Rename to...
8674 (error_at): ...this overload.
8675 (pedwarn_at_rich_loc): Rename to...
8676 (pedwarn): ...this overload.
8677 (permerror_at_rich_loc): Rename to...
8678 (permerror): ...this overload.
8679 (inform_at_rich_loc): Rename to...
8680 (inform): ...this overload.
8681 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8682 (diagnostic_n_impl_richloc): Rename to...
8683 (diagnostic_n_impl): ...this rich_location *-based decl.
8684 (inform_at_rich_loc): Rename to...
8685 (inform): ...this, and add an assertion.
8686 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8687 (warning_at_rich_loc): Rename to...
8688 (warning_at): ...this, and add an assertion.
8689 (warning_at_rich_loc_n): Rename to...
8690 (warning_n): ...this, and add an assertion.
8691 (warning_n): Update location_t-based implementation for removal of
8692 location_t-based diagnostic_n_impl.
8693 (pedwarn_at_rich_loc): Rename to...
8694 (pedwarn): ...this, and add an assertion.
8695 (permerror_at_rich_loc): Rename to...
8696 (permerror): ...this, and add an assertion.
8697 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8698 (error_at_rich_loc): Rename to...
8699 (error_at): ...this, and add an assertion.
8700 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8701 (driver::do_spec_on_infiles): Likewise.
8702 * substring-locations.c (format_warning_va): Update for renaming
8703 of inform_at_rich_loc.
8705 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8707 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8708 _Float<N>X built-in functions so that the variant without the
8709 "__builtin_" prefix is only enabled for the GNU C and Objective C
8710 languages when they are in non-strict ANSI/ISO mode.
8711 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8712 * target.def (floatn_builtin_p): Add a target hook to control
8713 whether _Float<N> and _Float<N>X built-in functions without the
8714 "__builtin_" prefix are enabled, and return true for C and
8715 Objective C in the default hook. Include langhooks.h in
8717 * targhooks.h (default_floatn_builtin_p): Likewise.
8718 * targhooks.c (default_floatn_builtin_p): Likewise.
8719 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8720 floatn_builtin_p target hook.
8721 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8723 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8724 Eric Botcazou <ebotcazou@adacore.com>
8726 PR rtl-optimization/81803
8727 * lra-constraints.c (curr_insn_transform): Also reload the whole
8728 register for a strict subreg no wider than a word if this is for
8729 a WORD_REGISTER_OPERATIONS target.
8731 2017-10-31 Jason Merrill <jason@redhat.com>
8733 * gdbinit.in: Skip over inlines from timevar.h.
8735 2017-10-31 Martin Liska <mliska@suse.cz>
8737 * doc/gcov.texi: Document new option.
8738 * gcov.c (print_usage): Likewise print it.
8739 (process_args): Support the argument.
8740 (format_count): New function.
8741 (format_gcov): Use the function.
8743 2017-10-31 Martin Liska <mliska@suse.cz>
8745 * gcov.c (struct name_map): do not use typedef.
8746 Define operator== and operator<.
8747 (name_search): Remove.
8748 (name_sort): Remove.
8749 (main): Do not allocate names.
8750 (process_file): Add vertical space.
8751 (generate_results): Use std::find.
8752 (release_structures): Do not release memory.
8753 (find_source): Use std::find.
8755 2017-10-31 Martin Liska <mliska@suse.cz>
8757 * gcov.c (struct line_info): Remove it's typedef.
8758 (line_info::line_info): Add proper ctor.
8759 (line_info::has_block): Do not use a typedef.
8760 (struct source_info): Do not use typedef.
8761 (circuit): Likewise.
8762 (get_cycles_count): Likewise.
8763 (output_intermediate_file): Iterate via vector iterator.
8764 (add_line_counts): Use std::vector methods.
8765 (accumulate_line_counts): Likewise.
8766 (output_lines): Likewise.
8768 2017-10-31 Martin Liska <mliska@suse.cz>
8770 * gcov.c (struct source_info): Remove typedef.
8771 (source_info::source_info): Add proper ctor.
8772 (accumulate_line_counts): Use struct, not it's typedef.
8773 (output_gcov_file): Likewise.
8774 (output_lines): Likewise.
8775 (main): Do not allocate an array.
8776 (output_intermediate_file): Use size of vector container.
8777 (process_file): Resize the vector.
8778 (generate_results): Do not preallocate, use newly added vector
8780 (release_structures): Do not release sources.
8781 (find_source): Use vector methods.
8782 (add_line_counts): Do not use typedef.
8784 2017-10-31 Martin Liska <mliska@suse.cz>
8786 * doc/gcov.texi: Document that.
8787 * gcov.c (add_line_counts): Mark lines with a non-executed
8789 (output_line_beginning): Handle such lines.
8790 (output_lines): Pass new argument.
8791 (output_intermediate_file): Print it in intermediate format.
8793 2017-10-31 Martin Liska <mliska@suse.cz>
8795 * color-macros.h: New file.
8796 * diagnostic-color.c: Factor out color related to macros to
8798 * doc/gcov.texi: Document -k option.
8799 * gcov.c (INCLUDE_STRING): Include string.h.
8800 (print_usage): Add -k option.
8801 (process_args): Parse it.
8802 (pad_count_string): New function.
8803 (output_line_beginning): Likewise.
8804 (DEFAULT_LINE_START): New macro.
8805 (output_lines): Support color output.
8807 2017-10-31 Martin Liska <mliska@suse.cz>
8809 PR gcov-profile/82633
8810 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8811 their interaction with GCOV infrastructure.
8812 * configure.ac: Add -fkeep-{inline,static}-functions to
8814 * configure: Regenerate.
8816 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8819 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8821 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8824 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8825 into a register if it does not fit into an immediate offset field.
8827 2017-10-31 Olivier Hainque <hainque@adacore.com>
8829 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8831 2017-10-31 Julia Koval <julia.koval@intel.com>
8833 * config.gcc: Add gfniintrin.h.
8834 * config/i386/gfniintrin.h: New.
8835 * config/i386/i386-builtin-types.def
8836 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8837 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8838 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8839 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8840 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8841 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8842 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8843 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8844 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8845 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8846 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8847 * config/i386/immintrin.h: Include gfniintrin.h.
8848 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8850 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
8852 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
8854 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8856 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
8857 (arm_ashldi3_1bit): Remove pattern.
8858 (ashrdi3): Remove shift by 1 expansion.
8859 (arm_ashrdi3_1bit): Remove pattern.
8860 (lshrdi3): Remove shift by 1 expansion.
8861 (arm_lshrdi3_1bit): Remove pattern.
8862 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
8863 cost of ashldi3 by 1.
8864 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
8865 (<shift>di3_neon): Likewise.
8867 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8869 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
8870 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
8871 and (*aarch64_simd_mov<VQ:mode>).
8872 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
8873 pattern alternative.
8874 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
8875 attributes to match pattern alternative.
8877 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
8879 * config.gcc (powerpc*-*-*): Add emmintrin.h.
8880 * config/rs6000/emmintrin.h: New file.
8881 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
8883 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8885 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
8886 * (movdi_vfp_cortexa8): Remove pattern.
8888 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8890 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
8892 (Specific, alpha*-dec-osf5.1): Remove.
8893 (Specific, mips-sgi-irix5): Remove.
8894 (Specific, mips-sgi-irix6): Remove.
8896 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8899 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
8900 arguments to clear_bit_region_be.
8902 2017-10-30 Jim Wilson <wilson@tuliptree.org>
8904 * gimplify.c: Include memmodel.h.
8906 2017-10-30 Martin Jambor <mjambor@suse.cz>
8908 * omp-grid.c (grid_attempt_target_gridification): Also insert a
8909 condition whether loop should be executed at all.
8911 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
8913 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
8914 gimple folding of vec_madd() intrinsics.
8915 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
8916 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
8917 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
8919 2017-10-30 Richard Biener <rguenther@suse.de>
8921 PR tree-optimization/82762
8923 2017-10-23 Richard Biener <rguenther@suse.de>
8925 PR tree-optimization/82129
8927 2017-08-01 Richard Biener <rguenther@suse.de>
8929 PR tree-optimization/81181
8930 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8931 (compute_antic): ... end of iteration here.
8933 2017-10-30 Joseph Myers <joseph@codesourcery.com>
8935 * doc/invoke.texi (C Dialect Options): Document -std=c17,
8936 -std=iso9899:2017 and -std=gnu17.
8937 * doc/standards.texi (C Language): Document C17 support.
8938 * doc/cpp.texi (Overview): Mention -std=c17.
8939 (Standard Predefined Macros): Document C11 and C17 values of
8940 __STDC_VERSION__. Do not refer to C99 support as incomplete.
8941 * doc/extend.texi (Inline): Do not list individual options for
8942 standards newer than C99.
8943 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
8945 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
8948 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
8950 * asan.c (asan_finish_file): Align asan globals array by shadow
8953 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8956 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
8957 (struct store_immediate_info): Add bitregion_start and bitregion_end
8959 (store_immediate_info::store_immediate_info): Add brs and bre
8960 arguments and initialize bitregion_{start,end} from those.
8961 (struct merged_store_group): Add bitregion_start, bitregion_end,
8962 align_base and mask fields. Drop unnecessary struct keyword from
8963 struct store_immediate_info. Add do_merge method.
8964 (clear_bit_region_be): Use memset instead of loop storing zeros.
8965 (merged_store_group::do_merge): New method.
8966 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
8967 stores as long as the surrounding bitregions have no gaps.
8968 (merged_store_group::merge_overlapping): Use do_merge.
8969 (merged_store_group::apply_stores): Test that bitregion_{start,end}
8970 is byte aligned, rather than requiring that start and width are
8971 byte aligned. Drop unnecessary struct keyword from
8972 struct store_immediate_info. Allocate and populate also mask array.
8973 Make start of the arrays relative to bitregion_start rather than
8974 start and size them according to bitregion_{end,start} difference.
8975 (struct imm_store_chain_info): Drop unnecessary struct keyword from
8976 struct store_immediate_info.
8977 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
8978 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
8979 struct keyword from struct store_immediate_info.
8980 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
8981 between stores as long as the surrounding bitregions have no gaps.
8983 (struct split_store): Add orig non-static data member.
8984 (split_store::split_store): Initialize orig to false.
8985 (find_constituent_stmts): Return store_immediate_info *, non-NULL
8986 if there is exactly a single original stmt. Change stmts argument
8987 to pointer from reference, if NULL, don't push anything to it. Add
8988 first argument, use it to optimize skipping over orig stmts that
8989 are known to be before bitpos already. Simplify.
8990 (split_group): Return unsigned int count how many stores are or
8991 would be needed rather than a bool. Add allow_unaligned argument.
8992 Change split_stores argument from reference to pointer, if NULL,
8993 only do a dry run computing how many stores would be produced.
8994 Rewritten algorithm to use both alignment and misalign if
8995 !allow_unaligned and handle bitfield stores with gaps.
8996 (imm_store_chain_info::output_merged_store): Set start_byte_pos
8997 from bitregion_start instead of start. Compute allow_unaligned
8998 here, if true, do 2 split_group dry runs to compute which one
8999 produces fewer stores and prefer aligned if equal. Punt if
9000 new count is bigger or equal than original before emitting any
9001 statements, rather than during that. Remove no longer needed
9002 new_ssa_names tracking. Replace num_stmts with
9003 split_stores.length (). Use 32-bit stack allocated entries
9004 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9005 if possible. Handle bitfields with gaps.
9006 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9007 Compute bitregion_{start,end} for the stores and construct
9008 store_immediate_info with that. Formatting fixes.
9010 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9013 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9014 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9016 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9018 * gimplify.c: Include tm_p.h.
9020 * common.opt (gcoff): Re-add as ignored option.
9021 (gcoff1, gcoff2, gcoff3): Likewise.
9023 * Makefile.in (OBJS): Delete sdbout.o.
9024 (GTFILES): Delete $(srcdir)/sdbout.c.
9025 * debug.h: Delete sdb_debug_hooks.
9026 * final.c: Delete sdbout.h include.
9027 (final_scan_insn): Delete SDB_DEBUG check.
9028 (rest_of_clean_state): Likewise.
9029 * output.h: Delete sdb_begin_function_line.
9032 * toplev.c: Delete sdbout.h include.
9033 (process_options): Delete SDB_DEBUG check.
9034 * tree-core.h (tree_type_common): Delete pointer field of
9036 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9037 TYPE_SYMTAB_POINTER.
9038 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9039 (TYPE_SYMTAB_IS_POINTER): Delete.
9040 (TYPE_SYMTAB_IS_DIE): Renumber.
9041 * xcoffout.c: Refer to former sdbout.c file.
9042 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9044 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9045 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9046 (Debugging Options): Delete -gcoff.
9047 (-gstabs): Delete SDB reference.
9049 (-gcoff@var{level}): Delete.
9050 * doc/passes.texi (Debugging information output): Delete SDB and
9051 sdbout.c references.
9052 * doc/tm.texi: Regenerate.
9053 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9054 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9056 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9057 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9058 (SDB_DEBUGGING_INFO): Delete.
9059 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9060 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9061 * target.def (output_source_filename): Delete COFF reference.
9063 * common.opt (gcoff): Delete.
9064 (gxcoff+): Update Negative chain.
9065 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9067 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9068 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9069 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9071 (expand_function_start): Change sdb reference to past tense.
9072 (expand_function_end): Change sdb reference to past tense.
9073 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9074 * opts.c (debug_type_names): Delete coff entry.
9075 (common_handle_option): Delete OPT_gcoff case.
9076 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9078 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9079 * config/cris/cris.h: Delete SDB reference in comment.
9080 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9081 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9082 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9083 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9085 (ix86_expand_prologue): Likewise.
9086 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9087 * config/ia64/ia64.h: Likewise.
9088 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9089 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9091 * config/mmix/mmix.h: Likewise.
9092 * config/nds32/nds32.c: Likewise.
9093 * config/stormy/storym16.h: Likewise.
9094 * config/visium/visium.h: Likewise.
9095 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9097 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9099 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9100 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9101 FRAME_POINTER_REGNUM point at high end of local var area.
9103 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9105 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9106 Move comment around. Do not reset best_edge for a copiable
9107 destination if the copy would cause a partition change.
9108 (better_edge_p): Remove redundant check.
9110 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9112 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9114 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9116 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9117 for math functions that have _Float<N> and _Float<N>X variants.
9118 (mathfn_built_in_2): Add support for math functions that have
9119 _Float<N> and _Float<N>X variants.
9120 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9121 (expand_builtin_mathfn_ternary): Add support for fma with
9122 _Float<N> and _Float<N>X variants.
9123 (expand_builtin): Likewise.
9124 (fold_builtin_3): Likewise.
9125 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9126 create math function _Float<N> and _Float<N>X variants as external
9128 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9129 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9130 the __builtin_ prefix and if not strict ansi, without the prefix.
9131 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9132 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9133 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9134 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9135 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9136 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9137 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9138 function signatures for fma _Float<N> and _Float<N>X variants.
9139 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9140 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9141 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9142 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9143 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9144 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9145 * gencfn-macros.c (print_case_cfn): Add support for math functions
9146 that have _Float<N> and _Float<N>X variants.
9147 (print_define_operator_list): Likewise.
9148 (fltfn_suffixes): Likewise.
9150 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9151 for math functions that have _Float<N> and _Float<N>X variants.
9152 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9153 and _Float<N>X variants.
9154 (COPYSIGN): Likewise.
9157 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9158 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9160 (integer_valued_read_call_p): Likewise.
9161 * fold-const-call.c (fold_const_call_ss): Likewise.
9162 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9163 _Float<N> and _Float<N>X variants.
9164 (fold_const_call_ssss): Add support for fma _Float<N> and
9165 _Float<N>X variants.
9166 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9167 support for copysign and fma _Float<N> and _Float<N>X variants.
9168 (backprop::process_builtin_call_use): Likewise.
9169 * tree-call-cdce.c (can_test_argument_range); Add support for
9170 sqrt _Float<N> and _Float<N>X variants.
9171 (edom_only_function): Likewise.
9172 (get_no_error_domain): Likewise.
9173 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9174 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9175 copysign _Float<N> and _Float<N>X variants.
9176 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9177 handled by machine independent code.
9178 (FMAF128): Likewise.
9179 * doc/cpp.texi (Common Predefined Macros): Document defining
9180 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9181 fma _Float<N> and _Float<N>X variants.
9183 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9186 * config/i386/i386-modes.def (CCFPU): Remove definition.
9187 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9188 (ix86_cc_modes_compatible): Ditto.
9189 (ix86_expand_carry_flag_compare): Ditto.
9190 (ix86_expand_int_movcc): Ditto.
9191 (ix86_expand_int_addcc): Ditto.
9192 (ix86_reverse_condition): Ditto.
9193 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9194 Return true/false for unordered/ordered fp comparisons.
9195 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9196 (ix86_prepare_fp_compare_args): Update for rename.
9197 (ix86_expand_fp_compare): Update for rename. Generate unordered
9198 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9199 (ix86_expand_sse_compare_and_jump): Ditto.
9200 * config/i386/predicates.md (fcmov_comparison_operator):
9201 Remove CCFPU mode handling.
9202 (ix86_comparison_operator): Ditto.
9203 (ix86_carry_flag_operator): Ditto.
9204 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9205 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9206 (*cmpu<mode>_cc_i387): Ditto.
9207 (FPCMP): Remove mode iterator.
9208 (unord): Remove mode attribute.
9209 (unord_subst): New define_subst transformation
9210 (unord): New define_subst attribute.
9212 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9213 (*cmpi<unord>xf_i387): Ditto.
9214 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9215 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9216 using unord_subst transformation.
9217 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9218 (round_saeonly): Also handle CCFP mode.
9219 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9220 Remove UNSPEC_SAHF unspec handling.
9222 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9224 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9226 2017-10-27 Jeff Law <law@redhat.com>
9228 * gimple-ssa-sprintf.c: Include domwalk.h.
9229 (class sprintf_dom_walker): New class, derived from dom_walker.
9230 (sprintf_dom_walker::before_dom_children): New function.
9231 (struct call_info): Moved into sprintf_dom_walker class
9232 (compute_formath_length, handle_gimple_call): Likewise.
9233 (sprintf_length::execute): Call the dominator walker rather
9234 than walking the statements.
9236 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9237 gimple statement locations.
9238 (check_array_bounds): Corresponding changes. Get the statement's
9239 location directly from wi->stmt.
9241 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9244 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9246 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9248 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9249 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9251 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9254 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9255 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9256 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9257 maybe_get_pool_constant.
9258 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9261 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9263 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9264 2.26 caveat. Update gas and gld versions.
9265 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9268 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9270 * cgraph.h (set_malloc_flag): Declare.
9271 * cgraph.c (set_malloc_flag_1): New function.
9272 (set_malloc_flag): Likewise.
9273 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9274 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9276 (read_ipa_call_summary): Add support for reading is_return_callee.
9277 (write_ipa_call_summary): Stream is_return_callee.
9278 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9279 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9280 ipa-prop.h, ipa-fnsummary.h.
9281 (pure_const_names): Change to static.
9282 (malloc_state_e): Define.
9283 (malloc_state_names): Define.
9284 (funct_state_d): Add field malloc_state.
9285 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9286 (check_retval_uses): New function.
9287 (malloc_candidate_p): Likewise.
9288 (analyze_function): Add support for malloc attribute.
9289 (pure_const_write_summary): Stream malloc_state.
9290 (pure_const_read_summary): Add support for reading malloc_state.
9291 (dump_malloc_lattice): New function.
9292 (propagate_malloc): New function.
9293 (warn_function_malloc): New function.
9294 (ipa_pure_const::execute): Call propagate_malloc and
9295 ipa_free_fn_summary.
9296 (pass_local_pure_const::execute): Add support for malloc attribute.
9297 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9298 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9300 2017-10-27 Martin Liska <mliska@suse.cz>
9302 PR gcov-profile/82457
9303 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9304 for fork-like functions to be properly instrumented.
9306 2017-10-27 Richard Biener <rguenther@suse.de>
9309 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9310 info when we redirected EH.
9312 2017-10-26 Michael Collison <michael.collison@arm.com>
9314 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9316 (<optab>_trunchf<GPI:mode>2: New pattern.
9317 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9318 * config/aarch64/iterators.md (wv): New mode attribute.
9319 (vf, VF): New mode attributes.
9320 (vgp, VGP): New mode attributes.
9321 (s): Update attribute with SImode and DImode prefixes.
9323 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9325 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9326 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9327 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9328 (nios2_option_overide): Initialize it. Don't allow R0-relative
9329 addressing with PIC.
9330 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9331 (nios2_symbolic_constant_p): Likewise.
9332 (nios2_legitimate_address_p): Likewise.
9333 (nios2_r0rel_section_name_p): New.
9334 (nios2_symbol_ref_in_r0rel_data_p): New.
9335 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9336 (r0rel_constant_p): New.
9337 (nios2_print_operand_address): Handle r0rel_constant_p.
9338 (nios2_cdx_narrow_form_p): Likewise.
9339 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9340 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9341 (Nios II Options): Document -mr0rel-sec.
9343 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9345 * config/nios2/nios2.c: Include xregex.h.
9346 (nios2_gprel_sec_regex): New.
9347 (nios2_option_overide): Initialize it. Don't allow GP-relative
9348 addressing with PIC.
9349 (nios2_small_section_name_p): Check for regex match.
9350 * config/nios2/nios2.opt (mgprel-sec=): New option.
9351 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9352 (Nios II Options): Document -mgprel-sec.
9354 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9356 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9358 2017-10-26 Tom de Vries <tom@codesourcery.com>
9360 PR tree-optimization/82707
9361 * gimple.c (gimple_copy): Fix unsharing of
9362 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9364 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9366 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9367 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9368 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9369 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9370 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9371 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9372 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9373 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9374 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9375 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9376 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9377 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9378 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9379 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9380 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9381 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9382 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9384 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9386 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9388 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9389 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9390 warning to rs6000.c. Remove the Undocumented flag, since it has
9392 (-mabi=ibmlongdouble): Likewise.
9393 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9394 already set, set the default format for long double.
9395 (rs6000_debug_reg_global): Print whether long double is IBM or
9397 (rs6000_option_override_internal): Rework setting long double
9398 format. Only warn if the user is changing the long double default
9399 and they did not use -Wno-psabi.
9400 * doc/invoke.texi (PowerPC options): Update the documentation for
9401 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9403 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9404 Alan Hayward <alan.hayward@arm.com>
9405 David Sherwood <david.sherwood@arm.com>
9407 * rtl.h (wider_subreg_mode): New function.
9408 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9409 rather than an unsigned int *.
9410 * ira-color.c (regno_max_ref_width): Replace with...
9411 (regno_max_ref_mode): ...this new variable.
9412 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9413 Use wider_subreg_mode.
9414 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9415 rather than an unsigned int *.
9416 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9417 (process_alt_operands): Likewise.
9418 (invariant_p): Likewise.
9419 * lra-spills.c (assign_mem_slot): Likewise.
9420 (add_pseudo_to_slot): Likewise.
9421 * lra.c (collect_non_operand_hard_regs): Likewise.
9422 (add_regs_to_insn_regno_info): Likewise.
9423 * reload1.c (regno_max_ref_width): Replace with...
9424 (regno_max_ref_mode): ...this new variable.
9425 (reload): Update accordingly. Update call to
9426 ira_sort_regnos_for_alter_reg.
9427 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9428 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9429 (scan_paradoxical_subregs): Likewise.
9431 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9433 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9434 (aarch64_frame): Add emit_frame_chain boolean.
9435 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9436 Move eh_return case to aarch64_layout_frame.
9437 (aarch64_layout_frame): Initialize emit_frame_chain.
9438 (aarch64_expand_prologue): Use emit_frame_chain.
9440 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9442 * config/aarch64/aarch64.c (aarch64_layout_frame):
9443 Ensure LR is always stored at the bottom of the callee-saves.
9444 Remove rarely used frame layout which saves callee-saves at top of
9445 frame, so the store of LR can be used as a valid probe in all cases.
9447 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9449 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9450 Improve unaligned TImode/TFmode base/offset split.
9452 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9453 Alan Hayward <alan.hayward@arm.com>
9454 David Sherwood <david.sherwood@arm.com>
9456 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9457 * combine.c (find_single_use_1): Likewise.
9458 (expand_field_assignment): Likewise.
9459 (move_deaths): Likewise.
9460 * lra-constraints.c (simplify_operand_subreg): Likewise.
9461 (curr_insn_transform): Likewise.
9462 * lra.c (collect_non_operand_hard_regs): Likewise.
9463 (add_regs_to_insn_regno_info): Likewise.
9464 * rtlanal.c (reg_referenced_p): Likewise.
9465 (covers_regno_no_parallel_p): Likewise.
9467 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9469 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9470 Don't print any bits outside the precision of the value.
9471 * wide-int.cc (test_printing): Add some new tests.
9473 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9475 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9476 supports -xbrace_comment option.
9477 * configure: Regenerate.
9478 * config.in: Regenerate.
9479 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9480 (ASM_CPU_SPEC): Use it.
9482 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9484 * target.def (static_rtx_alignment): New hook.
9485 * targhooks.h (default_static_rtx_alignment): Declare.
9486 * targhooks.c (default_static_rtx_alignment): New function.
9487 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9488 * doc/tm.texi: Regenerate.
9489 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9490 instead of targetm.constant_alignment. Remove call to
9492 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9493 (cris_preferred_mininum_alignment): New function, split out from...
9494 (cris_constant_alignment): ...here.
9495 (cris_static_rtx_alignment): New function.
9496 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9498 (ix86_constant_alignment): ...here.
9499 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9500 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9501 (mmix_static_rtx_alignment): New function.
9502 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9503 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9505 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9508 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9509 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9511 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9513 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9514 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9515 mark as a sign-extending load.
9516 (local_pic_load_u): Define.
9518 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9521 * fold-const.c (operand_equal_for_comparison_p): Also return true
9522 if ARG0 is a simple variant of ARG1 with narrower precision.
9523 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9525 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9527 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9529 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9530 scatter_static, scatter_per_elt.
9531 * x86-tune-costs.h: Add new cost entries.
9533 2017-10-25 Richard Biener <rguenther@suse.de>
9535 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9536 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9537 class pass_fre): Move to ...
9538 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9539 class pass_fre): ... here and adjust for statistics.
9541 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9544 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9545 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9546 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9547 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9550 2017-10-25 Richard Biener <rguenther@suse.de>
9552 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9553 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9554 eliminate_push_avail, eliminate_insert): Move inside...
9555 (class eliminate_dom_walker): ... this class in preparation
9557 (fini_eliminate): Remove by merging with ...
9558 (eliminate): ... this function. Adjust for class changes.
9559 (pass_pre::execute): Remove fini_eliminate call.
9560 (pass_fre::execute): Likewise.
9562 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9565 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9566 (VPERMI2, VPERMI2I): New mode iterators.
9567 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9568 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9570 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9571 mode iterator. Remove 3 old define_insn patterns.
9572 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9573 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9574 VPERMI2 mode iterator, remove the other two expanders.
9575 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9576 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9577 instructions, remove the other two patterns.
9578 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9579 mode iterator, remove the other two patterns.
9580 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9581 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9582 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9583 and adjust argument order accordingly.
9584 (ix86_expand_vec_perm): Adjust caller.
9585 (expand_vec_perm_1): Likewise.
9586 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9587 (expand_vec_perm_vpermt2_vpshub2): ... this.
9588 (ix86_expand_vec_perm_const_1): Adjust caller.
9589 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9592 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9593 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9594 (vec_shl_<mode>): Remove unused expander.
9595 (avx512bw_<shift_insn><mode>3): New define_insn.
9596 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9597 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9599 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9602 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9605 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9607 PR rtl-optimization/82396
9608 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9609 (autopref_multipass_init): Simplify initialization.
9610 (autopref_rank_data): Simplify sort order.
9611 * gcc/sched-int.h (autopref_multipass_data_): Remove
9612 multi_mem_insn_p, min_offset and max_offset.
9614 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9617 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9618 Check special value of flag_omit_frame_pointer.
9619 (aarch64_can_eliminate): Likewise.
9620 (aarch64_override_options_after_change_1): Simplify handling of
9621 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9623 2017-10-24 Richard Biener <rguenther@suse.de>
9625 PR tree-optimization/82697
9626 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9627 zero for conditional load and unconditional store.
9629 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9631 * doc/install.texi: Document bootstrap-cet.
9633 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9636 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9637 ENDBR instruction at function entrance if function is only
9640 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9643 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9644 patterns to better describe from which operation the CF is computed.
9645 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9646 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9647 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9648 is 0, use _0 suffixed expanders instead of emitting a comparison
9651 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9653 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9654 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9656 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9659 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9660 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9661 * loop-iv.c (iv_get_reaching_def): Likewise.
9662 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9663 variable is promoted and the partition contains undefined values.
9665 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9667 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9669 (nios2_address_cost): Define.
9670 (nios2_legitimize_address): Recognize (exp + constant) directly.
9671 (TARGET_ADDRESS_COST): Define.
9673 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9675 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9676 (nios2_symbolic_memory_operand_p): Declare.
9677 (nios2_split_large_constant): Declare.
9678 (nios2_split_symbolic_memory_operand): Declare.
9679 * config/nios2/nios2.c: Adjust includes.
9680 (nios2_symbolic_constant_allowed): New.
9681 (nios2_symbolic_constant_p): New.
9682 (nios2_plus_symbolic_constant_p): New.
9683 (nios2_valid_addr_expr_p): Recognize addresses involving
9685 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9686 (nios2_symbolic_memory_operand_p): New.
9687 (nios2_large_constant_p): New.
9688 (nios2_split_large_constant): New.
9689 (nios2_split_plus_large_constant): New.
9690 (nios2_split_symbolic_memory_operand): New.
9691 (nios2_legitimize_address): Code refactoring. Handle addresses
9692 involving symbolic constants.
9693 (nios2_emit_move_sequence): Likewise.
9694 (nios2_print_operand): Improve error output.
9695 (nios2_print_operand_address): Handle LO_SUM.
9696 (nios2_cdx_narrow_form_p): Likewise.
9697 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9698 operands involving symbolic constants.
9699 (movhi_internal, movsi_internal): Likewise.
9700 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9701 (extendhisi2, extendqi<mode>2): Likewise.
9703 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9705 * tree-pass.h (PROP_rtl_split_insns): Define.
9706 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9708 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9710 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9712 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9715 * target.def (const_not_ok_for_debug_p): Default to
9716 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9717 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9718 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9719 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9720 which targetm.const_not_ok_for_debug_p returned true.
9721 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9723 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9725 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9726 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9727 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9729 (ix86_const_not_ok_for_debug_p): New function.
9730 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9731 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9733 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9736 * system.h: Conditionally include "unique-ptr.h" if
9737 INCLUDE_UNIQUE_PTR is defined.
9738 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9739 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9741 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9743 * config/rl78/rl78.md: New define_expand "subdi3".
9745 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9748 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9749 DF_REF_INSN if DF_REF_INSN_INFO is false.
9751 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9753 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9754 xmm_move instead of sse_move.
9755 (sse_store_index): New function.
9756 (ix86_register_move_cost): Be more sensible about mismatch stall;
9757 model AVX moves correctly; make difference between sse->integer and
9759 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9760 moves; make difference between SSE and AVX.
9761 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9762 and zmm_move. Increase size of sse load and store tables;
9763 add unaligned load and store tables; add ssemmx_to_integer.
9764 * x86-tune-costs.h: Update all entries according to real
9765 move latencies from Agner Fog's manual and chip documentation.
9767 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9770 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9771 * config/i386/constraints.md (Wf): New constraint.
9772 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9773 (cmp<dwi>_doubleword): Removed.
9774 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9775 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9776 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9777 expand with cmp<dwi>_doubleword. For LTU and GEU use
9778 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9780 * common.opt (gcolumn-info): Enable by default.
9781 * doc/invoke.texi (gcolumn-info): Document new default.
9783 2017-10-23 Richard Biener <rguenther@suse.de>
9785 PR tree-optimization/82672
9786 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9787 Fold the stmt if we propagated into it.
9789 2017-10-23 Richard Biener <rguenther@suse.de>
9791 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9792 (bitmap_remove_expr_from_set): ... this. All callers call this
9793 for non-constant values.
9794 (bitmap_set_subtract): Rename to...
9795 (bitmap_set_subtract_expressions): ... this. Adjust and
9797 (bitmap_set_contains_value): Remove superfluous check.
9798 (bitmap_set_replace_value): Inline into single caller ...
9799 (bitmap_value_replace_in_set): ... here and simplify.
9800 (dependent_clean): Merge into ...
9801 (clean): ... this using an overload. Adjust.
9802 (prune_clobbered_mems): Adjust.
9803 (compute_antic_aux): Likewise.
9804 (compute_partial_antic_aux): Likewise.
9806 2017-10-23 Richard Biener <rguenther@suse.de>
9808 PR tree-optimization/82129
9810 2017-08-01 Richard Biener <rguenther@suse.de>
9812 PR tree-optimization/81181
9813 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9814 (compute_antic): ... end of iteration here.
9816 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9818 * target.def (starting_frame_offset): New hook.
9819 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9820 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9821 * doc/tm.texi.in: Regenerate.
9822 * hooks.h (hook_hwi_void_0): Declare.
9823 * hooks.c (hook_hwi_void_0): New function.
9824 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9825 STARTING_FRAME_OFFSET.
9826 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9827 * reload1.c (reload): Likewise.
9828 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9829 instead of STARTING_FRAME_OFFSET.
9830 * function.c (try_fit_stack_local): Likewise.
9831 (assign_stack_local_1): Likewise
9832 (instantiate_virtual_regs): Likewise.
9833 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9834 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9835 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9836 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9837 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9838 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9839 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9840 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9841 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9842 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9843 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9844 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9845 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9846 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9847 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9848 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9849 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9850 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
9851 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
9852 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
9853 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
9854 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
9855 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
9856 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
9857 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
9858 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
9859 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
9860 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
9861 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
9862 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
9863 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
9864 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
9865 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
9866 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
9867 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
9868 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
9869 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
9870 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
9871 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
9872 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
9873 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
9874 * config/avr/avr.c (avr_starting_frame_offset): Make static and
9875 return a HOST_WIDE_INT.
9876 (avr_builtin_setjmp_frame_value): Use it instead of
9877 STARTING_FRAME_OFFSET.
9878 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9879 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
9880 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
9882 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9883 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
9884 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
9885 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9886 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
9887 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
9888 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9889 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
9890 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
9891 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9892 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
9893 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
9895 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9896 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
9897 * config/mips/mips.c (mips_compute_frame_info): Refer to
9898 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9899 (mips_starting_frame_offset): New function.
9900 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9901 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
9902 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
9903 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
9904 and return a HOST_WIDE_INT.
9905 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9906 (mmix_initial_elimination_offset): Refer to
9907 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9908 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
9909 * config/pa/pa.c (pa_starting_frame_offset): New function.
9910 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
9911 (pa_expand_prologue): Likewise.
9912 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9913 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
9914 !FRAME_GROWS_DOWNWARD handling to...
9915 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9916 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
9917 !FRAME_GROWS_DOWNWARD handling to...
9918 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9919 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
9920 !FRAME_GROWS_DOWNWARD handling to...
9921 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9922 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
9924 (rs6000_starting_frame_offset): New function.
9925 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
9926 !FRAME_GROWS_DOWNWARD handling to...
9927 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9928 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
9929 !FRAME_GROWS_DOWNWARD handling to...
9930 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9931 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
9932 !FRAME_GROWS_DOWNWARD handling to...
9933 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9934 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
9935 (rs6000_starting_frame_offset): New function.
9936 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
9937 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
9938 * config/vax/vax.c (vax_starting_frame_offset): New function.
9939 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
9940 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9941 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
9942 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
9943 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9944 * system.h (STARTING_FRAME_OFFSET): Poison.
9946 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9948 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9949 SCALAR_TYPE_MODE instead of TYPE_MODE.
9951 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9952 Alan Hayward <alan.hayward@arm.com>
9953 David Sherwood <david.sherwood@arm.com>
9955 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
9957 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9958 Alan Hayward <alan.hayward@arm.com>
9959 David Sherwood <david.sherwood@arm.com>
9961 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
9963 2017-10-23 Richard Biener <rguenther@suse.de>
9965 PR tree-optimization/82129
9966 * tree-ssa-pre.c (bitmap_set_and): Remove.
9967 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
9968 canonicalizing expressions in the set to those with lowest
9969 ID rather than taking that from the first edge.
9971 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9973 * combine.c (rtx_equal_for_field_assignment_p): Use
9974 byte_lowpart_offset.
9976 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9977 Alan Hayward <alan.hayward@arm.com>
9978 David Sherwood <david.sherwood@arm.com>
9980 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
9981 to a SUBREG_PROMOTED_VAR.
9983 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9984 Alan Hayward <alan.hayward@arm.com>
9985 David Sherwood <david.sherwood@arm.com>
9987 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
9988 (expand_debug_source_expr): Likewise.
9989 * combine.c (combine_simplify_rtx): Likewise.
9990 * cse.c (fold_rtx): Likewise.
9991 * optabs.c (expand_float): Likewise.
9992 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9993 (simplify_binary_operation_1): Likewise.
9995 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9996 Alan Hayward <alan.hayward@arm.com>
9997 David Sherwood <david.sherwood@arm.com>
9999 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10000 (record_promoted_value): Likewise.
10001 * expr.c (expand_expr_real_2): Likewise.
10002 * ree.c (update_reg_equal_equiv_notes): Likewise.
10003 (combine_set_extension): Likewise.
10004 * rtlanal.c (low_bitmask_len): Likewise.
10005 * simplify-rtx.c (neg_const_int): Likewise.
10006 (simplify_binary_operation_1): Likewise.
10008 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10009 Alan Hayward <alan.hayward@arm.com>
10010 David Sherwood <david.sherwood@arm.com>
10012 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10013 * regcprop.c (maybe_mode_change): Likewise.
10014 * reload1.c (alter_reg): Likewise.
10016 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10018 * inchash.h (inchash::hash::add_wide_int): New function.
10019 * lto-streamer-out.c (hash_tree): Use it.
10021 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10023 * inchash.h (inchash::hash::add_wide_int): Rename to...
10024 (inchash::hash::add_hwi): ...this.
10025 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10026 (polymorphic_call_target_hasher::hash): Likewise.
10027 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10028 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10029 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10030 * lto-streamer-out.c (hash_tree): Likewise.
10031 * optc-save-gen.awk: Likewise.
10032 * tree.c (add_expr): Likewise.
10034 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10037 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10038 for ordered inequality comparisons even with TARGET_IEEE_FP.
10040 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10043 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10044 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10045 Expand with cmp<dwi>_doubleword.
10047 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10049 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10050 List CET intrinsics.
10051 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10052 specific to -fcf-protection option.
10054 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10056 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10057 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10058 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10059 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10060 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10061 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10062 (extra_objs): Add cet.o for Linux/x86 targets.
10063 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10064 * config/i386/cet.c: New file.
10065 * config/i386/cetintrin.h: Likewise.
10066 * config/i386/t-cet: Likewise.
10067 * config/i386/cpuid.h (bit_SHSTK): New.
10068 (bit_IBT): Likewise.
10069 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10070 pass IBT and SHSTK bits.
10071 * config/i386/i386-builtin-types.def
10072 (VOID_FTYPE_UNSIGNED_PVOID): New.
10073 (VOID_FTYPE_UINT64_PVOID): Likewise.
10074 * config/i386/i386-builtin.def: Add CET intrinsics.
10075 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10076 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10077 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10078 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10080 * config/i386/i386.c (rest_of_insert_endbranch): New.
10081 (pass_data_insert_endbranch): Likewise.
10082 (pass_insert_endbranch): Likewise.
10083 (make_pass_insert_endbranch): Likewise.
10084 (ix86_notrack_prefixed_insn_p): Likewise.
10085 (ix86_target_string): Add -mibt, -mshstk flags.
10086 (ix86_option_override_internal): Add flag_cf_protection
10088 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10089 (ix86_print_operand): Add 'notrack' prefix output.
10090 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10091 (ix86_expand_builtin): Expand CET intrinsics.
10092 (x86_output_mi_thunk): Add 'endbranch' instruction.
10093 * config/i386/i386.h (TARGET_IBT): New.
10094 (TARGET_IBT_P): Likewise.
10095 (TARGET_SHSTK): Likewise.
10096 (TARGET_SHSTK_P): Likewise.
10097 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10098 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10099 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10100 (builtin_setjmp_setup): New pattern.
10101 (builtin_longjmp): Likewise.
10102 (rdssp<mode>): Likewise.
10103 (incssp<mode>): Likewise.
10104 (saveprevssp): Likewise.
10105 (rstorssp): Likewise.
10106 (wrss<mode>): Likewise.
10107 (wruss<mode>): Likewise.
10108 (setssbsy): Likewise.
10109 (clrssbsy): Likewise.
10110 (nop_endbr): Likewise.
10111 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10113 * config/i386/immintrin.h: Include <cetintrin.h>.
10114 * config/i386/linux-common.h
10115 (file_end_indicate_exec_stack_and_cet): New prototype.
10116 (TARGET_ASM_FILE_END): New.
10118 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10120 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10121 latencies instead of having separate table; make difference between
10122 integer and float costs.
10123 * i386.h (processor_costs): Remove scalar_stmt_cost,
10124 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10125 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10127 * x86-tune-costs.h: Remove entries which has been removed in
10128 procesor_costs from all tables; make cond_taken_branch_cost
10129 and cond_not_taken_branch_cost COST_N_INSNS based.
10131 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10133 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10135 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10137 * config/i386/i386.md (isa): Remove fma_avx512f.
10138 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10139 <avx512>_fmadd_<mode>_mask3<round_name>,
10140 <avx512>_fmsub_<mode>_mask<round_name>,
10141 <avx512>_fmsub_<mode>_mask3<round_name>,
10142 <avx512>_fnmadd_<mode>_mask<round_name>,
10143 <avx512>_fnmadd_<mode>_mask3<round_name>,
10144 <avx512>_fnmsub_<mode>_mask<round_name>,
10145 <avx512>_fnmsub_<mode>_mask3<round_name>,
10146 <avx512>_fmaddsub_<mode>_mask<round_name>,
10147 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10148 <avx512>_fmsubadd_<mode>_mask<round_name>,
10149 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10150 (*vec_widen_umult_even_v16si<mask_name>,
10151 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10152 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10154 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10156 * extend.texi: Add 'nocf_check' documentation.
10157 * gimple.texi: Add second parameter to
10158 gimple_build_call_from_tree.
10159 * invoke.texi: Add -fcf-protection documentation.
10160 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10162 2017-10-20 Richard Biener <rguenther@suse.de>
10164 PR tree-optimization/82473
10165 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10166 the largest input type.
10168 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10170 * c-attribs.c (handle_nocf_check_attribute): New function.
10171 (c_common_attribute_table): Add 'nocf_check' handling.
10172 * gimple-parser.c: Add second argument NULL to
10173 gimple_build_call_from_tree.
10174 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10175 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10177 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10178 * common.opt: Add fcf-protection flag.
10179 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10180 * flag-types.h: Add enum cf_protection_level.
10181 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10182 Add 'nocf_check' attribute propagation to gimple call.
10183 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10184 (gimple_build_call_from_tree): Update prototype.
10185 (gimple_call_nocf_check_p): New function.
10186 (gimple_call_set_nocf_check): Likewise.
10187 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10188 * ipa-icf.c: Add nocf_check attribute in statement hash.
10189 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10190 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10191 * toplev.c (process_options): Add flag_cf_protection handling.
10193 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10195 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10197 2017-10-20 Richard Biener <rguenther@suse.de>
10199 PR tree-optimization/82603
10200 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10201 remove false predicated stores.
10203 2017-10-20 Richard Biener <rguenther@suse.de>
10205 * graphite-isl-ast-to-gimple.c
10206 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10207 Remove return value and simplify, dump copied stmt after lhs
10209 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10210 Reduce dump verbosity.
10211 (gsi_insert_earliest): Likewise.
10212 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10213 * graphite.c (print_global_statistics): Adjust dumping.
10214 (print_graphite_scop_statistics): Likewise.
10215 (print_graphite_statistics): Do not dump loops here.
10216 (graphite_transform_loops): But here.
10218 2017-10-20 Nicolas Roche <roche@adacore.com>
10220 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10221 * configure: Regenerate.
10223 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10226 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10227 functions when optimizing replace GIMPLE_RETURN stmts with
10228 calls to __builtin_unreachable ().
10231 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10232 for -fsanitize=thread link of executables.
10233 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10234 link of executables.
10237 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10238 New mode iterators.
10239 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10240 define_insns for logical vector shifts to use VI248_AVX512BW
10241 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10242 condition, useless isa and prefix attributes. Change the first
10243 2 of these define_insns to ...
10244 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10245 define_insn for avx512vl.
10246 (<shift_insn><mode>3): ... and this, new define_insn without
10247 masking for non-avx512vl.
10250 * config/i386/sse.md (*andnot<mode>3,
10251 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10252 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10253 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10254 not applied use empty suffix even for TARGET_AVX512VL.
10255 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10256 is applied, supply evex,evex or evex,evex,evex instead of just
10259 2017-10-20 Julia Koval <julia.koval@intel.com>
10261 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10262 (OPTION_MASK_ISA_GFNI_UNSET): New.
10263 (ix86_handle_option): Handle OPT_mgfni.
10264 * config/i386/cpuid.h (bit_GFNI): New.
10265 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10266 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10267 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10268 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10269 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10270 * config/i386/i386.opt: Add mgfni.
10272 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10274 * config/msp430/msp430.c (msp430_option_override): Disable
10275 -fdelete-null-pointer-checks.
10276 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10278 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10280 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10281 of x87 and SSE instructions.
10283 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10285 * asan.c (create_cond_insert_point): Do not update edge count.
10286 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10287 (afdo_propagate_circuit): Likewise.
10288 (afdo_calculate_branch_prob): Likewise.
10289 (afdo_annotate_cfg): Likewise.
10290 * basic-block.h (struct edge_def): Remove count.
10291 (edge_def::count): New accessor.
10292 * bb-reorder.c (rotate_loop): Update.
10293 (find_traces_1_round): Update.
10294 (connect_traces): Update.
10295 (sanitize_hot_paths): Update.
10296 * cfg.c (unchecked_make_edge): Update.
10297 (make_single_succ_edge): Update.
10298 (check_bb_profile): Update.
10299 (dump_edge_info): Update.
10300 (update_bb_profile_for_threading): Update.
10301 (scale_bbs_frequencies_int): Update.
10302 (scale_bbs_frequencies_gcov_type): Update.
10303 (scale_bbs_frequencies_profile_count): Update.
10304 (scale_bbs_frequencies): Update.
10305 * cfganal.c (connect_infinite_loops_to_exit): Update.
10306 * cfgbuild.c (compute_outgoing_frequencies): Update.
10307 (find_many_sub_basic_blocks): Update.
10308 * cfgcleanup.c (try_forward_edges): Update.
10309 (try_crossjump_to_edge): Update
10310 * cfgexpand.c (expand_gimple_cond): Update
10311 (expand_gimple_tailcall): Update
10312 (construct_exit_block): Update
10313 * cfghooks.c (verify_flow_info): Update
10314 (redirect_edge_succ_nodup): Update
10315 (split_edge): Update
10316 (make_forwarder_block): Update
10317 (duplicate_block): Update
10318 (account_profile_record): Update
10319 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10320 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10321 * cfgloopmanip.c (scale_loop_profile): Update.
10323 (lv_adjust_loop_entry_edge): Update.
10324 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10325 (force_nonfallthru_and_redirect): Update.
10326 (purge_dead_edges): Update.
10327 (rtl_flow_call_edges_add): Update.
10328 * cgraphunit.c (init_lowered_empty_function): Update.
10329 (cgraph_node::expand_thunk): Update.
10330 * gimple-pretty-print.c (dump_probability): Update.
10331 (dump_edge_probability): Update.
10332 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10333 * haifa-sched.c (sched_create_recovery_edges): Update.
10334 * hsa-gen.c (convert_switch_statements): Update.
10335 * ifcvt.c (dead_or_predicable): Update.
10336 * ipa-inline-transform.c (inline_transform): Update.
10337 * ipa-split.c (split_function): Update.
10338 * ipa-utils.c (ipa_merge_profiles): Update.
10339 * loop-doloop.c (add_test): Update.
10340 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10341 * lto-streamer-in.c (input_cfg): Update.
10342 (input_function): Update.
10343 * lto-streamer-out.c (output_cfg): Update.
10344 * modulo-sched.c (sms_schedule): Update.
10345 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10346 * predict.c (maybe_hot_edge_p): Update.
10347 (unlikely_executed_edge_p): Update.
10348 (probably_never_executed_edge_p): Update.
10349 (dump_prediction): Update.
10350 (drop_profile): Update.
10351 (propagate_unlikely_bbs_forward): Update.
10352 (determine_unlikely_bbs): Update.
10353 (force_edge_cold): Update.
10354 * profile.c (compute_branch_probabilities): Update.
10355 * reg-stack.c (better_edge): Update.
10356 * shrink-wrap.c (handle_simple_exit): Update.
10357 * tracer.c (better_p): Update.
10358 * trans-mem.c (expand_transaction): Update.
10359 (split_bb_make_tm_edge): Update.
10360 * tree-call-cdce.c: Update.
10361 * tree-cfg.c (gimple_find_sub_bbs): Update.
10362 (gimple_split_edge): Update.
10363 (gimple_duplicate_sese_region): Update.
10364 (gimple_duplicate_sese_tail): Update.
10365 (gimple_flow_call_edges_add): Update.
10366 (insert_cond_bb): Update.
10367 (execute_fixup_cfg): Update.
10368 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10369 * tree-complex.c (expand_complex_div_wide): Update.
10370 * tree-eh.c (lower_resx): Update.
10371 (unsplit_eh): Update.
10372 (cleanup_empty_eh_move_lp): Update.
10373 * tree-inline.c (copy_edges_for_bb): Update.
10374 (freqs_to_counts): Update.
10375 (copy_cfg_body): Update.
10376 * tree-ssa-dce.c (remove_dead_stmt): Update.
10377 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10378 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10379 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10380 (unloop_loops): Update.
10381 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10382 * tree-ssa-loop-split.c (connect_loops): Update.
10383 (split_loop): Update.
10384 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10385 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10386 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10387 * tree-ssa-reassoc.c (branch_fixup): Update.
10388 * tree-ssa-tail-merge.c (replace_block_by): Update.
10389 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10390 (compute_path_counts): Update.
10391 (update_profile): Update.
10392 (recompute_probabilities): Update.
10393 (update_joiner_offpath_counts): Update.
10394 (estimated_freqs_path): Update.
10395 (freqs_to_counts_path): Update.
10396 (clear_counts_path): Update.
10397 (ssa_fix_duplicate_block_edges): Update.
10398 (duplicate_thread_path): Update.
10399 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10400 (case_bit_test_cmp): Update.
10401 (collect_switch_conv_info): Update.
10402 (gen_inbound_check): Update.
10403 (do_jump_if_equal): Update.
10404 (emit_cmp_and_jump_insns): Update.
10405 * tree-tailcall.c (decrease_profile): Update.
10406 (eliminate_tail_call): Update.
10407 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10408 (vect_do_peeling): Update.
10409 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10410 * ubsan.c (ubsan_expand_null_ifn): Update.
10411 (ubsan_expand_ptr_ifn): Update.
10412 * value-prof.c (gimple_divmod_fixed_value): Update.
10413 (gimple_mod_pow2): Update.
10414 (gimple_mod_subtract): Update.
10415 (gimple_ic): Update.
10416 (gimple_stringop_fixed_value): Update.
10418 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10421 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10423 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10425 PR rtl-optimization/82395
10426 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10427 based on non_spilled_static_chain_regno_p.
10429 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10431 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10432 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10433 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10435 2017-10-19 Martin Sebor <msebor@redhat.com>
10437 PR tree-optimization/82596
10438 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10440 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10442 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10443 (get_mem_refs_of_builtin_call): Likewise.
10444 * builtins.c (expand_builtin_apply): Adjust call to
10445 allocate_dynamic_stack_space.
10446 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10447 the third argument to allocate_dynamic_stack_space, otherwise -1.
10448 (expand_builtin): Deal with all alloca variants.
10449 (is_inexpensive_builtin): Likewise.
10450 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10451 * calls.c (special_function_p): Deal with all alloca variants.
10452 (initialize_argument_information): Adjust call to
10453 allocate_dynamic_stack_space.
10454 (expand_call): Likewise.
10455 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10456 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10457 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10458 use it for the stack usage computation.
10459 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10460 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10461 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10462 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10463 (in_loop_p): Remove first argument and useless check.
10464 (pass_walloca::execute): Remove useless test and adjust call to above.
10465 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10466 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10467 (gimplify_call_expr): Deal with all alloca variants.
10468 * hsa-gen.c (gen_hsa_alloca): Likewise.
10469 (gen_hsa_insns_for_call): Likewise.
10470 * ipa-pure-const.c (special_builtin_state): Likewise.
10471 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10472 * tree-object-size.c (alloc_object_size): Likewise.
10473 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10474 (call_may_clobber_ref_p_1): Likewise.
10475 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10476 (ccp_fold_stmt): Likewise.
10477 (optimize_stack_restore): Likewise.
10478 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10479 (mark_all_reaching_defs_necessary_1): Likewise.
10480 (propagate_necessity): Likewise.
10481 (eliminate_unnecessary_stmts): Likewise.
10482 * tree.c (build_common_builtin_nodes): Build
10483 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10484 (build_alloca_call_expr): New function.
10485 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10486 (CASE_BUILT_IN_ALLOCA): Likewise.
10487 (build_alloca_call_expr): Declare.
10488 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10490 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10493 * dwarf2out.c (new_die_raw): New static inline function.
10494 (new_die): Use it to create the DIE.
10495 (add_AT_external_die_ref): Likewise.
10496 (clone_die): Likewise.
10497 (clone_as_declaration): Likewise.
10498 (dwarf2out_vms_debug_main_pointer): Likewise.
10499 (base_type_die): Likewise. Remove early return for corner cases.
10500 Do not call add_pubtype on the DIE here.
10501 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10502 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10503 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10504 native order exists for base types, attach the DIE manually and call
10505 add_pubtype on it. Do not equate a reverse order DIE to the type.
10507 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10509 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10510 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10512 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10514 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10516 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10517 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10518 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10519 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10520 * opts.c (sanitizer_opts): Add builtin.
10521 * ubsan.c (instrument_builtin): New function.
10522 (pass_ubsan::execute): Call it.
10523 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10524 * doc/invoke.texi: Document -fsanitize=builtin.
10526 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10527 builtins, store max (log2 (align), 0) into uchar field instead of
10528 align into uptr field.
10529 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10530 store uchar 0 field instead of uptr 0 field.
10531 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10532 instead of passing one address of struct with 2 locations pass
10533 two addresses of structs with 1 location each.
10534 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10535 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10536 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10537 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10538 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10539 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10540 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10541 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10543 2017-10-19 Martin Liska <mliska@suse.cz>
10546 * file-find.c (remove_prefix): Remove.
10547 * file-find.h (remove_prefix): Likewise.
10548 * gcc-ar.c: Remove smartness of lookup.
10550 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10552 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10553 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10554 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10556 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10559 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10560 (setcc + and to xor + setcc): New peephole2.
10562 2017-10-19 Tom de Vries <tom@codesourcery.com>
10564 * doc/sourcebuild.texi (Test Directives, Variants of
10565 dg-require-support): Add dg-require-stack-size.
10567 2017-10-19 Martin Liska <mliska@suse.cz>
10570 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10571 that have a large alignment.
10572 (gimplify_target_expr): Likewise.
10574 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10576 PR rtl-optimization/82602
10577 * ira.c (rtx_moveable_p): Return false for volatile asm.
10579 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10582 * config/i386/i386-modes.def (CCGZ): New CC mode.
10583 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10584 * config/i386/predicates.md (ix86_comparison_operator):
10586 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10587 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10588 with double-word subtraction.
10589 (put_condition_code): Handle CCGZmode.
10591 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10593 * wide-int.cc (debug (const wide_int &)): New.
10594 (debug (const wide_int *)): New.
10595 (debug (const widest_int &)): New.
10596 (debug (const widest_int *)): New.
10598 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10600 PR middle-end/82556
10601 * lra-constraints.c (curr_insn_transform): Use non-input operand
10602 instead of output one for matched reload.
10604 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10606 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10607 (tree-ssa-loop-ivopts.h): New header file.
10608 (struct builtin_info): New fields.
10609 (classify_builtin_1): Compute and record base and offset parts for
10610 memset builtin partition by calling strip_offset.
10611 (offset_cmp, fuse_memset_builtins): New functions.
10612 (finalize_partitions): Fuse adjacent memset partitions by calling
10614 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10615 Expose the interface.
10616 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10618 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10620 PR tree-optimization/82574
10621 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10622 that data reference must be executed exactly once per iteration
10623 against the outermost loop in nest.
10624 (classify_partition): Update call to above function.
10626 2017-10-18 Richard Biener <rguenther@suse.de>
10628 PR tree-optimization/82591
10629 * graphite.c (graphite_transform_loops): Move code gen message
10631 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10632 Here. Handle scop_to_isl_ast failing.
10633 (scop_to_isl_ast): Limit the number of ISL operations.
10635 2017-10-18 Richard Biener <rguenther@suse.de>
10637 * graphite-isl-ast-to-gimple.c
10638 (translate_isl_ast_to_gimple::set_rename): Simplify.
10639 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10640 (graphite_copy_stmts_from_block): ... here.
10641 (copy_bb_and_scalar_dependences): Simplify.
10642 (add_parameters_to_ivs_params): Canonicalize.
10643 (generate_entry_out_of_ssa_copies): Simplify.
10644 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10645 by passing in ISL dimension.
10646 (parameter_index_in_region_1): Rename to ...
10647 (parameter_index_in_region): ... this.
10648 (extract_affine): Adjust assert, pass down parameter index.
10649 (add_param_constraints): Use range-info when available.
10650 (build_scop_context): Adjust.
10651 * sese.c (new_sese_info): Adjust.
10652 (free_sese_info): Likewise.
10653 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10654 Remove unused typedefs.
10655 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10657 2017-10-18 Martin Liska <mliska@suse.cz>
10659 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10661 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10663 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10664 (s390_sched_init): Do not reset s390_sched_state if we entered the
10665 current basic block via a fallthru edge and all others are unlikely.
10667 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10669 * config/s390/s390.c (NUM_SIDES): New variable.
10670 (LONGRUNNING_THRESHOLD): New variable.
10671 (LATENCY_FACTOR): New variable.
10672 (s390_sched_score): Decrease score for long-running instructions on
10674 (s390_sched_variable_issue): Perform bookkeeping for long-running
10677 2017-10-18 Richard Biener <rguenther@suse.de>
10679 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10680 Simplify with removal of the parameter rename map.
10681 (set_rename): Likewise.
10682 (should_copy_to_new_region): Likewise.
10683 (graphite_copy_stmts_from_block): Likewise.
10684 (copy_bb_and_scalar_dependences): Remove initialization of
10685 unused copied_bb_map.
10686 (copy_def): Remove.
10687 (copy_internal_parameters): Likewise.
10688 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10689 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10690 Use INTEGRAL_TYPE_P.
10691 (parameter_index_in_region_1): Rename to ...
10692 (assign_parameter_index_in_region): ... this. Assert we have
10693 a parameter we handle.
10694 (scan_tree_for_params): Adjust.
10695 * sese.h (parameter_rename_map_t): Remove.
10696 (struct sese_info_t): Remove unused parameter_rename_map and
10697 copied_bb_map members.
10698 * sese.c (new_sese_info): Adjust.
10699 (free_sese_info): Likewise.
10701 2017-10-18 Martin Liska <mliska@suse.cz>
10704 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10705 on an abnormal edge.
10707 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10709 * doc/invoke.texi (ffunction-sections and fdata-sections):
10712 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10714 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10715 the use statement can throw internally.
10717 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10719 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10720 any RTX present on the RHS of a SET.
10721 * compare-elim.c (try_eliminate_compare): Restore comment.
10723 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10725 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10726 may be also called on tcc_type nodes.
10727 * langhooks.c (lhd_tree_size): Likewise.
10729 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10731 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10732 format_warning_at_substring.
10733 (maybe_warn): Convert source_range * param to a location_t. Pass
10734 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10735 (format_directive): Remove code to extract source_ranges and
10736 source_range * in favor of just a location_t.
10737 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10739 * substring-locations.c (format_warning_va): Convert
10740 source_range * param to a location_t.
10741 (format_warning_at_substring): Likewise.
10742 * substring-locations.h (format_warning_va): Likewise.
10743 (format_warning_at_substring): Likewise.
10745 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10747 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10749 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10750 and scatter/gather ops.
10752 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10753 vec_gather_load and vec_scatter_store.
10754 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10755 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10756 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10757 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10758 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10759 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10761 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10763 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10764 Detect FTST instruction and handle its register pops. Only pop
10765 second operand if can_pop_second_op is true.
10766 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10767 set can_pop_second_op to false in the compare_for_stack_reg call.
10769 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10770 output_fp_compare for stack register operands.
10771 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10772 instructions here. Do not emit stack register pops here. Assert
10773 that FCOMPP pops next to top stack register. Rewrite function.
10775 2017-10-17 Nathan Sidwell <nathan@acm.org>
10777 PR middle-end/82577
10778 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10779 use DECL_ASSEMBLER_NAME_RAW.
10781 PR middle-end/82546
10782 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10785 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10786 Wilco Dijkstra <wilco.dijkstra@arm.com>
10788 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10789 converstion to Pmode from the buf_addr.
10791 2017-10-17 Richard Biener <rguenther@suse.de>
10793 * graphite-dependences.c (scop_get_reads_and_writes): Change
10794 output parameters to references.
10796 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10798 PR 71026/tree-optimization
10799 * fold-const.c (distribute_real_division): Removed.
10800 (fold_binary_loc): Remove calls to distribute_real_divison.
10802 2017-10-17 Richard Biener <rguenther@suse.de>
10804 * graphite-scop-detection.c
10805 (scop_detection::stmt_has_simple_data_refs_p): Always use
10806 the full nest as region.
10807 (try_generate_gimple_bb): Likewise.
10808 * sese.c (scalar_evolution_in_region): Simplify now that
10809 SCEV can handle instantiation in regions.
10810 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10811 in the non-loop part of a function if requested.
10813 2017-10-17 Richard Biener <rguenther@suse.de>
10815 PR tree-optimization/82563
10816 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10818 (graphite_regenerate_ast_isl): Call it.
10819 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10821 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10823 PR tree-optimization/82549
10824 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10825 Formatting fixes. Instead of calling make_bit_field_ref with negative
10828 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10830 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10831 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10832 _mm_maskz_reduce_ss): New.
10833 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10834 __builtin_ia32_reducess_mask): Ditto..
10835 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10836 * config/i386/sse.md (reduces<mode>): Renamed to ...
10837 (reduces<mode><mask_scalar_name>): ... this.
10838 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10840 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10841 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10843 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10845 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10846 * selftest-run-tests.c (selftest::run_tests): Call
10847 selftest::unique_ptr_tests_cc_tests.
10848 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10849 * unique-ptr-tests.cc: New file.
10851 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
10854 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10856 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10857 (make_hard_regno_born, make_hard_regno_dead): Update
10858 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
10860 2017-10-16 Jeff Law <law@redhat.com>
10862 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10864 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
10866 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
10868 2017-10-16 Olivier Hainque <hainque@adacore.com>
10870 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
10871 with_cpu if we were configured for an e500v2 target cpu name.
10873 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
10875 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
10876 * doc/invoke.texi: Document +nodsp as a valid extension for
10879 2017-10-16 Martin Liska <mliska@suse.cz>
10881 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
10882 (test_set_range): Likewise.
10883 (test_range_functions): Rename to ...
10884 (test_bit_in_range): ... this.
10885 (sbitmap_c_tests): Add new test.
10887 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10889 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
10891 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
10892 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
10894 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10896 * config/aarch64/aarch64-builtins.c
10897 (aarch64_types_quadopu_lane_qualifiers): New.
10898 (TYPES_QUADOPU_LANE): New.
10899 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
10900 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
10901 (aarch64_<sur>dot_laneq<vsi2qi>): New.
10902 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
10903 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
10904 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
10905 (Vdottype, DOTPROD): New.
10906 (sur): Add SDOT and UDOT.
10908 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10910 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
10911 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
10912 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
10913 Add TARGET_DOTPROD.
10914 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
10915 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
10916 Enable TARGET_DOTPROD.
10917 (cortex-a75.cortex-a55): Likewise.
10918 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
10920 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10922 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
10923 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
10925 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
10927 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
10928 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
10929 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
10930 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
10931 * config/arm/types.md (neon_dot, neon_dot_q): New.
10932 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
10934 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10936 * config/arm/arm.h (TARGET_DOTPROD): New.
10937 * config/arm/arm.c (arm_arch_dotprod): New.
10938 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
10939 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
10940 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
10941 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
10942 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
10943 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
10944 * doc/invoke.texi (armv8.2-a): Document dotprod
10946 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10948 * i386.c (ix86_vec_cost): New function.
10949 (ix86_rtx_costs): Handle vector operations better.
10950 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
10951 * x86-tune-costs.h: Add new costs to all tables.
10953 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10955 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
10957 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
10958 divsd, sqrtss and sqrtsd
10959 * x86-tune-costs.h: Add new entries to all costs.
10960 (znver1_cost): Fix to match real instruction latencies.
10962 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10963 Michael Collison <michael.collison@arm.com>
10965 * compare-elim.c: Include emit-rtl.h.
10966 (can_merge_compare_into_arith): New function.
10967 (try_validate_parallel): Likewise.
10968 (try_merge_compare): Likewise.
10969 (try_eliminate_compare): Call the above when no previous clobber
10971 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
10974 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10976 PR middle-end/62263
10977 PR middle-end/82498
10978 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
10979 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
10981 PR middle-end/62263
10982 PR middle-end/82498
10983 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
10984 to be any operand_equal_p operands. For & (B - 1) require
10985 B to be power of 2. Recognize
10986 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
10988 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10991 * optabs.c (expand_memory_blockage): Fix call of
10992 targetm.have_memory_blockage.
10994 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10997 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
10998 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
10999 objects to extra_objs instead of overwriting it.
11001 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11003 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11004 Use any_fp_register_operand as operand[3] predicate. Simplify
11005 equality test for operands[2] and operands[4] memory location.
11006 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11007 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11008 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11009 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11010 any_fp_register_operand as operand[1] predicate. Simplify
11011 equality test for operands[0] and operands[3] memory location.
11012 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11013 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11014 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11016 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11018 * target-insns.def: Add memory_blockage.
11019 * optabs.c (expand_memory_blockage): New function.
11020 (expand_asm_memory_barrier): Rename ...
11021 (expand_asm_memory_blockage): ... to this.
11022 (expand_mem_thread_fence): Call expand_memory_blockage
11023 instead of expand_asm_memory_barrier.
11024 (expand_mem_singnal_fence): Ditto.
11025 (expand_atomic_load): Ditto.
11026 (expand_atomic_store): Ditto.
11027 * doc/md.texi (Standard Pattern Names For Generation):
11028 Document memory_blockage instruction pattern.
11030 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11032 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11033 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11034 * config/rl78/rl78.md: New define_expand "adddi3".
11036 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11038 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11041 2017-10-13 Jeff Law <law@redhat.com>
11043 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11045 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11048 * internal-fn.c (expand_mul_overflow): If both operands have
11049 the same highpart of -1 or 0 and the topmost bit of lowpart
11050 is different, overflow is if res <= 0 rather than res < 0.
11052 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11054 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11055 TARGET_P9_VECTOR code for unaligned_load case.
11057 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11059 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11061 2017-10-13 Nathan Sidwell <nathan@acm.org>
11063 * tree-core.h (tree_contains_struct): Make bool.
11064 * tree.c (tree_contains_struct): Likewise.
11065 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11066 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11067 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11068 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11069 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11071 2017-10-13 Richard Biener <rguenther@suse.de>
11073 * graphite-isl-ast-to-gimple.c
11074 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11075 parameters and dominance check.
11076 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11077 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11078 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11079 Do not update SSA form here or do intermediate IL verification.
11080 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11081 (graphite_initialize): Remove check on the number of loops in
11082 the function and inline into graphite_transform_loops.
11083 (graphite_finalize): Inline into graphite_transform_loops.
11084 (graphite_transform_loops): Perform SSA update and IL verification
11086 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11088 2017-10-13 Richard Biener <rguenther@suse.de>
11090 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11091 graphite_expression_type_precision): Avoid global constructor
11093 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11094 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11095 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11096 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11097 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11098 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11100 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11103 * config/i386/i386.h (ix86_red_zone_size): New.
11104 * config/i386/i386.md (push peephole2s): Replace
11105 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11107 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11108 Alan Hayward <alan.hayward@arm.com>
11109 David Sherwood <david.sherwood@arm.com>
11111 * combine.c (can_change_dest_mode): Reject changes in
11112 REGMODE_NATURAL_SIZE.
11114 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11115 Alan Hayward <alan.hayward@arm.com>
11116 David Sherwood <david.sherwood@arm.com>
11118 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11119 (expand_debug_source_expr): Likewise.
11120 * combine.c (combine_simplify_rtx): Likewise.
11121 * cse.c (fold_rtx): Likewise.
11122 * fwprop.c (canonicalize_address): Likewise.
11123 * targhooks.c (default_shift_truncation_mask): Likewise.
11125 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11126 Alan Hayward <alan.hayward@arm.com>
11127 David Sherwood <david.sherwood@arm.com>
11129 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11130 (widened_mode): Likewise.
11131 (expand_unop): Likewise.
11132 * ree.c (transform_ifelse): Likewise.
11133 (merge_def_and_ext): Likewise.
11134 (combine_reaching_defs): Likewise.
11135 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11137 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11138 Alan Hayward <alan.hayward@arm.com>
11139 David Sherwood <david.sherwood@arm.com>
11141 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11142 * combine.c (gen_lowpart_for_combine): Likewise.
11143 * dwarf2out.c (rtl_for_decl_location): Likewise.
11144 * final.c (alter_subreg): Likewise.
11145 * rtlhooks.c (gen_lowpart_general): Likewise.
11146 (gen_lowpart_if_possible): Likewise.
11148 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11149 Alan Hayward <alan.hayward@arm.com>
11150 David Sherwood <david.sherwood@arm.com>
11152 * calls.c (expand_call): Use subreg_lowpart_offset.
11153 * cse.c (cse_insn): Likewise.
11154 * regcprop.c (copy_value): Likewise.
11155 (copyprop_hardreg_forward_1): Likewise.
11157 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11160 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11161 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11162 =Q constraints to +Q and into insn condition add check
11163 that operands[0] and operands[1] are equal.
11164 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11165 =Q constraints to +Q and into insn condition add check
11166 that operands[0] is equal to either operands[1] or operands[2].
11169 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11170 instead of handling MINUS_EXPR twice (once for each argument),
11171 canonicalize operand order and handle just once, use rtype where
11172 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11175 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11176 any values of __C while still being pattern recognizable as a simple
11177 rotate instruction.
11179 2017-10-13 Richard Biener <rguenther@suse.de>
11181 PR tree-optimization/82451
11183 2017-10-02 Richard Biener <rguenther@suse.de>
11185 PR tree-optimization/82355
11186 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11187 a mapping for the enclosing loop but avoid generating one for
11188 the loop tree root.
11189 (copy_bb_and_scalar_dependences): Remove premature codegen
11190 error on PHIs in blocks duplicated into multiple places.
11191 * graphite-scop-detection.c
11192 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11193 in the region use it as loop and nest to analyze the DR in.
11194 (try_generate_gimple_bb): Likewise.
11195 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11196 (add_loop_constraints): For blocks in a loop not in the region
11197 create a dimension with a single iteration.
11198 * sese.h (gbb_loop_at_index): Remove assert.
11200 * cfgloop.c (loop_preheader_edge): For the loop tree root
11201 return the single successor of the entry block.
11202 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11203 Reset the SCEV hashtable and niters.
11204 * graphite-scop-detection.c
11205 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11206 assert that we only have POLYNOMIAL_CHREC that vary in loops
11207 contained in the region.
11208 (scop_detection::graphite_can_represent_expr): Adjust.
11209 (scop_detection::stmt_has_simple_data_refs_p): For loops
11210 not in the region set loop to NULL. The nest is now the
11211 entry edge to the region.
11212 (try_generate_gimple_bb): Likewise.
11213 * sese.c (scalar_evolution_in_region): Adjust for
11214 instantiate_scev change.
11215 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11216 Make nest parameter the edge into the region.
11217 (create_data_ref): Likewise.
11218 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11219 entry edge into a region and adjust instantiate_scev calls.
11220 (create_data_ref): Likewise.
11221 (graphite_find_data_references_in_stmt): Likewise.
11222 (find_data_references_in_stmt): Pass the loop preheader edge
11223 from the nest argument.
11224 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11225 parameter the edge into the region.
11226 (instantiate_parameters): Use the loop preheader edge as entry.
11227 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11229 (get_instantiated_value_entry): Make instantiate_below parameter
11230 the edge into the region.
11231 (instantiate_scev_name): Likewise. Adjust dominance checks,
11232 when we cannot use loop-based instantiation instantiate by
11233 walking use-def chains.
11234 (instantiate_scev_poly): Adjust.
11235 (instantiate_scev_binary): Likewise.
11236 (instantiate_scev_convert): Likewise.
11237 (instantiate_scev_not): Likewise.
11238 (instantiate_array_ref): Remove.
11239 (instantiate_scev_3): Likewise.
11240 (instantiate_scev_2): Likewise.
11241 (instantiate_scev_1): Likewise.
11242 (instantiate_scev_r): Do not blindly handle N-operand trees.
11243 Do not instantiate array-refs. Handle all constants and invariants.
11244 (instantiate_scev): Make instantiate_below parameter
11245 the edge into the region.
11246 (resolve_mixers): Use the loop preheader edge for the region
11247 parameter to instantiate_scev_r.
11248 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11250 2017-10-13 Richard Biener <rguenther@suse.de>
11252 PR tree-optimization/82525
11253 * graphite-isl-ast-to-gimple.c
11254 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11256 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11257 Fail code generation when we cannot represent the isl integer.
11258 (binary_op_to_tree): Elide modulo operations that are no-ops
11259 in the type we code generate. Remove now superfluous code
11262 2017-10-13 Richard Biener <rguenther@suse.de>
11264 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11265 (scop_detection::harmful_loop_in_region): Remove premature
11266 IV type restriction.
11267 (scop_detection::graphite_can_represent_scev): We can handle
11268 pointer IVs just fine.
11270 2017-10-13 Alan Modra <amodra@gmail.com>
11272 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11273 "Clobbers and Scratch Registers". Add paragraph on
11274 alternative to clobbers for scratch registers and OpenBLAS
11277 2017-10-13 Alan Modra <amodra@gmail.com>
11279 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11280 example of a memory input for a string of known length. Move
11281 commentary out of table. Add a number of new examples
11282 covering array memory inputs.
11284 2017-10-12 Martin Liska <mliska@suse.cz>
11286 PR tree-optimization/82493
11287 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11288 (test_range_functions): New function.
11289 (sbitmap_c_tests): Likewise.
11290 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11291 * selftest.h (sbitmap_c_tests): New function.
11293 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11295 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11297 * config/rs6000/amo.h: Fix spacing issue.
11299 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11302 * config/i386/i386.md (*ashl<mode>3_mask_1,
11303 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11304 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11307 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11309 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11310 (profile_probability): Set max_probability
11311 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11313 * profile-count.c (profile_probability::differs_from_p): Do not
11314 rely on max_probaiblity == 10000
11316 2017-10-12 Jeff Law <law@redhat.com>
11318 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11321 2017-10-12 Martin Sebor <msebor@redhat.com>
11325 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11326 (handle_alias_pairs): Call it.
11327 * common.opt (-Wattribute-alias): New option.
11328 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11329 * doc/invoke.texi (-Wattribute-alias): Document.
11331 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11334 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11336 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11338 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11339 (make_hard_regno_born, make_hard_regno_dead): Update
11340 bb_killed_pseudos and bb_gen_pseudos.
11342 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11344 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11346 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11348 * config/alpha/alpha.c (alpha_split_conditional_move):
11349 Use std::swap instead of manually swapping.
11350 (alpha_stdarg_optimize_hook): Ditto.
11351 (alpha_canonicalize_comparison): Ditto.
11353 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11355 * tree-loop-distribution.c (struct builtin_info): New struct.
11356 (struct partition): Refactor fields into struct builtin_info.
11357 (partition_free): Free struct builtin_info.
11358 (build_size_arg_loc, build_addr_arg_loc): Delete.
11359 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11360 information from struct builtin_info.
11361 (find_single_drs): New function refactored from classify_partition.
11362 Also moved builtin validity checks to this function.
11363 (compute_access_range, alloc_builtin): New functions.
11364 (classify_builtin_st, classify_builtin_ldst): New functions.
11365 (classify_partition): Refactor code into functions find_single_drs,
11366 classify_builtin_st and classify_builtin_ldst.
11367 (distribute_loop): Don't do runtime alias check when distributing
11369 (find_seed_stmts_for_distribution): New function.
11370 (pass_loop_distribution::execute): Refactor code finding seed
11371 stmts into above function. Support distribution for the innermost
11372 two-level loop nest. Adjust dump information.
11374 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11376 * tree-loop-distribution.c: Adjust the general comment.
11377 (NUM_PARTITION_THRESHOLD): New macro.
11378 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11379 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11380 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11381 in call to build_partition_graph.
11382 (finalize_partitions): New parameter. Make loop distribution more
11383 conservative by fusing more partitions.
11384 (distribute_loop): Don't do runtime alias check in case of loop nest
11386 (find_seed_stmts_for_distribution): New function.
11387 (prepare_perfect_loop_nest): New function.
11388 (pass_loop_distribution::execute): Refactor code finding seed stmts
11389 and loop nest into above functions. Support loop nest distribution.
11390 Adjust dump information accordingly.
11392 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11394 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11395 and set PTYPE_SEQUENTIAL for merged partition.
11397 2017-10-12 Richard Biener <rguenther@suse.de>
11399 PR tree-optimization/69728
11401 2017-09-19 Richard Biener <rguenther@suse.de>
11403 PR tree-optimization/69728
11404 * graphite-sese-to-poly.c (schedule_error): New global.
11405 (add_loop_schedule): Handle empty domain by failing the
11407 (build_original_schedule): Handle schedule_error.
11409 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11410 domain by returning an unchanged schedule.
11412 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11414 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11415 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11417 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11419 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11422 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11425 * expr.c (store_field): Don't optimize away bitsize == 0 store
11426 from CALL_EXPR with addressable return type.
11428 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11430 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11431 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11432 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11433 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11434 TARGET_ISEL instead of TARGET_ISEL<sel>.
11436 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11438 * config/rs6000/rs6000.c
11439 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11441 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11443 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11445 (reg_or_cint_operand): Fix comment.
11446 (reg_or_zero_operand): New predicate.
11447 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11448 * config/rs6000/rs6000.c (output_isel): Delete.
11449 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11450 instead of reg_or_cint_operand. Output instruction directly (not via
11452 (isel_unsigned_<mode>): Ditto.
11453 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11454 gpc_reg_operand. Add an instruction alternative for this. Output
11455 instruction directly.
11456 (*isel_reversed_unsigned_<mode>): Ditto.
11458 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11460 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11461 (TARGET_CANONICALIZE_COMPARISON): Define.
11463 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11466 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11467 Check whether the dest is REG before adding REG_EQUIV note.
11469 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11472 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11474 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11475 (make_hard_regno_born, make_hard_regno_dead): Update
11476 bb_killed_pseudos and bb_gen_pseudos.
11478 2017-10-11 Nathan Sidwell <nathan@acm.org>
11480 * incpath.h (enum incpath_kind): Name enum, prefix values.
11481 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11482 * incpath.c (heads, tails): Use INC_MAX.
11483 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11484 (merge_include_chains, split_quote_chain,
11485 register_include_chains): Update incpath_kind names.
11486 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11487 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11489 (add_framework_path, darwin_register_objc_includes): Likewise.
11490 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11492 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11494 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11495 Do not use float_operator operator predicate.
11496 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11497 * config/i386/predicates.md (float_operator): Remove predicate.
11499 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11501 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11502 (*jccxf_i387): Ditto.
11503 (*jcc<mode>_i387): Ditto.
11504 (*jccu<mode>_i387): Ditto.
11505 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11506 (*jcc_*_i387 splitters): Remove.
11507 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11508 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11509 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11512 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11514 * profile-count.h (slow_safe_scale_64bit): New function.
11515 (safe_scale_64bit): New inline.
11516 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11517 * profile-count.c: Include wide-int.h
11518 (slow_safe_scale_64bit): New.
11520 2017-10-11 Nathan Sidwell <nathan@acm.org>
11522 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11523 HAS_DECL_ASSEMBLER_NAME_P.
11524 * gimple-expr.c (gimple_decl_printable_name: Check
11525 HAS_DECL_ASSEMBLER_NAME_P too.
11526 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11527 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11528 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11529 * passes.c (rest_of_decl_compilation): Check
11530 HAS_DECL_ASSEMBLER_NAME_P too.
11531 * recog.c (verify_changes): Likewise.
11532 * tree-pretty-print.c (dump_decl_name): Likewise.
11533 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11535 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11536 (SET_DECL_ASSEMBLER_NAME): Use it.
11537 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11538 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11539 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11541 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11543 * config.gcc (i386, x86_64): Add extra objects.
11544 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11545 (ix86_min_insn_size): Declare.
11546 (ix86_issue_rate): Declare.
11547 (ix86_adjust_cost): Declare.
11548 (ia32_multipass_dfa_lookahead): Declare.
11549 (ix86_macro_fusion_p): Declare.
11550 (ix86_macro_fusion_pair_p): Declare.
11551 (ix86_bd_has_dispatch): Declare.
11552 (ix86_bd_do_dispatch): Declare.
11553 (ix86_core2i7_init_hooks): Declare.
11554 (ix86_atom_sched_reorder): Declare.
11555 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11556 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11557 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11558 (rip_relative_addr_p): Rename to ...
11559 (ix86_rip_relative_addr_p): ... this one; export.
11560 (memory_address_length): Update.
11561 (ix86_issue_rate): Move to x86-tune-sched.c.
11562 (ix86_flags_dependent): Move to x86-tune-sched.c.
11563 (ix86_agi_dependent): Move to x86-tune-sched.c.
11564 (exact_dependency_1): Move to x86-tune-sched.c.
11565 (exact_store_load_dependency): Move to x86-tune-sched.c.
11566 (ix86_adjust_cost): Move to x86-tune-sched.c.
11567 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11568 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11569 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11570 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11571 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11572 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11573 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11574 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11575 (min_insn_size): Rename to ...
11576 (ix86_min_insn_size): ... this one; export.
11577 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11578 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11579 (core2i7_first_cycle_multipass_backtrack): Move to
11580 x86-tune-sched-core.c.
11581 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11582 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11583 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11584 (ix86_avoid_jump_mispredicts): Update.
11585 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11586 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11587 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11588 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11589 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11590 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11591 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11592 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11593 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11594 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11595 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11596 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11597 (BIG): Move to ix86-tune-sched-bd.c.
11598 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11599 (enum insn_path): Move to ix86-tune-sched-bd.c.
11600 (get_mem_group): Move to ix86-tune-sched-bd.c.
11601 (is_cmp): Move to ix86-tune-sched-bd.c.
11602 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11603 (is_branch): Move to ix86-tune-sched-bd.c.
11604 (is_prefetch): Move to ix86-tune-sched-bd.c.
11605 (init_window): Move to ix86-tune-sched-bd.c.
11606 (allocate_window): Move to ix86-tune-sched-bd.c.
11607 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11608 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11609 (process_end_window): Move to ix86-tune-sched-bd.c.
11610 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11611 (find_constant): Move to ix86-tune-sched-bd.c.
11612 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11613 (has_immediate): Move to ix86-tune-sched-bd.c.
11614 (get_insn_path): Move to ix86-tune-sched-bd.c.
11615 (get_insn_group): Move to ix86-tune-sched-bd.c.
11616 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11617 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11618 (add_insn_window): Move to ix86-tune-sched-bd.c.
11619 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11620 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11621 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11622 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11623 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11624 (do_dispatch): Move to ix86-tune-sched-bd.c.
11625 (has_dispatch): Move to ix86-tune-sched-bd.c.
11626 * config/i386/t-i386: Add new object files.
11627 * config/i386/x86-tune-costs.h: New file.
11628 * config/i386/x86-tune-sched-atom.c: New file.
11629 * config/i386/x86-tune-sched-bd.c: New file.
11630 * config/i386/x86-tune-sched-core.c: New file.
11631 * config/i386/x86-tune-sched.c: New file.
11633 2017-10-11 Liu Hao <lh_mouse@126.com>
11635 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11636 the generic version below instead.
11637 (should_colorize): Recognize Windows consoles as terminals
11639 * pretty-print.c [__MINGW32__] (write_all): New function.
11640 [__MINGW32__] (find_esc_head): Likewise.
11641 [__MINGW32__] (find_esc_terminator): Likewise.
11642 [__MINGW32__] (eat_esc_sequence): Likewise.
11643 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11645 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11648 2017-10-11 Richard Biener <rguenther@suse.de>
11650 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11651 Properly call analyze_scalar_evolution with the loop of the stmt.
11653 2017-10-11 Richard Biener <rguenther@suse.de>
11655 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11656 * tree-core.h (tree_base): Add chrec_var union member.
11657 * tree.h (CHREC_VAR): Remove.
11658 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11659 * tree-chrec.h (build_polynomial_chrec): Adjust.
11660 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11661 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11663 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11665 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11666 * match.pd: ... here.
11667 ((T) X == (T) Y): Relax condition.
11669 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11671 PR tree-optimization/82472
11672 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11674 (break_alias_scc_partitions): Update postorder number.
11676 2017-10-11 Martin Liska <mliska@suse.cz>
11679 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11681 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11683 2017-10-11 Martin Liska <mliska@suse.cz>
11685 * print-rtl.c (print_insn): Move declaration of idbuf
11686 to same scope as name.
11688 2017-10-11 Martin Liska <mliska@suse.cz>
11693 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11694 volatile arguments.
11696 2017-10-11 Martin Liska <mliska@suse.cz>
11699 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11700 volatile arguments.
11702 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11704 * config.gcc (default_gnu_indirect_function): Default to yes for
11705 arm*-*-linux* with glibc.
11707 2017-10-11 Richard Biener <rguenther@suse.de>
11709 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11710 default-defs and types we do not want to analyze.
11711 (interpret_loop_phi): Replace unreachable code with an assert.
11712 (compute_scalar_evolution_in_loop): Remove and inline ...
11713 (analyze_scalar_evolution_1): ... here, replacing condition with
11714 what makes the intent clearer. Remove handling of cases
11715 get_scalar_evolution now handles.
11717 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11719 PR rtl-optimization/81434
11720 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11721 comment for main loop. In sched_group_found if, also add checks for
11722 pass and min_cost_group.
11724 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11726 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11727 (rs6000_insn_cost): New function.
11728 * config/rs6000/rs6000.md (cost): New attribute.
11730 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11731 H.J. Lu <hongjiu.lu@intel.com>
11735 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11736 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11737 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11738 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11739 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11740 OPTION_MASK_ISA_MMX and some other bit set require both
11741 mmx and the ISAs without the mmx bit.
11742 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11743 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11744 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11745 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11746 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11747 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11748 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11749 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11750 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11751 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11752 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11753 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11754 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11755 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11756 Add OPTION_MASK_ISA_MMX.
11758 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11760 * config.gcc (armv7*-*-freebsd*): New target.
11761 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11763 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11765 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11766 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11769 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11771 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11772 a template parameter.
11773 (WIDE_INT_REF_FOR): Update accordingly.
11774 * tree.h (wi::int_traits <const_tree>): Delete.
11775 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11776 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11777 (wi::tree_to_wide_ref): New typedef.
11778 (wi::to_wide): New function.
11779 * calls.c (get_size_range): Use wi::to_wide when operating on
11780 trees as wide_ints.
11781 * cgraph.c (cgraph_node::create_thunk): Likewise.
11782 * config/i386/i386.c (ix86_data_alignment): Likewise.
11783 (ix86_local_alignment): Likewise.
11784 * dbxout.c (stabstr_O): Likewise.
11785 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11786 * expr.c (const_vector_from_tree): Likewise.
11787 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11788 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11789 (fold_negate_expr_1, int_const_binop_1, const_binop)
11790 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11791 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11792 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11793 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11794 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11795 (fold_not_const, round_up_loc): Likewise.
11796 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11797 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11798 (alloca_call_type): Likewise.
11799 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11800 * godump.c (go_output_typedef): Likewise.
11801 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11802 * internal-fn.c (get_min_precision): Likewise.
11803 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11804 * ipa-polymorphic-call.c
11805 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11806 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11807 (ipa_modify_call_arguments): Likewise.
11808 * match.pd: Likewise.
11809 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11810 * print-tree.c (print_node_brief, print_node): Likewise.
11811 * stmt.c (expand_case): Likewise.
11812 * stor-layout.c (layout_type): Likewise.
11813 * tree-affine.c (tree_to_aff_combination): Likewise.
11814 * tree-cfg.c (group_case_labels_stmt): Likewise.
11815 * tree-data-ref.c (dr_analyze_indices): Likewise.
11816 (prune_runtime_alias_test_list): Likewise.
11817 * tree-dump.c (dequeue_and_dump): Likewise.
11818 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11819 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11820 * tree-pretty-print.c (dump_generic_node): Likewise.
11821 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11822 (simple_iv_with_niters): Likewise.
11823 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11824 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11825 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11826 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11827 (refine_value_range_using_guard, number_of_iterations_ne_max)
11828 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11829 (get_cst_init_from_scev, record_nonwrapping_iv)
11830 (scev_var_range_cant_overflow): Likewise.
11831 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11832 * tree-ssa-pre.c (compute_avail): Likewise.
11833 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11834 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11835 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11836 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11837 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11838 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11839 (dump_case_nodes, try_switch_expansion): Likewise.
11840 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11841 (vect_do_peeling): Likewise.
11842 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11843 * tree-vect-stmts.c (vectorizable_load): Likewise.
11844 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11845 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11846 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11847 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11848 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11849 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11850 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
11851 (evrp_dom_walker::before_dom_children): Likewise.
11852 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
11853 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
11854 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
11855 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
11856 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
11857 (get_range_pos_neg): Likewise.
11858 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
11859 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
11860 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
11861 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
11862 * config/avr/avr.c (avr_fold_builtin): Likewise.
11863 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
11864 * config/msp430/msp430.c (msp430_attr): Likewise.
11865 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
11866 * config/powerpcspe/powerpcspe-c.c
11867 (altivec_resolve_overloaded_builtin): Likewise.
11868 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
11869 (rs6000_expand_ternop_builtin): Likewise.
11870 * config/rs6000/rs6000-c.c
11871 (altivec_resolve_overloaded_builtin): Likewise.
11872 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
11873 (rs6000_expand_ternop_builtin): Likewise.
11874 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
11876 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11878 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
11879 when copying loop nest with only one inner loop.
11881 2017-10-10 Richard Biener <rguenther@suse.de>
11883 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
11884 blocks if SCEV is active.
11885 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
11887 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
11888 (scev_initialize): Assert we are not yet initialized.
11890 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11892 * tree-loop-distribution.c (generate_loops_for_partition): Remove
11893 inner loop's exit stmt by making it always exit the loop, otherwise
11894 we would generate an infinite empty loop.
11896 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11898 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
11899 renaming variables in new preheader if it's deleted.
11901 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11903 * tree-loop-distribution.c (struct partition): Remove unused field
11904 loops of the structure.
11905 (partition_alloc, partition_free): Ditto.
11906 (build_rdg_partition_for_vertex): Ditto.
11908 2017-10-09 Jeff Law <law@redhat.com>
11910 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
11911 return type to match prototype and documentation.
11913 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11915 * config/rs6000/rs6000.c (processor_costs): Move to ...
11916 * config/rs6000/rs6000.h: ... here.
11917 (rs6000_cost): Declare.
11919 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
11921 * except.c (setjmp_fn): New global variable.
11922 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
11923 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
11924 if DONT_USE_BUILTIN_SETJMP is defined.
11926 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11928 * target.def (insn_cost): New hook.
11929 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
11930 * doc/tm.texi: Regenerate.
11931 * rtlanal.c (insn_cost): Use the new hook.
11933 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11935 * combine.c (combine_validate_cost): Compute the new insn_cost,
11936 not just pattern_cost.
11937 (try_combine): Adjust comment.
11939 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11941 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
11943 * combine.c (uid_insn_cost): Adjust comment.
11944 (combine_validate_cost): Adjust comment. Use pattern_cost instead
11946 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
11947 * dse.c (find_shift_sequence): Ditto.
11948 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
11949 (bb_valid_for_noce_process_p): Use pattern_cost.
11950 * rtl.h (insn_rtx_cost): Delete.
11951 (pattern_cost): New prototype.
11952 (insn_cost): New prototype.
11953 * rtlanal.c (insn_rtx_cost): Rename to...
11954 (pattern_cost): ... this.
11957 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
11959 * config/i386/i386.md (*jcc_2): Remove insn pattern.
11960 (*jcc<mode>_0_r_i387): Ditto.
11961 (*jccxf_r_i387): Ditto.
11962 (*jcc<mode>_r_i387): Ditto.
11963 (*jccu<mode>_r_i387): Ditto.
11964 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
11965 (*jcc): Rename from *jcc_1.
11967 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11969 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
11970 deferred rescans after the lvx/stvx recombination pre-pass.
11972 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
11974 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
11975 memory operation instruction support.
11976 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
11977 (rs6000-ibm-aix[789]*): Likewise.
11978 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
11979 Document new functions.
11981 2017-10-09 Richard Biener <rguenther@suse.de>
11983 PR tree-optimization/82397
11984 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
11985 equality only for semantically equal trees.
11987 2017-10-09 Richard Biener <rguenther@suse.de>
11989 PR tree-optimization/82449
11990 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
11991 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
11992 allow constant addresses.
11993 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
11996 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11998 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12001 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12004 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12007 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12010 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12012 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12015 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12016 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12018 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12020 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12021 (WI_BINARY_PREDICATE_RESULT): Likewise.
12022 (wi::binary_traits::operator_result): New type.
12023 (wi::binary_traits::predicate_result): Likewise.
12024 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12025 (generic_wide_int::operator==, generic_wide_int::operator!=)
12026 (generic_wide_int::operator&, generic_wide_int::and_not)
12027 (generic_wide_int::operator|, generic_wide_int::or_not)
12028 (generic_wide_int::operator^, generic_wide_int::operator+
12029 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12031 (operator~, unary operator-, operator==, operator!=, operator&)
12032 (operator|, operator^, operator+, binary operator-, operator*): New
12034 * expr.c (get_inner_reference): Use wi::bit_and_not.
12035 * fold-const.c (fold_binary_loc): Likewise.
12036 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12037 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12038 (bit_value_binop): Likewise.
12039 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12040 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12041 (extract_range_from_binary_expr_1): Likewise.
12042 (masked_increment): Likewise.
12043 (simplify_bit_ops_using_ranges): Likewise.
12045 2017-10-09 Martin Jambor <mjambor@suse.cz>
12048 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12049 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12050 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12052 (hsa_fixup_mov_insn_type): New function.
12053 (hsa_op_with_type::get_in_type): Use it.
12054 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12055 immediates in an assert.
12056 (hsa_op_with_type::extend_int_to_32bit): New method.
12057 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12058 types. Convert to dest type if necessary.
12059 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12060 (reg_for_gimple_ssa): Pass false as min32int to
12061 hsa_type_for_scalar_tree_type.
12062 (gen_hsa_addr): Fixup type when creating addresable temporary.
12063 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12064 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12065 necessary. Call hsa_fixup_mov_insn_type.
12066 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12067 extend operands and convert to dest type if necessary.
12068 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12069 to dest type if necessary.
12070 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12071 if conversion nt necessary and size matches.
12072 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12073 to dest type if necessary.
12074 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12075 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12077 (gen_hsa_clrsb): Likewise.
12078 (gen_hsa_ffs): Likewise.
12079 (gen_hsa_divmod): Extend operands and convert to dest type if
12081 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12083 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12085 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12086 Remove empty default arguments. Use a brace block as output
12088 (conditional return): Ditto.
12090 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12091 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12093 (group_ending_nop): Ditto.
12094 (doloop_end): Ditto.
12095 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12096 (splitters for those): Ditto.
12098 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12100 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12101 a conditional jump (and the compare for it) so that pc_rtx is the
12103 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12104 for the deleted and renamed ctr<mode>_internal[234] patterns.
12105 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12106 Delete second conditional return pattern.
12107 (ctr<mode>_internal2): Delete this second bdnz pattern.
12108 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12109 (ctr<mode>_internal4): Delete this second bdz pattern.
12111 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12113 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12114 (always_initialized_rtx_for_ssa_name_p): New predicate.
12115 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12116 (finish_out_of_ssa): Free new field of SA.
12117 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12118 * tree-ssa-coalesce.c: Include tree-ssa.h.
12119 (get_parm_default_def_partitions): Remove extern keyword.
12120 (get_undefined_value_partitions): New function.
12121 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12122 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12123 uninitialized bits.
12124 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12126 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12128 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12130 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12132 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12133 for targets that preffer 128bit.
12135 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12137 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12139 2017-10-08 Olivier Hainque <hainque@adacore.com>
12141 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12142 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12144 (thumb_set_return_address): Likewise.
12146 2017-10-08 Olivier Hainque <hainque@adacore.com>
12148 * common/config/arm/arm-common.c (arm_except_unwind_info):
12149 Handle DWARF2_UNWIND_INFO.
12151 2017-10-07 Michael Collison <michael.collison@arm.com>
12153 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12156 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12158 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12159 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12160 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12161 defined, force the creation of a new block for a dispatch label.
12163 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12165 * invoke.texi (Wsuggest-attribute=cold): Document.
12166 * common.opt (Wsuggest-attribute=cold): New
12167 * ipa-pure-const.c (warn_function_cold): New function.
12168 * predict.c (compute_function_frequency): Use it.
12169 * predict.h (warn_function_cold): Declare.
12171 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12173 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12176 2017-10-06 Martin Liska <mliska@suse.cz>
12178 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12179 keyword for member functions.
12180 (struct sanopt_tree_couple): New struct.
12181 (struct sanopt_tree_couple_hash): New function.
12182 (struct sanopt_ctx): Add new hash_map.
12183 (has_dominating_ubsan_ptr_check): New function.
12184 (record_ubsan_ptr_check_stmt): Likewise.
12185 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12186 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12187 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12189 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12192 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12193 aarch64_simd_valid_immediate on CONST_VECTORs.
12194 (aarch64_reg_or_bic_imm): Likewise.
12196 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12198 PR rtl-optimization/82396
12199 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12201 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12203 * graphite-dependences.c (scop_get_reads): Move code to...
12204 (scop_get_must_writes): Move code to...
12205 (scop_get_may_writes): Move code to...
12206 (scop_get_reads_and_writes): ... here.
12207 (scop_get_dependences): Call scop_get_reads_and_writes.
12209 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12211 PR tree-optimization/82434
12212 * fold-const.h (can_native_encode_type_p,
12213 can_native_encode_string_p): Remove.
12214 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12215 don't encode anything, just return what would be otherwise returned.
12216 (native_encode_fixed, native_encode_complex, native_encode_vector):
12218 (native_encode_string): Likewise. Inline by hand
12219 can_native_encode_string_p.
12220 (can_native_encode_type_p): Remove.
12221 (can_native_encode_string_p): Remove.
12222 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12223 STRING_CSTs using can_native_encode_string_p, test all
12224 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12225 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12226 argument from native_encode_expr.
12227 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12228 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12231 2017-10-06 Richard Biener <rguenther@suse.de>
12233 PR tree-optimization/82397
12234 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12235 operand_equal_p but rely on data_ref_compare_tree for detecting
12237 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12238 to match up with dr_group_sort_cmp.
12240 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12243 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12245 * config/s390/s390-builtin-types.def: Regenerate.
12247 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12250 * config/s390/s390-builtin-types.def: Regenerate.
12251 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12252 Change flag from B_VXE to B_VX.
12253 (s390_vec_min_dbl): Remove B_VXE flag.
12255 2017-10-06 Richard Biener <rguenther@suse.de>
12257 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12258 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12259 translate_isl_ast_to_gimple::is_valid_rename,
12260 translate_isl_ast_to_gimple::get_rename,
12261 translate_isl_ast_to_gimple::get_def_bb_for_const,
12262 translate_isl_ast_to_gimple::get_new_name,
12263 translate_isl_ast_to_gimple::collect_all_ssa_names,
12264 translate_isl_ast_to_gimple::copy_loop_phi_args,
12265 translate_isl_ast_to_gimple::collect_all_ssa_names,
12266 translate_isl_ast_to_gimple::copy_loop_phi_args,
12267 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12268 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12269 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12270 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12271 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12272 translate_isl_ast_to_gimple::copy_cond_phi_args,
12273 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12274 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12275 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12276 translate_isl_ast_to_gimple::rename_uses,
12277 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12278 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12279 (set_rename_for_each_def): Likewise.
12280 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12281 here. Handle rewriting SCEV analyzable uses here.
12282 (copy_bb_and_scalar_dependences): Generate code for PHI
12284 (graphite_regenerate_ast_isl): Adjust.
12285 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12286 (add_write, add_read): New functions.
12287 (build_cross_bb_scalars_def): Use it and simplify.
12288 (build_cross_bb_scalars_use): Likewise.
12289 (graphite_find_cross_bb_scalar_vars): Inline into...
12290 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12291 simulating out-of-SSA. Compute liveout and add dependencies.
12292 (build_scops): Force an empty entry block.
12293 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12295 (sese_build_liveouts): Declare.
12296 (sese_trivially_empty_bb_p): Likewise.
12297 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12298 compute liveout and debug_liveout.
12299 (sese_bad_liveouts_use): Remove.
12300 (sese_reset_debug_liveouts_bb): Likewise.
12301 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12302 (sese_build_liveouts): Build liveout and debug_liveout and store
12304 (new_sese_info): Adjust.
12305 (free_sese_info): Likewise.
12306 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12307 do not build liveout here.
12308 (move_sese_in_condition): Adjust region entry.
12309 (scev_analyzable_p): Match up with chrec_apply requirements.
12310 (sese_trivially_empty_bb_p): New.
12311 * tree-into-ssa.c (get_reaching_def): Properly support generating
12312 default-defs for incremental rewrite of anonymous names.
12314 2017-10-06 Richard Biener <rguenther@suse.de>
12316 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12317 precision do not perform modulo reduction.
12319 2017-10-06 Richard Biener <rguenther@suse.de>
12321 PR tree-optimization/82436
12322 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12323 conservatively choose the vectorization factor when checking
12324 whether we can perform the required load permutation.
12325 (vect_transform_slp_perm_load): Assert when we may not fail.
12327 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12329 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12330 message for incompatible -msdata=* and -mcall-* options.
12332 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12334 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12335 rate for post-reload scheduling.
12337 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12339 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12341 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12343 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12344 to improve monte carlo in scimark.
12346 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12348 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12349 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12350 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12351 pentium4_cost, nocona_cost): Set reassociation width to 1.
12352 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12353 width to 2 for fp operations and 1 otherwise.
12354 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12356 (atom_cost): Set reassociation width to 2.
12357 (slm_cost, generic_cost): Set fp reassociation width
12358 to 2 and 1 otherwise.
12359 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12360 (core_cost): Set fp reassociation width to 4 and vector to 2.
12361 (ix86_reassociation_width): Rewrite using cost table; special case
12362 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12363 and TARGET_AVX128_OPTIMAL.
12364 * config/i386/i386.h (processor_costs): Add
12365 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12366 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12367 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12368 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12369 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12370 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12372 2017-10-05 Nathan Sidwell <nathan@acm.org>
12374 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12376 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12378 * config/arm/arm.c (arm_test_fpu_data): New.
12379 (arm_run_selftests): Call arm_test_fpu_data.
12381 2017-10-04 Nathan Sidwell <nathan@acm.org>
12383 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12385 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12387 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12389 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12390 check type for aarch64_simd_valid_immediate.
12391 (aarch64_output_simd_mov_immediate): Update prototype.
12392 (aarch64_simd_valid_immediate): Update prototype.
12393 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12394 support for ORR-immediate.
12395 (and<mode>3): modified pattern to add support for BIC-immediate.
12396 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12397 now checks for valid immediate for BIC and ORR based on new enum
12399 (aarch64_output_simd_mov_immediate): Function now used to output
12400 BIC/ORR imm as well based on new enum argument.
12401 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12403 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12404 (aarch64_reg_or_bic_imm): Likewise.
12406 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12408 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12409 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12411 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12415 PR rtl-optimization/82396
12416 * haifa-sched.c (autopref_multipass_init): Simplify
12418 (autopref_rank_data): Simplify sort order.
12419 * sched-int.h (autopref_multipass_data_): Remove
12420 multi_mem_insn_p, min_offset and max_offset.
12422 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12424 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12426 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12428 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12429 vect_doubleint_cvt.
12431 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12433 * doc/sourcebuild.texi: Document vect_long_mult.
12435 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12437 PR tree-optimization/82413
12438 * fold-const.c (build_range_check): Use widest_int when comparing
12439 the maximum ETYPE value with HIGH.
12441 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12443 PR rtl-optimization/82396
12444 * haifa-sched.c (autopref_multipass_init): Simplify
12446 (autopref_rank_data): Simplify sort order.
12447 * sched-int.h (autopref_multipass_data_): Remove
12448 multi_mem_insn_p, min_offset and max_offset.
12450 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12452 PR tree-optimization/82381
12453 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12454 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12457 PR tree-optimization/82374
12458 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12459 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12460 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12461 current_function_decl to the new decl.
12463 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12465 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12466 helper macro for IEEE float128 hardware built-in functions.
12467 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12469 (TRUNCF128_ODD): Likewise.
12470 (ADDF128_ODD): Likewise.
12471 (SUBF128_ODD): Likewise.
12472 (MULF128_ODD): Likewise.
12473 (DIVF128_ODD): Likewise.
12474 (FMAF128_ODD): Likewise.
12475 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12476 UNSPEC_TRUNC_ROUND_TO_ODD.
12477 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12478 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12479 floating point round to odd instructions.
12480 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12481 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12482 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12483 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12484 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12485 (trunc<mode>sf2_hw): Change the truncate with round to odd
12486 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12487 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12488 to odd hardware instructions.
12489 (sub<mode>3_odd): Likewise.
12490 (mul<mode>3_odd): Likewise.
12491 (div<mode>3_odd): Likewise.
12492 (sqrt<mode>2_odd): Likewise.
12493 (fma<mode>4_odd): Likewise.
12494 (fms<mode>4_odd): Likewise.
12495 (nfma<mode>4_odd): Likewise.
12496 (nfms<mode>4_odd): Likewise.
12497 (trunc<mode>df2_odd): Change the truncate with round to odd
12498 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12500 * doc/extend.texi (PowerPC built-in functions): Update documentation
12501 for existing IEEE float128-bit built-in functions. Add built-in
12502 functions that generate the IEEE 128-bit floating point round to
12505 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12507 PR rtl-optimization/77729
12508 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12509 to (X&(C1&~C2))|C2 transformations.
12511 2017-10-03 Martin Jambor <mjambor@suse.cz>
12513 PR tree-optimization/82363
12514 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12515 mismatch, mark lacc written regardless of racc.
12517 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12519 PR tree-optimization/82381
12520 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12521 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12522 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12523 fallthrough into reassoc_stmt_dominates_stmt_p.
12526 * combine.c (combine_instructions): Don't combine in unreachable
12529 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12532 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12533 function to not use the have_cpu variable. Do not set cpu_index,
12534 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12535 or the default cpu.
12536 (rs6000_valid_attribute_p): Remove duplicate initializations of
12537 old_optimize and func_optimize.
12538 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12539 (rs6000_activate_target_options): Make global.
12540 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12543 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12545 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12546 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12547 Set *poffset to 0 and *psize and *pmax_size to -1 if
12548 *poffset + *psize overflows in HOST_WIDE_INT.
12550 PR tree-optimization/82387
12551 PR tree-optimization/82388
12552 PR tree-optimization/82389
12553 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12554 instead of live_bytes non-NULL.
12556 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12559 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12562 2017-10-02 Richard Biener <rguenther@suse.de>
12564 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12565 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12566 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12568 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12570 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12571 requested precision matches the type's.
12572 * calls.c (alloc_max_size): Calculate the new candidate size as
12573 a widest_int and use wi::to_widest when comparing it with the
12574 current candidate size.
12575 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12576 zero rather than integer_zero_node.
12577 * match.pd: Check for a no-op conversion before using wi::add
12578 rather than after. Use tree_to_uhwi when summing small shift
12579 counts into an unsigned int.
12581 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12582 Alan Hayward <alan.hayward@arm.com>
12583 David Sherwood <david.sherwood@arm.com>
12586 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12587 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12588 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12589 POINTER_AND_FP_REGS.
12591 2017-10-02 Richard Biener <rguenther@suse.de>
12593 PR tree-optimization/82355
12594 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12595 a mapping for the enclosing loop but avoid generating one for
12596 the loop tree root.
12597 (copy_bb_and_scalar_dependences): Remove premature codegen
12598 error on PHIs in blocks duplicated into multiple places.
12599 * graphite-scop-detection.c
12600 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12601 in the region use it as loop and nest to analyze the DR in.
12602 (try_generate_gimple_bb): Likewise.
12603 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12604 (add_loop_constraints): For blocks in a loop not in the region
12605 create a dimension with a single iteration.
12606 * sese.h (gbb_loop_at_index): Remove assert.
12608 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12610 * omp-expand.c (adjust_context_scope): New function.
12611 (expand_parallel_call): Call adjust_context_scope.
12613 2017-10-01 Jeff Law <law@redhat.com>
12615 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12616 dom_opt_dom_walker class with direct access to private members.
12617 Add comments. Call test_for_singularity.
12618 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12619 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12620 m_dummy_cond anymore.
12621 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12623 (pass_dominator:execute): Build the dummy_cond here and pass it
12624 to the dom_opt_dom_walker ctor.
12625 (test_for_singularity): New function.
12627 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12628 Maya Rashish <coypu@sdf.org>
12630 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12631 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12632 (alpha*-*-netbsd*) Use nbsd_tm_file.
12633 (arm*-*-netbsdelf*) Likewise.
12634 (i[34567]86-*-netbsdelf*) Likewise.
12635 (x86_64-*-netbsd*) Likewise.
12636 (mips*-*-netbsd*) Likewise.
12637 (powerpc-*-netbsd*) Likewise.
12638 (sh*-*-netbsd*) Likewise.
12639 (sparc-*-netbsdelf*) Likewise.
12640 (sparc64-*-netbsd*) Likewise.
12641 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12643 (vax-*-netbsdelf*) Likewise.
12644 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12645 (UINT_FAST8_TYPE) Likewise.
12646 (INT_FAST16_TYPE) Check CHAR_FAST16.
12647 (UINT_FAST16_TYPE) Likewise.
12649 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12652 * config/i386/i386.md
12653 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12654 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12655 *divmodsi4_zext_2): New define_insn_and_split.
12656 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12657 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12658 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12659 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12660 New define_insn_and_split.
12661 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12662 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12663 operands[1] having DImode when mode is SImode.
12665 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12666 always SImode for DIV and MOD in REG_EQUAL notes.
12668 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12670 PR middle-end/82319
12671 * match.pd: Fix handling of NaNs in pattern.
12673 2017-09-29 Jeff Law <law@redhat.com>
12675 * sbitmap.c (bitmap_bit_in_range_p): New function.
12676 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12677 * tree-ssa-dse.c (live_bytes_read): New function.
12678 (dse_classify_store): Ignore reads of dead bytes.
12680 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12681 typos and whitespace errors.
12682 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12683 * config/s390/s390.c (s390_emit_prologue): Likewise.
12685 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12688 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12689 with a symbol for LRA.
12691 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12693 PR rtl-optimization/82338
12694 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12696 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12698 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12699 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12700 (qsort_chk): Declare.
12701 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12702 (qsort_chk): New function.
12704 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12706 PR tree-optimization/82337
12707 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12708 phi definition if the PHI result appears in an abnormal PHI.
12709 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12710 basis appears in an abnormal PHI.
12712 2017-09-29 Richard Biener <rguenther@suse.de>
12714 * graphite-isl-ast-to-gimple.c
12715 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12716 (binary_op_to_tree): Use it.
12717 (get_rename_from_scev): Likewise.
12718 (copy_loop_phi_nodes): Likewise.
12719 (copy_bb_and_scalar_dependences): Likewise.
12720 (translate_pending_phi_nodes): Likewise.
12722 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12725 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12726 for movabsq $(i32 << shift), r64.
12728 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12730 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12731 index when encoding %esp as %rsp to avoid 0x67 prefix.
12733 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12735 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12736 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12738 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12740 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12741 Extensions with more than 16 double VFP registers.
12742 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12743 to_clear_mask and all code related to it. Replace the remaining
12744 entry by a sbitmap and adapt code accordingly.
12746 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12748 * brig-builtins.def: Change pure attributes to const.
12750 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12752 * config.gcc (default_gnu_indirect_function): Default to yes for
12753 sparc*-*-linux* with glibc.
12755 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12757 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12758 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12759 when creating .init_array and .fini_array sections with priority
12762 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12765 * config/aarch64/aarch64.c
12766 (aarch64_builtin_support_vector_misalignment): Always return false
12767 when misalignment is unknown.
12769 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12771 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12772 this function to return false if the definition used by the swap
12773 instruction is artificial, or if the memory address from which the
12774 constant value is loaded is not represented by a base address held
12775 in a register or if the base address register is a frame or stack
12776 pointer. Additionally, return false if the base address of the
12777 loaded constant is a SYMBOL_REF but is not considered to be a
12779 (replace_swapped_load_constant): New function.
12780 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12781 loaded constant vector with a load of a swapped constant vector.
12783 2017-09-27 Carl Love <cel@us.ibm.com>
12785 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12786 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12787 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12790 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12792 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12793 first, always call autopref_rank_data otherwise.
12795 2017-09-27 Richard Biener <rguenther@suse.de>
12797 * graphite-scop-detection.c (find_scop_parameters): Move
12798 loop bound handling ...
12799 (gather_bbs::before_dom_children): ... here, avoiding the need
12800 to build scop_info->loop_nest.
12801 (record_loop_in_sese): Remove.
12802 * sese.h (sese_info_t::loop_nest): Remove.
12803 * sese.c (new_sese_info): Do not allocate loop_nest.
12804 (free_sese_info): Do not free loop_nest.
12806 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12809 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12810 lhs from calls if the lhs has addressable type.
12812 2017-09-27 Richard Biener <rguenther@suse.de>
12814 * graphite.h (scop::max_alias_set): New member.
12815 * graphite-scop-detection.c: Remove references to non-existing
12816 --param in comments.
12817 (build_alias_sets): Record the maximum alias set used for drs.
12818 (build_scops): Support zero as unlimited for
12819 --param graphite-max-arrays-per-scop.
12820 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12821 and inline into ...
12822 (build_poly_sr_1): ... here. Compute alias set based on the
12823 maximum alias set used for drs rather than
12824 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12826 2017-09-27 Richard Biener <rguenther@suse.de>
12828 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12829 --param loop-block-tile-size=0 to disable tiling.
12831 2017-09-27 Richard Biener <rguenther@suse.de>
12833 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12834 (graphite-max-nb-scop-params): Document special value zero.
12835 * domwalk.h (dom_walker::STOP): New symbolical constant.
12836 (dom_walker::dom_walker): Add optional parameter for bb to
12838 (dom_walker::~dom_walker): Declare.
12839 (dom_walker::before_dom_children): Document STOP return value.
12840 (dom_walker::m_user_bb_to_rpo): New member.
12841 (dom_walker::m_bb_to_rpo): Likewise.
12842 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12843 mapping here if not provided by the user.
12844 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12845 provided by the user.
12846 (dom_walker::STOP): Define.
12847 (dom_walker::walk): Do not compute bb to RPO mapping here.
12848 Support STOP return value from before_dom_children to stop
12850 * graphite-optimize-isl.c (optimize_isl): If the schedule
12851 is the same still generate code if -fgraphite-identity
12852 or -floop-parallelize-all are given.
12853 * graphite-scop-detection.c: Include cfganal.h.
12854 (gather_bbs::gather_bbs): Get and pass through bb to RPO
12856 (gather_bbs::before_dom_children): Return STOP for BBs
12858 (build_scops): Compute bb to RPO mapping and pass it to
12859 the domwalk. Treat --param graphite-max-nb-scop-params=0
12860 as not limiting the number of params.
12861 * graphite.c (graphite_initialize): Remove limit on the
12862 number of basic-blocks in a function.
12863 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
12864 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
12865 default value of 10.
12867 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12869 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
12870 Adjust code to eliminate needing to do the shift right 32-bits
12871 operation after XSCVDPSPN.
12873 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12875 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
12876 ((X / Y) != 0 -> X >= Y): Likewise.
12878 2017-09-26 Carl Love <cel@us.ibm.com>
12880 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
12881 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
12882 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
12883 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
12884 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
12885 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
12886 definitions and overloading.
12887 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
12888 statement for P9V_BUILTIN_XST_LEN_R.
12889 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
12890 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
12891 define_expand and define_insn for the instructions and builtins.
12892 * doc/extend.texi: Update the built-in documentation file for the new
12893 built-in functions.
12894 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
12895 define_insn for the instructions
12897 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
12900 * gcc/config/netbsd-protos.h: New file.
12901 * gcc/config/netbsd.c: New file.
12902 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
12903 * gcc/config/t-netbsd: New file.
12904 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
12905 (tmake_file) Add t-netbsd.
12906 (extra_objs) Add netbsd.o.
12908 2017-09-26 Janus Weil <janus@gcc.gnu.org>
12912 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
12914 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12916 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
12917 sign extension from a vector register to a GPR by doing a 32-bit
12918 direct move and then an EXTSW.
12919 (extendsi<mode>2 splitter): Likewise.
12920 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
12921 right or vector extract after doing XSCVDPSPN. Use
12922 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
12924 (movdi_from_sf_zero_ext): Likewise.
12925 (reload_gpr_from_vsxsf): Likewise.
12926 (p8_mfvsrd_4_disf): Delete, no longer used.
12927 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
12928 then needing to move the SFmode to a GPR to use the XSCVDPSP
12929 instruction instead of FRSP and XSCVDPSPN.
12930 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
12931 it is adjacent to the other XSCVSPDP insns.
12932 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
12933 SFmode to be in traditional Altivec registers.
12934 (vsx_xscvdpspn): Eliminate useless alternative constraint.
12935 (vsx_xscvspdpn): Likewise.
12936 (vsx_xscvspdpn_scalar): Likewise.
12938 2017-09-26 Martin Jambor <mjambor@suse.cz>
12940 * tree-sra.c (compare_access_positions): Put integral types first,
12941 stabilize sorting of integral types, remove conditions putting
12942 non-full-precision integers last.
12943 (sort_and_splice_var_accesses): Disable scalarization if a
12944 non-integert would be represented by a non-full-precision integer.
12946 2017-09-26 Joseph Myers <joseph@codesourcery.com>
12948 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
12949 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
12950 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
12951 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
12952 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
12953 conditionals inside the function instead of around it. Call
12954 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
12955 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
12957 2017-09-26 Richard Biener <rguenther@suse.de>
12959 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
12961 (scop_detection::build_scop_breadth): ... this. Removed.
12962 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
12963 (scop_detection::harmful_stmt_in_bb): Likewise.
12964 (scop_detection::graphite_can_represent_stmt): Likewise.
12965 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
12966 (scop_detection::can_represent_loop): Remove recursion, fold in ...
12967 (scop_detection::can_represent_loop_1): ... this. Removed.
12968 (scop_detection::harmful_loop_in_region): Simplify after inlining
12969 the above and remove more quadraticness.
12970 (build_scops): Adjust.
12971 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
12974 2017-09-26 Jakub Jelinek <jakub@redhat.com>
12977 * config/i386/i386.c (ix86_print_operand_address_as): Only test
12978 REGNO (base) == SP_REG if base is a REG.
12980 PR middle-end/35691
12981 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
12982 if it is different SSA_NAME.
12983 (optimize_range_tests_cmp_bitwise): New function.
12984 (optimize_range_tests): Call it.
12986 2017-09-26 Richard Biener <rguenther@suse.de>
12988 PR tree-optimization/82321
12989 * graphite.c (canonicalize_loop_closed_ssa): Properly check
12990 for the def being inside the loop.
12992 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12994 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
12996 * config/s390/s390-builtins.def: Fix constraint on op4.
12998 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13000 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13001 independent expanders.
13002 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13003 ("vec_ordered", "vec_unordered"): New expanders.
13005 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13007 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13010 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13012 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13013 vec_unpacks_lo_v16qi.
13014 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13016 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13018 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13019 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13020 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13022 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13024 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13026 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13028 ("*vec_slb<mode>"): New insn pattern.
13029 ("vec_shr_<mode>"): New expander.
13030 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13031 and force the shift count operand to V16QImode.
13032 ("vec_srb<mode>"): Set shift count mode to V16QI.
13034 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13036 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13037 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13038 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13040 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13043 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13045 2017-09-26 Richard Biener <rguenther@suse.de>
13047 PR tree-optimization/82320
13048 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13051 2017-09-25 Jeff Law <law@redhat.com>
13053 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13054 prototype for new argument.
13055 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13056 mostly extracted from rs6000_emit_allocate_stack.
13057 (rs6000_emit_probe_stack_range_stack_clash): New function.
13058 (rs6000_emit_allocate_stack): Call
13059 rs6000_emit_probe_stack_range_stack_clash as needed.
13060 (rs6000_emit_probe_stack_range): Add additional argument
13061 to call to gen_probe_stack_range{si,di}.
13062 (output_probe_stack_range): New.
13063 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13064 (output_probe_stack_range_stack_clash): New.
13065 (rs6000_emit_prologue): Emit notes into dump file as requested.
13066 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13067 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13068 Add additional operand and pass it to output_probe_stack_range.
13070 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13072 PR tree-optimization/82163
13073 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13074 (checking_verify_loop_closed_ssa): New parameter.
13075 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13076 (check_loop_closed_ssa_stmt): Delete.
13077 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13078 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13079 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13082 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13084 * brig-builtins.def: Treat HSAIL barrier builtins as
13085 setjmp/longjump style functions.
13087 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13089 * target.def (constant_alignment): New hook.
13090 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13091 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13092 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13093 * doc/tm.texi: Regenerate.
13094 * targhooks.h (default_constant_alignment): Declare.
13095 (constant_alignment_word_strings): Likewise.
13096 * targhooks.c (default_constant_alignment): New function.
13097 (constant_alignment_word_strings): Likewise.
13098 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13099 instead of CONSTANT_ALIGNMENT.
13100 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13101 (force_const_mem): Likewise.
13102 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13103 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13104 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13105 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13106 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13108 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13109 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13110 constant_alignment_word_strings.
13111 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13112 (CONSTANT_ALIGNMENT): Likewise.
13113 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13114 (arm_constant_alignment): New function.
13115 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13116 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13117 constant_alignment_word_strings.
13118 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13119 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13120 constant_alignment_word_strings.
13121 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13122 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13123 (cris_constant_alignment): New function.
13124 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13125 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13126 (epiphany_constant_alignment): New function.
13127 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13128 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13129 constant_alignment_word_strings.
13130 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13131 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13132 constant_alignment_word_strings.
13133 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13134 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13135 constant_alignment_word_strings.
13136 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13137 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13138 * config/i386/i386.c (ix86_constant_alignment): Make static.
13139 Use the same interface as the target hook.
13140 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13141 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13142 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13143 constant_alignment_word_strings.
13144 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13145 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13146 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13147 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13148 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13149 constant_alignment_word_strings.
13150 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13151 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13152 constant_alignment_word_strings.
13153 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13154 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13155 constant_alignment_word_strings.
13156 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13157 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13159 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13160 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13161 * config/mips/mips.c (mips_constant_alignment): New function.
13162 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13163 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13164 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13165 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13166 (mmix_constant_alignment): Make static. Use the same interface
13167 as the target hook.
13168 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13169 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13170 constant_alignment_word_strings.
13171 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13172 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13173 constant_alignment_word_strings.
13174 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13175 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13176 constant_alignment_word_strings.
13177 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13178 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13179 (rs6000_constant_alignment): New function.
13180 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13181 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13182 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13183 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13184 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13185 (rs6000_constant_alignment): New function.
13186 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13187 * config/s390/s390.c (s390_constant_alignment): New function.
13188 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13189 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13190 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13191 constant_alignment_word_strings.
13192 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13193 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13194 (sparc_constant_alignment): New function.
13195 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13196 * config/spu/spu.c (spu_constant_alignment): New function.
13197 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13198 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13199 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13200 constant_alignment_word_strings.
13201 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13202 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13203 constant_alignment_word_strings.
13204 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13205 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13206 constant_alignment_word_strings.
13207 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13208 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13209 (visium_constant_alignment): New function.
13210 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13211 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13212 (xtensa_constant_alignment): New function.
13213 * system.h (CONSTANT_ALIGNMENT): Poison.
13215 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13217 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13218 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13219 (rs6000_builtin_valid_without_lhs): New helper function.
13220 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13221 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13223 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13225 * target.h (vec_perm_indices): Use unsigned short rather than
13227 (auto_vec_perm_indices): Likewise.
13228 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13229 Use unsigned int rather than unsigned char.
13230 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13232 2017-09-25 Richard Biener <rguenther@suse.de>
13234 * cfgloop.h (sort_sibling_loops): Declare.
13235 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13236 (sort_sibling_loops): New function sorting the sibling loop list
13238 * graphite.c (graphite_transform_loops): Sort sibling loops.
13240 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13242 * target.def (vec_perm_const_ok): Change sel parameter to
13244 * optabs-query.c (can_vec_perm_p): Update accordingly.
13245 * doc/tm.texi: Regenerate.
13246 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13247 auto_vec_perm_indices and remove separate nelt field.
13248 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13249 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13250 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13251 (aarch64_expand_vec_perm_const): Update accordingly.
13252 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13253 to vec_perm_indices.
13254 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13255 auto_vec_perm_indices and remove separate nelt field.
13256 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13257 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13258 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13260 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13261 to vec_perm_indices.
13262 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13263 sel to vec_perm_indices.
13264 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13265 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13266 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13268 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13271 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13274 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13275 on the FUNCTION_DECL function context if it has a DIE that is a
13278 2017-09-25 Richard Biener <rguenther@suse.de>
13280 PR tree-optimization/82285
13281 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13284 2017-09-25 Tom de Vries <tom@codesourcery.com>
13288 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13291 2017-09-25 Richard Biener <rguenther@suse.de>
13293 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13294 ISL errors other than isl_error_quota happen. Dump if the
13295 schedule is the same.
13296 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13297 errors instead of aborting inside ISL.
13299 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13302 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13303 of libgcc_eh for m64.
13304 * config/i386/darwin64.h: Likewise.
13306 2017-09-25 Richard Biener <rguenther@suse.de>
13308 PR middle-end/82144
13309 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13310 attribute for incomplete types nor twice for complete ones.
13312 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13315 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13316 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13319 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13322 * config/i386/i386.opt (mprefer-avx256): Use
13323 ix86_target_flags variable.
13324 * config/i386/i386.c (ix86_target_string): Move
13325 -mprefer-avx256 to flag2_opts.
13327 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13329 PR middle-end/35691
13330 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13331 and x != -1 | y != -1 into (x & y) != -1.
13333 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13335 * config.gcc: Add new case statement to set
13336 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13337 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13338 s390x-*-linux* case statements. Added aarch64 to the list of
13339 supported architectures.
13341 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13343 PR tree-optimization/82289
13344 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13345 STMT_VINFO_RELEVANT_P.
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-vrp.c (extract_range_from_multiplicative_op_1): Assert
13352 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13353 (extract_range_from_binary_expr_1): Don't call
13354 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13356 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13357 Alan Hayward <alan.hayward@arm.com>
13358 David Sherwood <david.sherwood@arm.com>
13360 * target.def (preferred_vector_alignment): New hook.
13361 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13363 * doc/tm.texi: Regenerate.
13364 * targhooks.h (default_preferred_vector_alignment): Declare.
13365 * targhooks.c (default_preferred_vector_alignment): New function.
13366 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13368 (DR_TARGET_ALIGNMENT): New macro.
13369 (aligned_access_p): Update commentary.
13370 (vect_known_alignment_in_bytes): New function.
13371 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13373 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13374 Calculate the misalignment based on the target alignment rather than
13376 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13377 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13378 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13379 the target alignment, rather than masking the element misalignment
13380 with the number of elements in a vector. Also use the target
13381 alignment when calculating the maximum number of peels.
13382 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13383 instead of TYPE_ALIGN_UNIT.
13384 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13385 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13386 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13387 (vect_create_data_ref_ptr): Likewise.
13388 (vect_setup_realignment): Realign by ANDing with
13389 -DR_TARGET_MISALIGNMENT.
13390 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13391 the number of peels based on DR_TARGET_ALIGNMENT.
13392 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13393 with the guaranteed alignment boundary when deciding whether
13395 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13396 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13397 (ensure_base_align): Remove stmt_info parameter. Get the
13398 target base alignment from DR_TARGET_ALIGNMENT.
13399 (vectorizable_store): Update call accordingly. Interpret
13400 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13402 (vectorizable_load): Likewise.
13404 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13405 Alan Hayward <alan.hayward@arm.com>
13406 David Sherwood <david.sherwood@arm.com>
13408 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13409 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13410 (vect_enhance_data_refs_alignment): Likewise.
13412 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13414 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13415 error. Only quit immediately if parsing is complete.
13416 (BEGIN): Initialize fatal_err and parse_done.
13417 (begin fpu, end fpu): Check number of arguments.
13418 (begin arch, end arch): Likewise.
13419 (begin cpu, end cpu): Likewise.
13420 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13421 (optalias): Likewise.
13423 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13425 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13426 * config/arm/arm-isa.h: Delete. Move definitions to ...
13427 * arm-cpus.in: ... here. Use new feature and fgroup values.
13428 * config/arm/arm.c (arm_option_override): Use lower case for feature
13430 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13431 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13432 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13434 (print_isa_bits_for): New function.
13435 (gen_isa): New function.
13436 (gen_comm_data): Use print_isa_bits_for.
13437 (define feature): New keyword.
13438 (define fgroup): New keyword.
13439 * config/arm/t-arm (TM_H): Remove.
13440 (GTM_H): Add arm-isa.h.
13441 (arm-isa.h): Add rule to generate file.
13442 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13443 case for feature bit names.
13445 2017-09-22 Richard Biener <rguenther@suse.de>
13447 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13449 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13450 print of no dependency loops ...
13451 * graphite.c (graphite_transform_loops): ... here.
13452 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13454 (same_close_phi_node, remove_duplicate_close_phi,
13455 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13456 (canonicalize_loop_closed_ssa): ... here and simplify.
13457 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13458 (optimize_isl): Use dump_printf_loc to tell when we stopped
13459 optimizing because of an ISL timeout.
13461 2017-09-22 Richard Biener <rguenther@suse.de>
13463 PR tree-optimization/82291
13464 * tree-if-conv.c (predicate_mem_writes): Make sure to
13465 remove writes in blocks predicated with false.
13467 2017-09-22 Richard Biener <rguenther@suse.de>
13469 * sese.c: Include cfganal.h.
13470 (if_region_set_false_region): Remove.
13471 (create_if_region_on_edge): Likewise.
13472 (move_sese_in_condition): Re-implement without destroying
13475 2017-09-22 Richard Biener <rguenther@suse.de>
13477 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13478 Verify both BBs contain loop PHI nodes before dispatching to
13479 copy_loop_phi_args.
13480 (graphite_regenerate_ast_isl): Do not recompute dominators,
13481 do not verify three times. Restructure for clarity.
13482 * graphite-scop-detection.c (same_close_phi_node,
13483 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13484 defined_in_loop_p, canonicalize_loop_closed_ssa,
13485 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13486 checking and SSA rewrite, move to ...
13487 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13488 (graphite_initialize): Do not pass in ctx, do not reset the
13489 SCEV cache, compute only dominators.
13490 (graphite_transform_loops): Allocate ISL ctx after
13491 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13492 Maintain post-dominators only around build_scops.
13493 * sese.c (if_region_set_false_region): Make static. Free
13494 and recompute dominators.
13495 (move_sese_in_condition): Assert we don't get called with
13496 post-dominators computed.
13497 * sese.h (if_region_set_false_region): Remove.
13499 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13501 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13502 mode attribute for TARGET_AVX512VL.
13504 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13506 * config/i386/i386.opt (mprefer-avx256): New option.
13507 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13509 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13510 for TARGET_PREFER_AVX256.
13511 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13513 2017-09-21 Jeff Law <law@redhat.com>
13515 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13516 Fix dump output if the only stack space is for pushed registers.
13518 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13520 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13523 2017-09-21 Martin Sebor <msebor@redhat.com>
13526 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13527 code (in C++) or code that triggers warnings.
13529 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13531 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13533 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13535 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13536 * sched-rgn.c: Ditto.
13537 * sel-sched-ir.c: Ditto.
13539 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13541 * toplev.h (set_random_seed): Adjust return type.
13542 * toplev.c (init_local_tick): Move eager initialization of random_seed
13543 to get_random_seed. Adjust comment.
13544 (init_random_seed): Inline to get_random_seed, delete.
13545 (get_random_seed): Initialize random_seed lazily.
13546 (set_random_seed): Do not return previous value.
13547 (print_switch_value): Do not call get_random_seed.
13549 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13551 * cgraph.c (delete_function_version): New, broken out from...
13552 (cgraph_node::delete_function_version): ...here. Rename to
13553 cgraph_node::delete_function_version_by_decl. Update all uses.
13554 (cgraph_node::remove): Call delete_function_version.
13556 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13559 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13560 VAR_DECLs to which addressable non-volatile parameters are mapped
13561 and for id->retvar after the return value assignment. Clear
13562 id->retval and id->retbnd after inlining.
13564 2017-09-21 Richard Biener <rguenther@suse.de>
13566 PR tree-optimization/82276
13567 PR tree-optimization/82244
13568 * tree-vrp.c (build_assert_expr_for): Set
13569 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13571 (remove_range_assertions): Revert earlier change.
13573 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13576 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13578 2017-09-21 Richard Biener <rguenther@suse.de>
13580 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13581 Restore valid IL after code generation errors.
13582 * graphite.c (graphite_transform_loops): Diagnose code
13583 generation issues as MSG_MISSED_OPTIMIZATION and continue
13584 with processing SCOPs.
13586 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13587 Alan Hayward <alan.hayward@arm.com>
13588 David Sherwood <david.sherwood@arm.com>
13590 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13591 rather than choosing between plus_constant and gen_rtx_<CODE>.
13592 * expr.c (emit_push_insn): Likewise.
13593 (expand_expr_real_2): Likewise.
13595 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13596 Alan Hayward <alan.hayward@arm.com>
13597 David Sherwood <david.sherwood@arm.com>
13599 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13601 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13602 Alan Hayward <alan.hayward@arm.com>
13603 David Sherwood <david.sherwood@arm.com>
13605 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13606 calling tree_to_uhwi.
13608 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13609 Alan Hayward <alan.hayward@arm.com>
13610 David Sherwood <david.sherwood@arm.com>
13612 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13613 INTEGER_CST rather than a negative test for ADDR_EXPR.
13615 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13616 Alan Hayward <alan.hayward@arm.com>
13617 David Sherwood <david.sherwood@arm.com>
13619 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13620 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13622 2017-09-21 Richard Biener <rguenther@suse.de>
13624 PR tree-optimization/71351
13625 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13626 graphite_create_new_loop_guard): Remove, fold remaining parts
13628 (translate_isl_ast_node_for): ... here and simplify.
13630 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13633 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13634 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13635 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13636 alternative always use QI mode, for -Os imov (=R,R) alternative
13637 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13640 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13641 Jeff Law <law@redhat.com>
13643 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13644 (allocate_stack_space): New function, partially extracted from
13645 s390_emit_prologue.
13646 (s390_emit_prologue): Track offset to most recent stack probe.
13647 Code to allocate space moved into allocate_stack_space.
13648 Dump actions when no stack is allocated.
13649 (s390_prologue_plus_offset): New function.
13650 (s390_emit_stack_probe): Likewise.
13652 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13654 * common.opt (Wa, Wl, Wp, g, gz=): Add
13656 (gno-column-info): Remove.
13657 (gcolumn-info): Drop RejectNegative.
13658 (gno-): New prefix.
13659 (gno-record-gcc-switches): Remove.
13660 (grecord-gcc-switches): Drop RejectNegative.
13661 (gno-split-dwarf): Remove.
13662 (gsplit-dwarf): Drop RejectNegative.
13663 (gno-strict-dwarf): Remove.
13664 (gstrict-dwarf): Drop RejectNegative.
13665 * config/darwin.opt (gfull, gused): Add RejectNegative.
13666 * dwarf2out.c (gen_producer_string): Drop
13667 gno-record-gcc-switches handler.
13668 * optc-gen.awk: Add g to prefixes with negative forms.
13669 * opts-common.c (remapping_prefix_p): New.
13670 (find_opt): Check it.
13671 (generate_canonical_option): Test g prefix.
13672 (option_map): Add -gno- mapping.
13673 (add_misspelling_candidates): Check remapping_prefix_p.
13675 2017-09-20 Jeff Law <law@redhat.com>
13677 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13678 thinko in stack clash protection support.
13680 * explow.c (compute_stack_clash_protection_loop_data): Use
13681 CONST_INT_P instead of explicit test. Verify object is a
13682 CONST_INT_P before looking at INTVAL.
13683 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13684 instead of explicit test.
13686 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13689 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13690 address instead of to r1 and r11.
13692 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13694 * config.gcc: Support "knm".
13695 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13696 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13698 * config/i386/i386.c (m_KNM): Define.
13699 (processor_target_table): Add "knm".
13701 (ix86_option_override_internal): Add "knm".
13702 (ix86_issue_rate): Add PROCESSOR_KNM.
13703 (ix86_adjust_cost): Ditto.
13704 (ia32_multipass_dfa_lookahead): Ditto.
13705 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13706 (fold_builtin_cpu): Add M_INTEL_KNM.
13707 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13708 (processor_type): Add PROCESSOR_KNM.
13709 * config/i386/x86-tune.def: Add m_KNM.
13710 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13712 2017-09-20 Richard Biener <rguenther@suse.de>
13714 PR tree-optimization/80213
13715 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13716 are allowed in empty BBs as well.
13717 (canonicalize_loop_closed_ssa): Also look for other complex
13719 (scop_detection::get_sese): Include the loop-closed PHI block
13721 (scop_detection::merge_sese): Remove code adding extra blocks.
13722 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13723 (build_scops): Assert the final returned scop is invalid.
13725 2017-09-20 Richard Biener <rguenther@suse.de>
13727 PR tree-optimization/82264
13728 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13730 (vn_phi_lookup): Likewise.
13731 (vn_phi_insert): Likewise.
13733 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13735 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13736 that fits into uhwi or shwi, add DW_AT_const_value regardless
13737 of early_dwarf without going through RTL, using add_AT_unsigned
13740 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13741 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13742 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13743 (DEBUG_MACRO_SECTION): Likewise.
13744 (DEBUG_LTO_MACRO_SECTION): Likewise.
13745 (DEBUG_STR_DWO_SECTION): Likewise.
13746 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13747 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13748 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13749 (DEBUG_LTO_LINE_STR_SECTION): Define.
13750 (init_sections_and_labels): Initialize debug_line_str_section
13751 variable. Initialize debug_loc_section for -gdwarf-5 to
13752 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13754 2017-09-20 Richard Biener <rguenther@suse.de>
13756 * graphite-sese-to-poly.c (extract_affine): Properly handle
13757 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13759 2017-09-20 Richard Biener <rguenther@suse.de>
13761 PR tree-optimization/81373
13762 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13763 Force SESE live-out defs to be handled even if they are
13766 2017-09-19 Jeff Law <law@redhat.com>
13768 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13769 nothing for stack adjustments with REG_STACK_CHECK.
13770 * sched-deps.c (parse_add_or_inc): Reject insns with
13771 REG_STACK_CHECK from dependency breaking.
13772 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13773 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13774 * reg-notes.def (STACK_CHECK): New note.
13776 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13777 (ix86_expand_prologue): Dump stack clash info as needed.
13778 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13780 * function.c (dump_stack_clash_frame_info): New function.
13781 * function.h (dump_stack_clash_frame_info): Prototype.
13782 (enum stack_clash_probes): New enum.
13784 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13785 flag_stack_clash_protection.
13786 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13787 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13788 (arm_frame_pointer_required): Likewise.
13789 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13790 (ia64_expand_prologue): Likewise.
13791 * config/mips/mips.c (mips_expand_prologue): Likewise.
13792 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13793 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13794 (sparc_flat_expand_prologue): Likewise.
13795 * config/spu/spu.c (spu_expand_prologue): Likewise.
13797 * explow.c: Include "params.h".
13798 (anti_adjust_stack_and_probe_stack_clash): New function.
13799 (get_stack_check_protect): Likewise.
13800 (compute_stack_clash_protection_loop_data): Likewise.
13801 (emit_stack_clash_protection_loop_start): Likewise.
13802 (emit_stack_clash_protection_loop_end): Likewise.
13803 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13804 Use anti_adjust_stack_and_probe_stack_clash.
13805 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13806 (emit_stack_clash_protection_loop_start): Likewise.
13807 (emit_stack_clash_protection_loop_end): Likewise.
13808 * rtl.h (get_stack_check_protect): Prototype.
13809 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13810 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13811 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13813 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13815 * doc/tm.texi: Rebuilt.
13816 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13817 get_stack_check_protect.
13818 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13819 * config/arm/arm.c (arm_expand_prologue): Likewise.
13820 (arm_frame_pointer_required): Likewise.
13821 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13822 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13823 * config/mips/mips.c (mips_expand_prologue): Likewise.
13824 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13825 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13826 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13827 (sparc_flat_expand_prologue): Likewise.
13829 * common.opt (-fstack-clash-protection): New option.
13830 * flag-types.h (enum stack_check_type): Note difference between
13831 -fstack-check= and -fstack-clash-protection.
13832 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13833 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13834 * toplev.c (process_options): Issue warnings/errors for cases
13835 not handled with -fstack-clash-protection.
13836 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13837 (-fstack-check): Note additional problem with -fstack-check=generic.
13838 Note that -fstack-check is primarily for Ada and refer users
13839 to -fstack-clash-protection for stack-clash-protection.
13840 Document new params for stack clash protection.
13842 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13844 * config/i386/i386.c (ix86_split_long_move): Do not handle
13845 address used for LEA in a special way.
13847 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13849 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
13851 2017-09-19 Martin Sebor <msebor@redhat.com>
13854 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
13855 of incompatible types.
13857 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
13859 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13860 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
13861 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13862 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
13864 2017-09-19 Richard Biener <rguenther@suse.de>
13866 PR tree-optimization/82244
13867 * tree-vrp.c (remove_range_assertions): Do not propagate
13868 a constant to abnormals but replace the assert with a copy.
13870 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
13872 PR rtl-optimization/57878
13873 PR rtl-optimization/68988
13874 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
13875 avoidance test involving non_reload_pseudos. Move frequency test
13876 below the general fragmentation avoidance test.
13878 2017-09-19 Richard Biener <rguenther@suse.de>
13880 PR tree-optimization/69728
13881 * graphite-sese-to-poly.c (schedule_error): New global.
13882 (add_loop_schedule): Handle empty domain by failing the
13884 (build_original_schedule): Handle schedule_error.
13886 2017-09-19 Richard Biener <rguenther@suse.de>
13888 * graphite-scop-detection.c (scop_detection::can_represent_loop):
13889 Do not iterate to sibling loops but only to siblings of inner
13892 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
13895 * config/m68k/m68k.md (moveq feeding equality comparison): Check
13896 that the registers are different.
13898 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
13900 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
13901 to processor_model and "amdfam17h" to arch_names_table.
13902 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
13904 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13907 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
13909 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13910 Alan Hayward <alan.hayward@arm.com>
13911 David Sherwood <david.sherwood@arm.com>
13913 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
13915 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
13917 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
13918 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
13919 vect_schedule_slp_instance.
13920 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
13921 Update call to vect_slp_analyze_node_operations. Simplify return
13923 (vect_slp_analyze_bb_1): Update call accordingly.
13924 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
13925 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
13926 (vect_schedule_slp): Update call accordingly.
13928 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13929 Alan Hayward <alan.hayward@arm.com>
13930 David Sherwood <david.sherwood@arm.com>
13932 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
13933 with types that aren't in fact scalar.
13935 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13937 * tree-vect-slp.c (vect_record_max_nunits): New function,
13939 (vect_build_slp_tree_1): ...here.
13940 (vect_build_slp_tree_2): Call it for phis too.
13942 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13944 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
13945 to vect_get_vec_def_for_operand when getting the mask operand.
13947 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13948 Alan Hayward <alan.hayward@arm.com>
13949 David Sherwood <david.sherwood@arm.com>
13951 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
13954 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13955 Alan Hayward <alan.hayward@arm.com>
13956 David Sherwood <david.sherwood@arm.com>
13958 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
13959 calculation for vector booleans.
13961 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13962 Alan Hayward <alan.hayward@arm.com>
13963 David Sherwood <david.sherwood@arm.com>
13965 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
13967 (vect_transform_stmt): ...here.
13968 (vect_analyze_stmt): Use it instead of calling
13969 vectorizable_live_operation directly.
13971 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
13973 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
13974 non-SIMT targets in acc vector loops.
13976 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
13978 * configure.ac: Add arc and check if assembler supports gdwarf2.
13979 * configure: Regenerate.
13981 2017-09-18 Richard Biener <rguenther@suse.de>
13983 PR tree-optimization/82220
13984 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
13985 epilogue niters from the min_profitable_iters compute.
13987 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13990 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
13991 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
13992 (ix86_init_pic_reg): Revert 2017-09-01 changes.
13994 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
13997 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
13998 switching to a new text section.
14000 2017-09-18 Richard Biener <rguenther@suse.de>
14002 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14004 (build_alias_set): Reject aliases with no access function.
14006 2017-09-18 Richard Biener <rguenther@suse.de>
14008 PR tree-optimization/79622
14009 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14011 (build_cross_bb_scalars_use): Likewise.
14013 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14015 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14017 2017-09-18 Alan Modra <amodra@gmail.com>
14020 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14021 stack_pointer_rtx for count 0. Update comments. Break up
14022 large rtl expression.
14024 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14026 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14027 Increase to 20 bytes.
14028 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14029 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14030 or avx version of the stub.
14032 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14035 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14036 compute the minimum stack alignment. Also update preferred stack
14037 boundary for leaf functions.
14039 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14041 PR tree-optimization/82228
14042 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14045 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14047 * common/config/nds32/nds32-common.c
14048 (nds32_option_optimization_table): Refine formatting.
14049 (nds32_option_optimization_table): Use -fsched-pressure and
14050 -fomit-frame-pointer for specific optimization level.
14052 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14054 * config/nds32/nds32.c: Refine formatting and comments.
14055 * config/nds32/nds32.h: Likewise.
14056 * config/nds32/nds32.md: Likewise.
14057 * config/nds32/nds32-cost.c: Likewise.
14058 * config/nds32/nds32-isr.c: Likewise.
14059 * config/nds32/nds32-md-auxiliary.c: Likewise.
14060 * config/nds32/nds32-multiple.md: Likewise.
14061 * config/nds32/nds32-predicates.c: Likewise.
14063 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14064 Jakub Jelinek <jakub@redhat.com>
14066 Add support for -std=c++2a.
14067 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14069 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14071 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14074 * doc/extend.texi (Common Function Attributes): Add
14075 references to ARM, AArch64, and S/390 specific attributes.
14076 (Function Specific Option Pragmas): Add AArch64 and S/390
14077 to list of back ends that support the target pragma.
14079 2017-09-15 Nathan Sidwell <nathan@acm.org>
14081 * doc/standards.texi: Fix C++17 description. Update URLs for
14084 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14086 * common.opt (Wcast-align=strict): New warning option.
14087 * doc/invoke.texi: Document -Wcast-align=strict.
14089 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14091 * cgraph.h (cgraph_thunk_info): Add comments.
14092 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14093 assert for VIRTUAL_* arguments stricter.
14095 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14097 PR tree-optimization/71026
14098 * match.pd: Move RDIV patterns from fold-const.c
14099 * fold-const.c (distribute_real_division): Removed.
14100 (fold_binary_loc): Remove calls to distribute_real_divison.
14102 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14104 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14105 c++1z and gnu++1z as deprecated. Change other references to
14106 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14107 Change -Wc++1z-compat to -Wc++17-compat.
14108 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14109 * dwarf2out.c (highest_c_language): Handle C++17.
14110 (gen_compile_unit_die): Likewise.
14112 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14114 PR rtl-optimization/82192
14115 * combine.c (make_extraction): Don't look through non-paradoxical
14116 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14119 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14120 Alan Hayward <alan.hayward@arm.com>
14121 David Sherwood <david.sherwood@arm.com>
14123 * target.def (function_arg_offset): New hook.
14124 * targhooks.h (default_function_arg_offset): Declare.
14125 * targhooks.c (default_function_arg_offset): New function.
14126 * function.c (locate_and_pad_parm): Use
14127 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14128 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14129 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14130 * doc/tm.texi: Regenerate.
14131 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14132 * config/spu/spu.c (spu_function_arg_offset): New function.
14133 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14134 * system.h (FUNCTION_ARG_OFFSET): Poison.
14136 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14137 Alan Hayard <alan.hayward@arm.com>
14138 David Sherwood <david.sherwood@arm.com>
14140 * target.def (truly_noop_truncation): New hook.
14141 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14142 than TRULY_NOOP_TRUNCATION.
14143 * hooks.h (hook_bool_uint_uint_true): Declare.
14144 * hooks.c (hook_bool_uint_uint_true): New function.
14145 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14146 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14147 * doc/tm.texi: Regenerate.
14148 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14149 rather than TRULY_NOOP_TRUNCATION in comments.
14150 (simplify_comparison): Likewise.
14151 (record_truncated_value): Likewise.
14152 * expmed.c (extract_bit_field_1): Likewise.
14153 (extract_split_bit_field): Likewise.
14154 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14155 instead of TRULY_NOOP_TRUNCATION.
14156 * function.c (assign_parm_setup_block): Likewise.
14157 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14158 * rtlhooks.c: Include target.h.
14159 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14160 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14161 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14162 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14163 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14164 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14165 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14166 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14167 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14168 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14169 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14170 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14171 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14172 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14173 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14174 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14175 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14176 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14177 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14178 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14179 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14180 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14181 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14182 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14183 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14184 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14185 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14186 rather than TRULY_NOOP_TRUNCATION in comments.
14187 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14188 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14189 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14190 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14191 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14192 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14193 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14194 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14195 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14196 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14197 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14198 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14199 rather than TRULY_NOOP_TRUNCATION in comments.
14200 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14201 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14202 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14203 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14204 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14205 TRULY_NOOP_TRUNCATION condition.
14206 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14207 (TRULY_NOOP_TRUNCATION): Delete.
14208 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14209 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14210 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14211 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14212 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14213 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14214 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14215 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14216 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14217 rather than TRULY_NOOP_TRUNCATION in comments.
14218 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14219 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14220 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14221 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14222 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14223 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14225 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14228 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14229 (*cmp_ior): Likewise.
14230 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14231 (*ior_scc_scc_cmp): Likewise.
14232 (*and_scc_scc): Likewise.
14233 (*and_scc_scc_cmp): Likewise.
14235 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14236 Alan Hayard <alan.hayward@arm.com>
14237 David Sherwood <david.sherwood@arm.com>
14239 * target.def (can_change_mode_class): New hook.
14240 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14241 (hard_regno_nregs): Likewise.
14242 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14243 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14244 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14245 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14246 (LOAD_EXTEND_OP): Update accordingly.
14247 * doc/tm.texi: Regenerate.
14248 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14249 CANNOT_CHANGE_MODE_CLASS.
14250 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14251 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14252 * combine.c (simplify_set): Update accordingly.
14253 * emit-rtl.c (validate_subreg): Likewise.
14254 * recog.c (general_operand): Likewise.
14255 * regcprop.c (mode_change_ok): Likewise.
14256 * reload1.c (choose_reload_regs): Likewise.
14257 (inherit_piecemeal_p): Likewise.
14258 * rtlanal.c (simplify_subreg_regno): Likewise.
14259 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14260 instead of CANNOT_CHANGE_MODE_CLASS.
14261 (reload_cse_simplify_operands): Likewise.
14262 * reload.c (push_reload): Use targetm.can_change_mode_class
14263 instead of CANNOT_CHANGE_MODE_CLASS.
14264 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14265 REG_CANNOT_CHANGE_MODE_P.
14266 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14267 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14268 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14269 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14270 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14271 (arm_can_change_mode_class): New function.
14272 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14273 than CANNOT_CHANGE_MODE_CLASS in comments.
14274 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14275 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14276 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14277 (ix86_can_change_mode_class): ...this new function, inverting the
14278 sense of the return value.
14279 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14280 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14281 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14282 (ia64_can_change_mode_class): New function.
14283 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14284 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14285 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14286 (m32c_can_change_mode_class): ...this new function, inverting the
14287 sense of the return value.
14288 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14289 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14290 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14291 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14292 (mips_can_change_mode_class): ...this new function, inverting the
14293 sense of the return value.
14294 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14295 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14296 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14297 (msp430_can_change_mode_class): New function.
14298 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14299 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14300 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14301 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14302 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14303 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14304 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14305 (pa_cannot_change_mode_class): Replace with...
14306 (pa_can_change_mode_class): ...this new function, inverting the
14307 sense of the return value.
14308 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14309 than CANNOT_CHANGE_MODE_CLASS in comments.
14310 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14311 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14312 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14313 (pdp11_cannot_change_mode_class): Replace with...
14314 (pdp11_can_change_mode_class): ...this new function, inverting the
14315 sense of the return value.
14316 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14317 * config/powerpcspe/powerpcspe-protos.h
14318 (rs6000_cannot_change_mode_class_ptr): Delete.
14319 * config/powerpcspe/powerpcspe.c
14320 (rs6000_cannot_change_mode_class_ptr): Delete.
14321 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14322 (rs6000_option_override_internal): Assign to
14323 targetm.can_change_mode_class instead of
14324 rs6000_cannot_change_mode_class_ptr.
14325 (rs6000_cannot_change_mode_class): Replace with...
14326 (rs6000_can_change_mode_class): ...this new function, inverting the
14327 sense of the return value.
14328 (rs6000_debug_cannot_change_mode_class): Replace with...
14329 (rs6000_debug_can_change_mode_class): ...this new function.
14330 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14331 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14332 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14333 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14334 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14336 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14337 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14338 (rs6000_option_override_internal): Assign to
14339 targetm.can_change_mode_class instead of
14340 rs6000_cannot_change_mode_class_ptr.
14341 (rs6000_cannot_change_mode_class): Replace with...
14342 (rs6000_can_change_mode_class): ...this new function, inverting the
14343 sense of the return value.
14344 (rs6000_debug_cannot_change_mode_class): Replace with...
14345 (rs6000_debug_can_change_mode_class): ...this new function.
14346 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14347 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14348 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14349 (s390_can_change_mode_class): ...this new function, inverting the
14350 sense of the return value.
14351 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14352 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14353 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14354 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14355 (sh_cannot_change_mode_class): Replace with...
14356 (sh_can_change_mode_class): ...this new function, inverting the
14357 sense of the return value.
14358 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14359 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14360 (sparc_can_change_mode_class): New function.
14361 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14362 * config/spu/spu.c (spu_can_change_mode_class): New function.
14363 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14364 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14365 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14366 (visium_can_change_mode_class): New function.
14367 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14369 2017-09-15 Richard Biener <rguenther@suse.de>
14371 PR tree-optimization/82217
14372 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14373 but not undefined case.
14375 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14378 * postreload.c (reload_cse_simplify_operands): Skip
14379 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14381 2017-09-15 Richard Biener <rguenther@suse.de>
14383 PR tree-optimization/68823
14384 * graphite-scop-detection.c (build_alias_set): If we have a
14385 possible dependence check whether we can handle them by just
14386 looking at the DRs DR_ACCESS_FNs.
14387 (build_scops): If build_alias_set fails, fail the SCOP.
14389 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14391 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14392 to support float128 built-in functions that require the ISA 3.0
14394 (BU_FLOAT128_3_HW): Likewise.
14395 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14396 built-in functions.
14397 (FMAF128): Likewise.
14399 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14400 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14401 floating point instructions.
14402 (rs6000_invalid_builtin): Likewise.
14403 (rs6000_builtin_mask_names): Likewise.
14404 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14405 (RS6000_BTM_FLOAT128_HW): Likewise.
14406 (RS6000_BTM_COMMON): Likewise.
14407 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14409 * doc/extend.texi (RS/6000 built-in functions): Document the
14410 IEEE 128-bit floating point square root and fused multiply-add
14411 built-in functions.
14413 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14415 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14416 reg (r2) isn't in the set of registers defined in the prologue.
14418 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14419 Alan Hayward <alan.hayward@arm.com>
14420 David Sherwood <david.sherwood@arm.com>
14422 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14423 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14424 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14425 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14426 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14428 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14429 max_vectorization_factor.
14430 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14432 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14433 Alan Hayward <alan.hayward@arm.com>
14434 David Sherwood <david.sherwood@arm.com>
14436 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14437 (vect_worthwhile_without_simd_p): Declare.
14438 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14439 (vectorizable_reduction): Use it.
14440 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14441 (vectorizable_operation): Likewise.
14443 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14444 Alan Hayward <alan.hayward@arm.com>
14445 David Sherwood <david.sherwood@arm.com>
14447 * tree-vectorizer.h (vect_get_num_copies): New function.
14448 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14449 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14450 (vectorizable_induction): Likewise.
14451 (vectorizable_live_operation): Likewise.
14452 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14453 (vectorizable_bswap): Likewise.
14454 (vectorizable_call): Likewise.
14455 (vectorizable_conversion): Likewise.
14456 (vectorizable_assignment): Likewise.
14457 (vectorizable_shift): Likewise.
14458 (vectorizable_operation): Likewise.
14459 (vectorizable_store): Likewise.
14460 (vectorizable_load): Likewise.
14461 (vectorizable_condition): Likewise.
14462 (vectorizable_comparison): Likewise.
14463 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14465 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14466 Alan Hayward <alan.hayward@arm.com>
14467 David Sherwood <david.sherwood@arm.com>
14469 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14470 of vect_init_vector.
14472 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14473 Alan Hayward <alan.hayward@arm.com>
14474 David Sherwood <david.sherwood@arm.com>
14476 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14477 an inline wrapper that provides a location.
14478 (gimple_build_vector): Likewise.
14479 * gimple-fold.c (gimple_build_vector_from_val): New function.
14480 (gimple_build_vector): Likewise.
14481 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14482 functions to build the initial value. Always return a gimple value.
14483 (get_initial_defs_for_reduction): Likewise. Only compute
14485 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14486 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14487 (vectorizable_induction): Use gimple_build_vector rather than
14490 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14491 Alan Hayward <alan.hayward@arm.com>
14492 David Sherwood <david.sherwood@arm.com>
14494 * target.h (vec_perm_indices): New typedef.
14495 (auto_vec_perm_indices): Likewise.
14496 * optabs-query.h: Include target.h
14497 (can_vec_perm_p): Take a vec_perm_indices *.
14498 * optabs-query.c (can_vec_perm_p): Likewise.
14499 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14500 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14501 * tree-vect-generic.c (lower_vec_perm): Likewise.
14502 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14503 (vect_grouped_load_supported): Likewise.
14504 (vect_shift_permute_load_chain): Likewise.
14505 (vect_permute_store_chain): Use auto_vec_perm_indices.
14506 (vect_permute_load_chain): Likewise.
14507 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14508 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14509 Update uses of can_vec_perm_p.
14510 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14511 mode with a number of elements. Take a vec_perm_indices *.
14512 (vect_create_epilog_for_reduction): Update accordingly.
14513 Use auto_vec_perm_indices.
14514 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14515 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14516 (vect_transform_slp_perm_load): Likewise.
14517 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14518 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14519 (vect_gen_perm_mask_checked): Likewise.
14520 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14521 (vect_gen_perm_mask_checked): Likewise.
14522 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14523 (vectorizable_store): Likewise.
14524 (vectorizable_load): Likewise.
14525 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14526 (vectorizable_bswap): Likewise.
14528 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14529 Alan Hayward <alan.hayward@arm.com>
14530 David Sherwood <david.sherwood@arm.com>
14532 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14533 * tree.c (build_vector): Likewise.
14534 (build_vector_from_ctor): Update accordingly.
14535 (build_vector_from_val): Likewise.
14536 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14537 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14538 * tree-vect-generic.c (add_rshift): Likewise.
14539 (expand_vector_divmod): Likewise.
14540 (optimize_vector_constructor): Likewise.
14541 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14542 (vect_transform_slp_perm_load): Likewise.
14543 (vect_schedule_slp_instance): Likewise.
14544 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14545 (vectorizable_call): Likewise.
14546 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14547 * expmed.c (make_tree): Likewise.
14548 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14549 a vector passed to build_vector.
14550 (fold_convert_const): Likewise.
14551 (exact_inverse): Likewise.
14552 (fold_ternary_loc): Likewise.
14553 (fold_relational_const): Likewise.
14554 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14555 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14556 (const_unop): Likewise. Store the reduction accumulator in a
14557 variable rather than an array.
14558 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14559 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14560 the new vector, rather than constructing it after the input arrays.
14561 (native_interpret_vector): Use auto_vec<tree> when building
14562 a vector passed to build_vector. Add elements in order.
14563 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14564 auto_vec<tree> when building a vector passed to build_vector.
14565 (vect_create_epilog_for_reduction): Likewise.
14566 (vectorizable_induction): Likewise.
14567 (get_initial_def_for_reduction): Likewise. Fix indentation of
14569 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14571 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14572 passed to build_vector.
14574 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14575 Alan Hayward <alan.hayward@arm.com>
14576 David Sherwood <david.sherwood@arm.com>
14578 * tree-core.h (tree_base::u): Add an "nelts" field.
14579 (tree_vector): Use VECTOR_CST_NELTS as the length.
14580 * tree.c (tree_size): Likewise.
14581 (make_vector): Initialize VECTOR_CST_NELTS.
14582 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14583 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14584 TYPE_VECTOR_SUBPARTS.
14585 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14586 as the number of units, setting it from VECTOR_CST_NELTS.
14587 (const_vector_from_tree): Likewise.
14588 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14589 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14590 (fold_negate_expr_1): Likewise.
14591 (fold_convert_const): Likewise.
14592 (const_binop): Likewise. Differentiate the number of output and
14594 (const_unop): Likewise.
14595 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14596 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14597 in cases that did the opposite.
14599 2017-09-14 Richard Biener <rguenther@suse.de>
14601 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14604 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14606 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14608 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14611 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14612 if and where to split a bb, except for splitting before debug insn
14613 sequences followed by non-label real insn. Delete debug insns
14614 in between basic blocks.
14616 * combine.c (make_compound_operation_int): Formatting fixes.
14618 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14619 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14620 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14621 * config/sol2.h (LINK_EH_SPEC): Likewise.
14622 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14623 * config/s390/linux.h (LINK_SPEC): Likewise.
14624 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14625 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14626 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14627 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14628 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14629 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14630 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14631 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14633 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14635 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14637 (ENDFILE_LINUX_SPEC): Likewise.
14638 (LINK_EH_SPEC): Likewise.
14639 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14640 (LINK_OS_LINUX_SPEC32): Likewise.
14641 (LINK_OS_LINUX_SPEC64): Likewise.
14642 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14643 (LINK_OS_LINUX_SPEC): Likewise.
14645 2017-09-13 Martin Liska <mliska@suse.cz>
14647 PR middle-end/82154
14648 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14649 CASE_HIGH is NULL_TREE.
14651 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14652 Alan Hayward <alan.hayward@arm.com>
14653 David Sherwood <david.sherwood@arm.com>
14655 * target.def (secondary_memory_needed): New hook.
14656 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14657 instead of SECONDARY_MEMORY_NEEDED.
14658 (secondary_memory_needed_mode): Likewise.
14659 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14660 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14661 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14662 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14663 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14664 * doc/tm.texi: Regenerate.
14665 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14666 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14667 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14668 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14669 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14670 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14671 mode argument first and change the reg_class arguments to reg_class_t.
14672 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14673 Make static. Update the call to inline_secondary_memory_needed.
14674 (ix86_register_move_cost): Update the call to
14675 inline_secondary_memory_needed.
14676 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14677 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14679 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14680 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14682 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14683 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14684 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14685 and match hook interface. Add comment from mips.h.
14686 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14687 * config/mmix/mmix.md (truncdfsf2): Refer to
14688 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14690 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14691 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14692 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14693 (pa_secondary_memory_needed): New function.
14694 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14695 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14696 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14697 (pdp11_secondary_memory_needed): Make static and match hook interface.
14698 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14699 * config/powerpcspe/powerpcspe-protos.h
14700 (rs6000_secondary_memory_needed_ptr): Delete.
14701 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14703 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14704 (rs6000_option_override_internal): Assign to
14705 targetm.secondary_memory_needed rather than
14706 rs6000_secondary_memory_needed_ptr.
14707 (rs6000_secondary_memory_needed): Match hook interface.
14708 (rs6000_debug_secondary_memory_needed): Likewise.
14709 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14710 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14711 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14712 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14713 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14714 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14716 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14717 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14718 (rs6000_option_override_internal): Assign to
14719 targetm.secondary_memory_needed rather than
14720 rs6000_secondary_memory_needed_ptr.
14721 (rs6000_secondary_memory_needed): Match hook interface.
14722 (rs6000_debug_secondary_memory_needed): Likewise.
14723 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14724 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14725 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14726 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14727 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14728 (sparc_secondary_memory_needed): New function.
14729 * lra-constraints.c (check_and_process_move): Refer to
14730 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14732 (curr_insn_transform): Likewise.
14733 (process_alt_operands): Use targetm.secondary_memory_needed
14734 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14735 (check_secondary_memory_needed_p): Likewise.
14736 (choose_split_class): Likewise.
14737 * reload.c: Unconditionally include code that was previously
14738 conditional on SECONDARY_MEMORY_NEEDED.
14739 (push_secondary_reload): Use targetm.secondary_memory_needed
14740 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14741 (push_reload): Likewise.
14742 * reload1.c: Unconditionally include code that was previously
14743 conditional on SECONDARY_MEMORY_NEEDED.
14744 (choose_reload_regs): Use targetm.secondary_memory_needed
14745 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14746 (gen_reload): Likewise.
14747 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14749 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14750 Alan Hayward <alan.hayward@arm.com>
14751 David Sherwood <david.sherwood@arm.com>
14753 * target.def (secondary_memory_needed_mode): New hook:
14754 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14755 * targhooks.h (default_secondary_memory_needed_mode): New function.
14756 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14757 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14758 * doc/tm.texi: Regenerate.
14759 * lra-constraints.c (check_and_process_move): Use
14760 targetm.secondary_memory_needed_mode instead of
14761 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14762 (curr_insn_transform): Likewise.
14763 * reload.c (get_secondary_mem): Likewise.
14764 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14765 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14767 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14768 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14769 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14770 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14771 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14773 * config/powerpcspe/powerpcspe-protos.h
14774 (rs6000_secondary_memory_needed_mode): Delete.
14775 * config/powerpcspe/powerpcspe.c
14776 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14777 (rs6000_secondary_memory_needed_mode): Make static.
14778 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14779 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14781 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14783 (rs6000_secondary_memory_needed_mode): Make static.
14784 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14785 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14786 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14787 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14788 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14790 (sparc_secondary_memory_needed_mode): New function.
14791 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14793 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14795 * config/aarch64/constraints.md (Umq): New constraint.
14796 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14798 (mov<mode>): Update condition.
14800 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14802 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14803 when bitposition is the same.
14805 2017-09-13 Richard Biener <rguenther@suse.de>
14807 * dwarf2out.c (output_die_symbol): Remove.
14808 (output_die): Do not output a DIEs symbol.
14810 2017-09-13 Richard Biener <rguenther@suse.de>
14812 PR middle-end/82128
14813 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14814 default-def to avoid breaking iterator update with the weird
14815 interaction with cgraph_update_edges_for_call_stmt_node.
14817 2017-09-13 Richard Biener <rguenther@suse.de>
14819 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14820 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14821 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14822 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14824 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14826 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14827 Disable pc relative literal load irrespective of
14828 TARGET_FIX_ERR_A53_84341 for default.
14830 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14832 * config/sparc/sparc.c (output_return): Output the source location of
14833 the insn in the delay slot, if any.
14834 (output_sibcall): Likewise.
14836 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14839 * common.opt (-static-pie): New alias.
14840 (shared): Negate static-pie.
14841 (-no-pie): Update help text.
14843 (static-pie): New option.
14844 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14845 -static-pie support.
14846 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14847 (LINK_EH_SPEC): Likewise.
14848 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14849 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14850 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14851 * gcc.c (LINK_COMMAND_SPEC): Likewise.
14852 (init_gcc_specs): Likewise.
14853 (init_spec): Likewise.
14854 (display_help): Update help message for -pie.
14855 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
14858 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
14860 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
14861 (movdi_aarch64): Likewise.
14862 (movti_aarch64): Likewise.
14864 2017-09-12 Simon Wright <simon@pushface.org>
14867 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
14868 calculation of the minor version, always output as 0.
14870 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14873 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14874 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
14875 on it early, rather than manual conversion late. For
14876 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
14877 instead of performing manual conversion.
14879 2017-09-12 Carl Love <cel@us.ibm.com>
14881 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
14882 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
14884 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14885 VMULOSW): Add definitions.
14886 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14887 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14888 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14889 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14890 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14892 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14894 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
14896 (movti_aarch64): Likewise.
14897 (movdf_aarch64): Likewise.
14898 (movtf_aarch64): Likewise.
14899 (load_pairdi): Likewise.
14900 (store_pairdi): Likewise.
14901 (load_pairdf): Likewise.
14902 (store_pairdf): Likewise.
14903 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
14904 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
14905 (ldr_got_small_<mode>): Likewise.
14906 (ldr_got_small_28k_<mode>): Likewise.
14907 (ldr_got_tiny): Likewise.
14908 * config/aarch64/iterators.md (ldst_sz): New.
14909 (ldpstp_sz): Likewise.
14910 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
14912 (thunderx_load): Split load_8 to load_16.
14913 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
14915 (thunderx2t99_storepair_basic): Split store_8 to store_16.
14916 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
14917 (xgene1_store_pair): Split store_8 to store_16.
14918 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
14919 (falkor_st_0_st_sd): Split store_8 to store_16.
14921 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14923 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
14924 and store1/2/3/4 to store_4/8/12/16.
14925 * config/aarch64/aarch64.md: Update for rename.
14926 * config/arm/arm.md: Likewise.: Likewise.
14927 * config/arm/arm.c: Likewise.
14928 * config/arm/thumb1.md: Likewise.
14929 * config/arm/thumb2.md: Likewise.
14930 * config/arm/vfp.md: Likewise.
14931 * config/arm/arm-generic.md: Likewise.
14932 * config/arm/arm1020e.md: Likewise.
14933 * config/arm/arm1026ejs.md: Likewise.
14934 * config/arm/arm1136jfs.md: Likewise.
14935 * config/arm/arm926ejs.md: Likewise.
14936 * config/arm/cortex-a15.md: Likewise.
14937 * config/arm/cortex-a17.md: Likewise.
14938 * config/arm/cortex-a5.md: Likewise.
14939 * config/arm/cortex-a53.md: Likewise.
14940 * config/arm/cortex-a57.md: Likewise.
14941 * config/arm/cortex-a7.md: Likewise.
14942 * config/arm/cortex-a8.md: Likewise.
14943 * config/arm/cortex-a9.md: Likewise.
14944 * config/arm/cortex-m4.md: Likewise.
14945 * config/arm/cortex-m7.md: Likewise.
14946 * config/arm/cortex-r4.md: Likewise.
14947 * config/arm/exynos-m1.md: Likewise.
14948 * config/arm/fa526.md: Likewise.
14949 * config/arm/fa606te.md: Likewise.
14950 * config/arm/fa626te.md: Likewise.
14951 * config/arm/fa726te.md: Likewise.
14952 * config/arm/fmp626.md: Likewise.
14953 * config/arm/iwmmxt.md: Likewise.
14954 * config/arm/ldmstm.md: Likewise.
14955 * config/arm/marvell-pj4.md: Likewise.
14956 * config/arm/xgene1.md: Likewise.
14957 * config/aarch64/thunderx.md: Likewise.
14958 * config/aarch64/thunderx2t99.md: Likewise.
14959 * config/aarch64/falkor.md: Likewise.
14961 2017-09-12 Martin Liska <mliska@suse.cz>
14963 * attribs.c (private_lookup_attribute): New function.
14964 * attribs.h (private_lookup_attribute): Declared here.
14965 (lookup_attribute): Called from this place.
14967 2017-09-12 Richard Biener <rguenther@suse.de>
14969 PR tree-optimization/82157
14970 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
14971 stmts with side-effects.
14973 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14974 Alan Hayward <alan.hayward@arm.com>
14975 David Sherwood <david.sherwood@arm.com>
14977 * target.def (hard_regno_nregs): New hook.
14978 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
14979 * targhooks.h (default_hard_regno_nregs): Declare.
14980 * targhooks.c (default_hard_regno_nregs): New function.
14981 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
14982 (TARGET_HARD_REGNO_NREGS): ...this hook.
14983 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
14984 (CLASS_MAX_NREGS): Likewise.
14985 * doc/tm.texi: Regenerate.
14986 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
14987 instead of HARD_REGNO_NREGS.
14988 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
14989 HARD_REGNO_NREGS in the comment.
14990 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
14991 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
14992 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
14993 Return an unsigned int.
14994 (TARGET_HARD_REGNO_NREGS): Redefine.
14995 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
14996 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
14997 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14998 (arc_hard_regno_nregs): New function.
14999 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15000 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15001 (arm_hard_regno_nregs): New function.
15002 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15003 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15004 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15005 (TARGET_HARD_REGNO_NREGS): Redefine.
15006 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15007 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15008 (HARD_REGNO_NREGS): Delete.
15009 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15010 (cr16_hard_regno_nregs): New function.
15011 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15012 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15013 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15014 (cris_hard_regno_nregs): New function.
15015 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15016 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15017 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15018 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15019 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15020 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15021 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15022 (frv_hard_regno_nregs): Make static. Take and return an
15024 (frv_class_max_nregs): Remove outdated copy of documentation.
15025 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15026 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15027 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15028 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15029 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15030 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15031 (TARGET_HARD_REGNO_NREGS): Redefine.
15032 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15033 (CLASS_MAX_NREGS): Update comment.
15034 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15035 (ia64_hard_regno_nregs): New function.
15036 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15037 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15038 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15039 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15040 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15042 (m32c_hard_regno_nregs): Likewise. Make static.
15043 (TARGET_HARD_REGNO_NREGS): Redefine.
15044 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15045 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15046 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15047 (m68k_hard_regno_nregs): New function.
15048 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15049 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15050 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15051 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15052 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15053 Take and return an unsigned int.
15054 (TARGET_HARD_REGNO_NREGS): Redefine.
15055 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15056 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15057 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15058 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15059 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15060 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15061 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15062 (msp430_hard_regno_nregs): Make static. Take and return an
15064 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15065 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15066 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15067 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15068 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15069 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15070 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15071 (TARGET_HARD_REGNO_NREGS): Redefine.
15072 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15073 (PA_HARD_REGNO_NREGS): ...this.
15074 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15075 (PA_HARD_REGNO_NREGS): ...this.
15076 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15077 (pa_hard_regno_nregs): New function.
15078 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15079 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15080 (pdp11_hard_regno_nregs): New function.
15081 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15082 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15083 (rs6000_hard_regno_nregs_hook): New function.
15084 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15085 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15086 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15087 Take and return an unsigned int. Move earlier in file.
15088 (TARGET_HARD_REGNO_NREGS): Redefine.
15089 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15090 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15091 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15092 (rl78_hard_regno_nregs): Make static. Take and return an
15094 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15095 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15096 (rs6000_hard_regno_nregs_hook): New function.
15097 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15098 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15099 (TARGET_HARD_REGNO_NREGS): Redefine.
15100 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15101 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15102 instead of HARD_REGNO_NREGS.
15103 (s390_hard_regno_nregs): New function.
15104 (s390_hard_regno_mode_ok): Add comment from s390.h.
15105 (TARGET_HARD_REGNO_NREGS): Redefine.
15106 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15107 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15108 (sh_hard_regno_nregs): New function.
15109 (sh_pass_in_reg_p): Use it.
15110 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15111 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15112 (sparc_hard_regno_nregs): New function.
15113 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15114 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15115 (spu_function_arg_advance): Use it, supplying a valid register number.
15116 (TARGET_HARD_REGNO_NREGS): Redefine.
15117 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15118 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15119 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15120 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15121 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15122 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15123 (CLASS_MAX_NREGS): Remove copy of old documentation.
15124 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15125 (visium_hard_regno_nregs): New function.
15126 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15127 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15128 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15129 xtensa_hard_regno_nregs): New function.
15130 * system.h (HARD_REGNO_NREGS): Poison.
15132 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15134 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15135 hard_regno_nregs instead of HARD_REGNO_NREGS.
15136 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15137 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15138 (c6x_expand_epilogue): Likewise.
15139 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15140 (frv_read_iacc_argument): Likewise.
15141 * config/sh/sh.c: Include regs.h.
15142 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15143 (regs_used): Likewise.
15144 (output_stack_adjust): Likewise.
15145 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15146 * expmed.c: Include regs.h.
15147 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15148 * ree.c: Include regs.h.
15149 (combine_reaching_defs): Use hard_regno_nregs instead of
15151 (add_removable_extension): Likewise.
15153 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15155 * regs.h (hard_regno_nregs): Turn into a function.
15156 (end_hard_regno): Update accordingly.
15157 * caller-save.c (setup_save_areas): Likewise.
15158 (save_call_clobbered_regs): Likewise.
15159 (replace_reg_with_saved_mem): Likewise.
15160 (insert_restore): Likewise.
15161 (insert_save): Likewise.
15162 * combine.c (can_change_dest_mode): Likewise.
15163 (move_deaths): Likewise.
15164 (distribute_notes): Likewise.
15165 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15166 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15167 (rs6000_split_multireg_move): Likewise.
15168 (rs6000_register_move_cost): Likewise.
15169 (rs6000_memory_move_cost): Likewise.
15170 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15171 (rs6000_split_multireg_move): Likewise.
15172 (rs6000_register_move_cost): Likewise.
15173 (rs6000_memory_move_cost): Likewise.
15174 * cselib.c (cselib_reset_table): Likewise.
15175 (cselib_lookup_1): Likewise.
15176 * emit-rtl.c (set_mode_and_regno): Likewise.
15177 * function.c (aggregate_value_p): Likewise.
15178 * ira-color.c (setup_profitable_hard_regs): Likewise.
15179 (check_hard_reg_p): Likewise.
15180 (calculate_saved_nregs): Likewise.
15181 (assign_hard_reg): Likewise.
15182 (improve_allocation): Likewise.
15183 (calculate_spill_cost): Likewise.
15184 * ira-emit.c (modify_move_list): Likewise.
15185 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15186 (ira_hard_reg_in_set_p): Likewise.
15187 * ira.c (setup_reg_mode_hard_regset): Likewise.
15188 (clarify_prohibited_class_mode_regs): Likewise.
15189 (check_allocation): Likewise.
15190 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15191 (lra_setup_reg_renumber): Likewise.
15192 (setup_try_hard_regno_pseudos): Likewise.
15193 (spill_for): Likewise.
15194 (assign_hard_regno): Likewise.
15195 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15196 * lra-constraints.c (in_class_p): Likewise.
15197 (lra_constraint_offset): Likewise.
15198 (simplify_operand_subreg): Likewise.
15199 (lra_constraints): Likewise.
15200 (split_reg): Likewise.
15201 (split_if_necessary): Likewise.
15202 (invariant_p): Likewise.
15203 (inherit_in_ebb): Likewise.
15204 * lra-lives.c (process_bb_lives): Likewise.
15205 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15206 (get_hard_regs): Likewise.
15207 (do_remat): Likewise.
15208 * lra-spills.c (assign_spill_hard_regs): Likewise.
15209 * mode-switching.c (create_pre_exit): Likewise.
15210 * postreload.c (reload_combine_recognize_pattern): Likewise.
15211 * recog.c (peep2_find_free_register): Likewise.
15212 * regcprop.c (kill_value_regno): Likewise.
15213 (set_value_regno): Likewise.
15214 (copy_value): Likewise.
15215 (maybe_mode_change): Likewise.
15216 (find_oldest_value_reg): Likewise.
15217 (copyprop_hardreg_forward_1): Likewise.
15218 * regrename.c (check_new_reg_p): Likewise.
15219 (regrename_do_replace): Likewise.
15220 * reload.c (push_reload): Likewise.
15221 (combine_reloads): Likewise.
15222 (find_dummy_reload): Likewise.
15223 (operands_match_p): Likewise.
15224 (find_reloads): Likewise.
15225 (find_equiv_reg): Likewise.
15226 (reload_adjust_reg_for_mode): Likewise.
15227 * reload1.c (count_pseudo): Likewise.
15228 (count_spilled_pseudo): Likewise.
15229 (find_reg): Likewise.
15230 (clear_reload_reg_in_use): Likewise.
15231 (free_for_value_p): Likewise.
15232 (allocate_reload_reg): Likewise.
15233 (choose_reload_regs): Likewise.
15234 (reload_adjust_reg_for_temp): Likewise.
15235 (emit_reload_insns): Likewise.
15236 (delete_output_reload): Likewise.
15237 * rtlanal.c (subreg_get_info): Likewise.
15238 * sched-deps.c (sched_analyze_reg): Likewise.
15239 * sel-sched.c (init_regs_for_mode): Likewise.
15240 (mark_unavailable_hard_regs): Likewise.
15241 (choose_best_reg_1): Likewise.
15242 (verify_target_availability): Likewise.
15243 * valtrack.c (dead_debug_insert_temp): Likewise.
15244 * var-tracking.c (track_loc_p): Likewise.
15245 (emit_note_insn_var_location): Likewise.
15246 * varasm.c (make_decl_rtl): Likewise.
15247 * reginfo.c (choose_hard_reg_mode): Likewise.
15248 (init_reg_modes_target): Refer directly to
15249 this_target_regs->x_hard_regno_nregs.
15251 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15253 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15254 instead of hard_regno_nregs.
15256 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15258 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15259 end_hard_regno instead of hard_regno_nregs.
15260 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15261 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15262 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15263 * ira-color.c (improve_allocation): Likewise.
15264 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15265 * lra-lives.c (mark_regno_live): Likewise.
15266 (mark_regno_dead): Likewise.
15267 * lra-remat.c (operand_to_remat): Likewise.
15268 * lra.c (collect_non_operand_hard_regs): Likewise.
15269 * postreload.c (reload_combine_note_store): Likewise.
15270 (move2add_valid_value_p): Likewise.
15271 * reload.c (regno_clobbered_p): Likewise.
15273 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15275 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15277 * config/v850/v850.c (v850_reorg): Likewise.
15278 * reload.c (refers_to_regno_for_reload_p): Likewise.
15279 (find_equiv_reg): Likewise.
15280 * reload1.c (reload_reg_reaches_end_p): Likewise.
15282 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15284 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15286 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15287 * config/arm/arm.c (output_move_neon): Likewise.
15288 (arm_attr_length_move_neon): Likewise.
15289 (neon_split_vcombine): Likewise.
15290 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15291 (c6x_mark_reg_written): Likewise.
15292 (c6x_dwarf_register_span): Likewise.
15293 * config/i386/i386.c (ix86_save_reg): Likewise.
15294 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15295 (rws_access_reg): Likewise.
15296 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15297 * mode-switching.c (create_pre_exit): Likewise.
15298 * ree.c (combine_reaching_defs): Likewise.
15299 (add_removable_extension): Likewise.
15300 * regcprop.c (find_oldest_value_reg): Likewise.
15301 (copyprop_hardreg_forward_1): Likewise.
15302 * reload.c (reload_inner_reg_of_subreg): Likewise.
15303 (push_reload): Likewise.
15304 (combine_reloads): Likewise.
15305 (find_dummy_reload): Likewise.
15306 (reload_adjust_reg_for_mode): Likewise.
15307 * reload1.c (find_reload_regs): Likewise.
15308 (forget_old_reloads_1): Likewise.
15309 (reload_reg_free_for_value_p): Likewise.
15310 (reload_adjust_reg_for_temp): Likewise.
15311 (emit_reload_insns): Likewise.
15312 (delete_output_reload): Likewise.
15313 * sel-sched.c (choose_best_reg_1): Likewise.
15314 (choose_best_pseudo_reg): Likewise.
15316 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15317 Alan Hayward <alan.hayward@arm.com>
15318 David Sherwood <david.sherwood@arm.com>
15320 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15321 * target.def (slow_unaligned_access): New hook.
15322 * targhooks.h (default_slow_unaligned_access): Declare.
15323 * targhooks.c (default_slow_unaligned_access): New function.
15324 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15325 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15326 * doc/tm.texi: Regenerate.
15327 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15328 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15329 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15331 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15332 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15334 (rs6000_slow_unaligned_access): New function.
15335 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15336 (expand_block_compare): Likewise.
15337 (expand_strn_compare): Likewise.
15338 (rs6000_rtx_costs): Likewise.
15339 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15340 (riscv_slow_unaligned_access): Likewise.
15341 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15342 (riscv_slow_unaligned_access_p): ...this and make static.
15343 (riscv_option_override): Update accordingly.
15344 (riscv_slow_unaligned_access): New function.
15345 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15346 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15347 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15348 (rs6000_slow_unaligned_access): New function.
15349 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15350 (rs6000_rtx_costs): Likewise.
15351 * config/rs6000/rs6000-string.c (expand_block_compare)
15352 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15353 of SLOW_UNALIGNED_ACCESS.
15354 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15355 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15356 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15357 of SLOW_UNALIGNED_ACCESS.
15358 * expmed.c (simple_mem_bitfield_p): Likewise.
15359 * expr.c (alignment_for_piecewise_move): Likewise.
15360 (emit_group_load_1): Likewise.
15361 (emit_group_store): Likewise.
15362 (copy_blkmode_from_reg): Likewise.
15363 (emit_push_insn): Likewise.
15364 (expand_assignment): Likewise.
15365 (store_field): Likewise.
15366 (expand_expr_real_1): Likewise.
15367 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15368 * lra-constraints.c (simplify_operand_subreg): Likewise.
15369 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15370 * gimple-ssa-store-merging.c: Likewise in block comment at start
15372 * tree-ssa-strlen.c: Include target.h.
15373 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15374 of SLOW_UNALIGNED_ACCESS.
15375 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15377 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15379 PR rtl-optimization/82185
15380 * expmed.c (emit_store_flag_int): Only test tem if it has been
15383 2017-09-12 Richard Biener <rguenther@suse.de>
15385 PR middle-end/82149
15386 * match.pd ((FTYPE) N CMP CST): Fix typo.
15388 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15390 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15392 (mips_near_type_p): Add 'short_call' attribute as a synonym
15394 * doc/extend.texi (short_call): Document new function attribute.
15396 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15399 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15400 assertion check that in the condition.
15401 (get_atomic_generic_size): Likewise. Before testing if parameter
15402 has pointer type, if it has array type, call for C++
15403 default_conversion to perform array-to-pointer conversion.
15405 2017-09-12 Richard Biener <rguenther@suse.de>
15407 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15408 for operations we cannot scalarize.
15410 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15412 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15413 vectors heap vectors. Clean up comments.
15414 Make visited_bbs a reference.
15415 (profitable_jump_thread_path): Make GC
15416 vectors heap vectors. Clean up comments.
15418 (convert_and_register_jump_thread_path): Make GC vectors heap
15420 (check_subpath_and_update_thread_path): Same. Clean up comments.
15421 Make visited_bbs a reference.
15422 (handle_phi): Abstract common code to to
15423 register_jump_thread_path_if_profitable.
15424 Rename VAR_BB to DEF_BB.
15426 Make GC vectors heap vectors.
15427 Make visited_bbs a reference.
15428 (handle_assignment): Same.
15429 (register_jump_thread_path_if_profitable): New.
15430 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15432 Make GC vectors heap vectors. Clean up comments.
15433 Make visited_bbs a reference.
15434 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15435 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15438 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15441 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15442 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15444 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15446 Revert r251800 and r251799.
15448 2017-09-11 Martin Jambor <mjambor@suse.cz>
15451 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15452 arguments in advance.
15453 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15454 use it to find predecessor edges.
15455 (naive_outof_ssa): Collect vector of predecessors.
15457 2017-09-08 Jason Merrill <jason@redhat.com>
15459 PR c++/70029 - ICE with ref-qualifier and -flto
15460 * langhooks.h (struct lang_hooks_for_types): Add
15461 copy_lang_qualifiers.
15462 * attribs.c (build_type_attribute_qual_variant): Use it.
15463 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15465 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15466 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15468 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15471 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15472 (*mulsi3_sp64): New instruction.
15473 (mulsi3): New expander.
15475 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15477 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15479 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15481 * sancov.c: Include memmodel.h.
15483 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15486 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15489 2017-09-07 Carl Love <cel@us.ibm.com>
15491 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15492 the sldi instruction.
15494 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15496 * sancov.c: Include tm_p.h.
15498 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15501 * output.h (switch_to_other_text_partition): New declaration.
15502 * varasm.c (switch_to_other_text_partition): New function.
15503 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15504 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15505 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15506 to the other text partition before emitting LCL label and switch back
15507 after emitting the word after it.
15509 2017-09-07 Richard Biener <rguenther@suse.de>
15511 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15512 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15513 critical edges here, after loop init.
15514 (pass_data_pre): Remove PROP_no_crit_edges flags.
15515 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15516 for valueization of call args to avoid leaking VN_TOP.
15517 (visit_use): Assert we do not visit default defs.
15518 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15519 Use error_mark_node to more easily detect leaking VN_TOP.
15520 All default-defs are varying, not VN_TOP. Mark them visited.
15521 (run_scc_vn): Make code match comment.
15523 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15525 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15526 OPTION_MASK_FLOAT128_KEYWORD.
15527 (POWERPC_MASKS): Likewise.
15528 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15529 support for the -mfloat128-type option, and make -mfloat128
15530 default on PowerPC Linux systems. Define or undefine
15531 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15532 Define __float128 to be __ieee128 if IEEE 128-bit support is
15533 enabled, or undefine it.
15534 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15535 Delete defining __FLOAT128_TYPE__.
15536 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15537 -mfloat128-type option and make -mfloat128 default on PowerPC
15539 (TARGET_FLOAT128_TYPE): Likewise.
15540 (-mfloat128-type): Likewise.
15541 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15542 Delete the -mfloat128-type option and make -mfloat128 default on
15543 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15544 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15545 128-bit floating point is enabled. Change tests from using
15546 -mfloat128-type to -mfloat128.
15547 (rs6000_mangle_type): Use the correct mangling for the __float128
15548 type even if normal long double is restricted to 64-bits.
15549 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15551 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15552 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15553 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15555 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15556 documentation for -mfloat128.
15558 2017-09-06 Olivier Hainque <hainque@adacore.com>
15560 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15562 2017-09-06 Wish Wu <wishwu007@gmail.com>
15563 Jakub Jelinek <jakub@redhat.com>
15565 * asan.c (initialize_sanitizer_builtins): Add
15566 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15567 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15568 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15569 BT_FN_VOID_UINT64_PTR variables.
15570 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15571 (BT_FN_VOID_UINT16_UINT16): Likewise.
15572 (BT_FN_VOID_UINT32_UINT32): Likewise.
15573 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15574 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15575 (BT_FN_VOID_UINT64_PTR): Likewise.
15576 * common.opt (flag_sanitize_coverage): New variable.
15577 (fsanitize-coverage=trace-pc): Remove.
15578 (fsanitize-coverage=): Add.
15579 * flag-types.h (enum sanitize_coverage_code): New enum.
15580 * fold-const.c (fold_range_test): Disable non-short-circuit
15581 optimization if flag_sanitize_coverage.
15582 (fold_truth_andor): Likewise.
15583 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15584 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15585 (coverage_sanitizer_opts): New array.
15586 (get_closest_sanitizer_option): Add OPTS argument, handle also
15587 OPT_fsanitize_coverage_.
15588 (parse_sanitizer_options): Adjusted to also handle
15589 OPT_fsanitize_coverage_.
15590 (common_handle_option): Add OPT_fsanitize_coverage_.
15591 * sancov.c (instrument_comparison, instrument_switch): New function.
15592 (sancov_pass): Add trace-cmp support.
15593 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15594 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15595 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15596 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15597 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15598 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15599 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15600 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15601 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15602 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15604 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15606 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15607 error. Only quit immediately if parsing is complete.
15608 (BEGIN): Initialize fatal_err and parse_done.
15609 (begin fpu, end fpu): Check number of arguments.
15610 (begin arch, end arch): Likewise.
15611 (begin cpu, end cpu): Likewise.
15612 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15613 (optalias): Likewise.
15615 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15617 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15618 * config/arm/arm-isa.h: Delete. Move definitions to ...
15619 * arm-cpus.in: ... here. Use new feature and fgroup values.
15620 * config/arm/arm.c (arm_option_override): Use lower case for feature
15622 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15623 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15624 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15626 (print_isa_bits_for): New function.
15627 (gen_isa): New function.
15628 (gen_comm_data): Use print_isa_bits_for.
15629 (define feature): New keyword.
15630 (define fgroup): New keyword.
15631 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15632 (arm-isa.h): Add rule to generate file.
15633 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15634 case for feature bit names.
15636 2017-09-06 Richard Biener <rguenther@suse.de>
15638 * tree-ssa-pre.c (NECESSARY): Remove.
15639 (create_expression_by_pieces): Do not touch pass-local flags.
15640 (insert_into_preds_of_block): Likewise.
15641 (do_pre_regular_insertion): Likewise.
15642 (eliminate_insert): Likewise.
15643 (eliminate_dom_walker::before_dom_children): Likewise.
15644 (fini_eliminate): Do not look at inserted_exprs.
15645 (mark_operand_necessary): Remove.
15646 (remove_dead_inserted_code): Replace with simple work-list
15647 algorithm based on inserted_exprs and SSA uses.
15648 (pass_pre::execute): Re-order fini_eliminate and
15649 remove_dead_inserted_code.
15651 2017-09-06 Olivier Hainque <hainque@adacore.com>
15653 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15654 for VxWorks 7. Adjust surrounding comments.
15656 2017-09-06 Richard Biener <rguenther@suse.de>
15658 * gimple-ssa-strength-reduction.c
15659 (find_candidates_dom_walker::before_dom_children): Also allow
15662 2017-09-06 Richard Biener <rguenther@suse.de>
15664 PR tree-optimization/82108
15665 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15666 for gap in the non-permutation SLP case.
15668 2017-09-06 Martin Jambor <mjambor@suse.cz>
15670 PR tree-optimization/82078
15671 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15672 add_access_to_work_queue...
15673 (build_accesses_from_assign): ...here.
15674 (propagate_all_subaccesses): Make sure racc is the group
15675 representative, if there is one.
15677 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15679 PR middle-end/82095
15680 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15683 2017-09-06 Richard Biener <rguenther@suse.de>
15685 * gimple-ssa-strength-reduction.c
15686 (find_candidates_dom_walker::before_doom_children): Use a
15687 type and not a mode check.
15689 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15692 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15694 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15696 2017-09-05 Jeff Law <law@redhat.com>
15698 PR tree-optimization/64910
15699 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15700 cases where we have 3 or more operands.
15702 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15704 PR middle-end/81768
15705 * omp-low.c (lower_omp_for): Recompute tree invariant if
15706 gimple_omp_for_initial/final is ADDR_EXPR.
15708 PR middle-end/81768
15709 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15710 into gimple val before gimplification fo the COND_EXPR.
15712 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15714 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15715 REGION_COPY argument.
15716 (thread_through_all_blocks): Remove unused argument to
15717 duplicate_thread_path.
15719 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15720 Alan Hayward <alan.hayward@arm.com>
15721 David Sherwood <david.sherwood@arm.com>
15723 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15724 Take a scalar_mode rather than a machine_mode.
15725 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15726 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15727 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15728 (aarch64_gen_adjusted_ldpstp): Likewise.
15729 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15731 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15732 Alan Hayward <alan.hayward@arm.com>
15733 David Sherwood <david.sherwood@arm.com>
15735 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15736 Take a scalar_int_mode instead of a machine_mode.
15737 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15738 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15739 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15740 (aarch64_simd_attr_length_rglist): Delete.
15741 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15742 a scalar_int_mode instead of a machine_mode.
15743 (aarch64_add_offset): Likewise.
15744 (aarch64_internal_mov_immediate): Likewise
15745 (aarch64_add_constant_internal): Likewise.
15746 (aarch64_add_constant): Likewise.
15747 (aarch64_movw_imm): Likewise.
15748 (aarch64_rtx_arith_op_extract_p): Likewise.
15749 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15750 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15751 Remove assert that the mode isn't a vector.
15752 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15753 (aarch64_expand_mov_immediate): Update calls after above changes.
15754 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15755 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15756 (aarch64_move_imm): Likewise.
15757 (aarch64_can_const_movi_rtx_p): Likewise.
15758 (aarch64_strip_extend): Likewise.
15759 (aarch64_extr_rtx_p): Likewise.
15760 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15761 a CONST_INT when the mode parameter is VOIDmode.
15762 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15764 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15766 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15767 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15768 (bitwise_type_for_mode): Update accordingly.
15770 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15772 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15773 * stor-layout.c (mode_for_size_tree): Likewise.
15774 (mode_for_array): Update accordingly.
15775 (layout_decl): Likewise.
15776 (compute_record_mode): Likewise. Only set the mode once.
15778 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15780 * target.def (get_mask_mode): Change return type to opt_mode.
15782 * doc/tm.texi: Regenerate.
15783 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15784 * targhooks.c (default_get_mask_mode): Likewise.
15785 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15786 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15787 targetm.get_mask_mode.
15788 * tree.c (build_truth_vector_type): Likewise.
15790 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15792 * machmode.h (mode_for_vector): Return an opt_mode.
15793 * stor-layout.c (mode_for_vector): Likewise.
15794 (mode_for_int_vector): Update accordingly.
15795 (layout_type): Likewise.
15796 * config/i386/i386.c (emit_memmov): Likewise.
15797 (ix86_expand_set_or_movmem): Likewise.
15798 (ix86_expand_vector_init): Likewise.
15799 (ix86_get_mask_mode): Likewise.
15800 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15802 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15803 * expmed.c (extract_bit_field_1): Likewise.
15804 * expr.c (expand_expr_real_2): Likewise.
15805 * optabs-query.c (can_vec_perm_p): Likewise.
15806 (can_vec_mask_load_store_p): Likewise.
15807 * optabs.c (expand_vec_perm): Likewise.
15808 * targhooks.c (default_get_mask_mode): Likewise.
15809 * tree-vect-stmts.c (vectorizable_store): Likewise.
15810 (vectorizable_load): Likewise.
15811 (get_vectype_for_scalar_type_and_size): Likewise.
15813 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15815 * machmode.h (mode_for_int_vector): New function.
15816 * stor-layout.c (mode_for_int_vector): Likewise.
15817 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15818 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15819 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15820 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15821 (s390_expand_vcond): Likewise.
15823 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15825 * machmode.h (opt_machine_mode): New type.
15826 (opt_mode<T>): Allow construction from anything that can be
15828 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15829 (mode_for_size): Return an opt_machine_mode.
15830 * stor-layout.c (mode_for_size): Likewise.
15831 (mode_for_size_tree): Update call accordingly.
15832 (bitwise_mode_for_mode): Likewise.
15833 (make_fract_type): Likewise.
15834 (make_accum_type): Likewise.
15835 * caller-save.c (replace_reg_with_saved_mem): Update call
15837 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15838 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15839 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15840 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15841 * expmed.c (extract_bit_field_1): Likewise.
15842 * reload.c (get_secondary_mem): Likewise.
15843 * varasm.c (assemble_integer): Likewise.
15844 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15847 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15849 * machmode.h (decimal_float_mode_for_size): New function.
15850 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
15851 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
15852 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
15853 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
15854 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
15855 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
15857 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15859 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
15860 (get_builtin_sync_mode): Likewise.
15861 (expand_ifn_atomic_compare_exchange): Likewise.
15862 (expand_builtin_atomic_clear): Likewise.
15863 (expand_builtin_atomic_test_and_set): Likewise.
15864 (fold_builtin_atomic_always_lock_free): Likewise.
15865 * calls.c (compute_argument_addresses): Likewise.
15866 (emit_library_call_value_1): Likewise.
15867 (store_one_arg): Likewise.
15868 * combine.c (combine_instructions): Likewise.
15869 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
15870 * config/arm/arm.c (arm_function_value): Likewise.
15871 (aapcs_allocate_return_reg): Likewise.
15872 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
15873 * config/i386/i386.c (construct_container): Likewise.
15874 (ix86_gimplify_va_arg): Likewise.
15875 (ix86_expand_sse_cmp): Likewise.
15876 (emit_memmov): Likewise.
15877 (emit_memset): Likewise.
15878 (expand_small_movmem_or_setmem): Likewise.
15879 (ix86_expand_pextr): Likewise.
15880 (ix86_expand_pinsr): Likewise.
15881 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
15882 * config/microblaze/microblaze.c (microblaze_block_move_straight):
15884 * config/mips/mips.c (mips_function_value_1) Likewise.
15885 (mips_block_move_straight): Likewise.
15886 (mips_expand_ins_as_unaligned_store): Likewise.
15887 * config/powerpcspe/powerpcspe.c
15888 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15889 (rs6000_darwin64_record_arg_flush): Likewise.
15890 * config/rs6000/rs6000.c
15891 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15892 (rs6000_darwin64_record_arg_flush): Likewise.
15893 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
15894 (sparc_function_value_1): Likewise.
15895 * config/spu/spu.c (adjust_operand): Likewise.
15896 (spu_emit_branch_or_set): Likewise.
15897 (arith_immediate_p): Likewise.
15898 * emit-rtl.c (gen_lowpart_common): Likewise.
15899 * expr.c (expand_expr_real_1): Likewise.
15900 * function.c (assign_parm_setup_block): Likewise.
15901 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
15902 * reload1.c (alter_reg): Likewise.
15903 * stor-layout.c (mode_for_vector): Likewise.
15904 (layout_type): Likewise.
15906 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15908 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
15909 (spu_convert_move): Likewise.
15910 * lower-subreg.c (resolve_simple_move): Likewise.
15912 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15915 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
15916 define_insn to a define_expand.
15917 (altivec_vsum2sws_direct): New define_insn.
15918 (altivec_vsumsws): Convert from a define_insn to a define_expand.
15920 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15922 * config/arm/arm.c (arm_option_params_internal): Improve setting of
15925 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
15930 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
15931 realign stack if stack alignment needed is less than incoming
15934 2017-09-05 Marek Polacek <polacek@redhat.com>
15937 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
15940 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15942 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
15944 2017-09-05 Richard Biener <rguenther@suse.de>
15946 PR tree-optimization/82084
15947 * fold-const.c (can_native_encode_string_p): Handle wide characters.
15949 2017-09-05 Richard Biener <rguenther@suse.de>
15951 PR tree-optimization/82102
15952 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
15954 2017-09-05 Martin Liska <mliska@suse.cz>
15956 PR tree-optimization/82032
15957 * tree-cfg.c (generate_range_test): New function.
15958 * tree-cfg.h (generate_range_test): Declared here.
15959 * tree-cfgcleanup.c (convert_single_case_switch): New function.
15960 (cleanup_control_expr_graph): Use it.
15961 * tree-switch-conversion.c (try_switch_expansion): Remove
15963 (emit_case_nodes): Use generate_range_test.
15965 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15968 * config/i386/i386.md (*<btsc><mode>_mask): Add
15969 TARGET_USE_BT to insn constraint.
15970 (*btr<mode>_mask): Ditto.
15972 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
15974 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
15975 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
15977 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
15980 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
15981 TARGET_NEON and TARGET_IWMMXT.
15982 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
15983 TARGET_NEON and TARGET_IWMMXT.
15984 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
15986 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15988 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
15989 (ix86_rewrite_tls_address): Ditto.
15990 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
15991 (ix86_rewrite_tls_address_1): Ditto.
15992 (ix86_rewrite_tls_address): Ditto.
15993 * config/i386/predicates.md (tls_address_pattern): New predicate.
15994 * config/i386/i386.md (TLS address splitter): New splitter.
15996 2017-09-04 Richard Biener <rguenther@suse.de>
15998 PR tree-optimization/82084
15999 * fold-const.h (can_native_encode_string_p): Declare.
16000 * fold-const.c (can_native_encode_string_p): Factor out from ...
16001 (native_encode_string): ... here.
16002 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16003 vectorizing stores from constants we later cannot handle.
16005 2017-09-04 Marek Polacek <polacek@redhat.com>
16008 * doc/invoke.texi: Update -Wtautological-compare documentation.
16010 2017-09-04 Jeff Law <law@redhat.com>
16012 PR tree-optimization/64910
16013 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16014 swap the first and last operand if the last is a constant.
16016 2017-09-04 Marek Polacek <polacek@redhat.com>
16019 * convert.c (do_narrow): When sanitizing signed integer overflows,
16020 bail out for signed types.
16021 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16023 2017-09-04 Richard Biener <rguenther@suse.de>
16025 PR tree-optimization/82060
16026 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16027 Move devirtualization after stmt folding and before EH/AB/noreturn
16028 cleanup to get the stmt refs canonicalized. Use a bool instead
16029 of gimple_modified_p since that doesn't work for NOPs. Schedule
16030 NOPs generated by folding for removal.
16032 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16033 Alan Hayward <alan.hayward@arm.com>
16034 David Sherwood <david.sherwood@arm.com>
16036 * coretypes.h (pad_direction): New enum.
16037 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16038 (FUNCTION_ARG_PADDING): Likewise.
16039 * target.def (function_arg_padding): New hook.
16040 * targhooks.h (default_function_arg_padding): Declare.
16041 * targhooks.c (default_function_arg_padding): New function.
16042 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16043 (TARGET_FUNCTION_ARG_PADDING): ...this.
16044 * doc/tm.texi: Regenerate.
16045 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16046 instead of direction.
16047 (compute_argument_addresses): Likewise.
16048 (load_register_parameters): Likewise.
16049 (emit_library_call_value_1): Likewise.
16050 (store_one_arg): Use targetm.calls.function_arg_padding instead
16051 of FUNCTION_ARG_PADDING.
16052 (must_pass_in_stack_var_size_or_pad): Likewise.
16053 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16054 (emit_group_store): Likewise.
16055 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16056 instead of FUNCTION_ARG_PADDING.
16057 (emit_push_insn): Likewise, and propagate enum change throughout
16059 * function.h (direction): Delete.
16060 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16062 * function.c (assign_parm_find_stack_rtl): Likewise.
16063 (assign_parm_setup_block_p): Likewise.
16064 (assign_parm_setup_block): Likewise.
16065 (gimplify_parameters): Likewise.
16066 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16067 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16069 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16070 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16071 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16072 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16073 (aarch64_function_arg_padding): ...this new function.
16074 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16075 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16076 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16077 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16078 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16079 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16080 (arm_pad_arg_upward): Replace with...
16081 (arm_function_arg_padding): ...this new function.
16082 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16084 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16085 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16086 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16087 (ia64_hpux_function_arg_padding): Replace with...
16088 (ia64_function_arg_padding): ...this new function. Use pad_direction
16089 instead of direction. Check for TARGET_HPUX.
16090 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16091 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16092 (iq2000_function_arg_padding): New function.
16093 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16094 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16095 (mips_function_arg_padding): ...this new function.
16096 (mips_pad_reg_upward): Update accordingly.
16097 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16098 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16099 targetm.calls.function_arg_padding.
16100 (FUNCTION_ARG_PADDING): Delete.
16101 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16102 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16103 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16104 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16105 (nios2_block_reg_padding): Return pad_direction instead of direction.
16106 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16107 instead of direction.
16108 (nios2_function_arg_padding): Likewise. Make static.
16109 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16110 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16111 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16112 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16113 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16114 (pa_function_arg_padding): Make static. Return pad_direction instead
16116 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16117 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16118 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16119 instead of direction. Use targetm.calls.function_arg_padding.
16120 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16121 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16122 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16123 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16124 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16126 (function_arg_padding): Rename to...
16127 (rs6000_function_arg_padding): ...this. Make static. Return
16128 pad_direction instead of direction.
16129 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16130 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16131 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16132 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16133 instead of direction. Use targetm.calls.function_arg_padding.
16134 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16135 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16136 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16137 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16138 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16139 (function_arg_padding): Rename to...
16140 (rs6000_function_arg_padding): ...this. Make static. Return
16141 pad_direction instead of direction.
16142 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16143 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16144 * config/s390/s390.c (s390_function_arg_padding): New function.
16145 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16146 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16147 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16148 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16149 (function_arg_padding): Rename to...
16150 (sparc_function_arg_padding): ...this. Make static. Return
16151 pad_direction instead of direction.
16152 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16153 * config/spu/spu.c (spu_function_arg_padding): New function.
16154 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16155 * system.h (FUNCTION_ARG_PADDING): Poison.
16157 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16158 Alan Hayward <alan.hayward@arm.com>
16159 David Sherwood <david.sherwood@arm.com>
16161 * target.def (modes_tieable_p): New hook.
16162 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16163 (TARGET_MODES_TIEABLE_P): ...this.
16164 * doc/tm.texi.in: Regenerate.
16165 * hooks.h (hook_bool_mode_mode_true): Declare.
16166 * hooks.c (hook_bool_mode_mode_true): New function.
16167 * combine.c (subst): Use targetm.modes_tieable_p instead of
16169 * dse.c (find_shift_sequence): Likewise.
16170 * expmed.c (extract_low_bits): Likewise.
16171 * lower-subreg.c: Include target.h.
16172 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16174 * rtlanal.c (rtx_cost): Likewise.
16175 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16176 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16177 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16178 (TARGET_MODES_TIEABLE_P): Redefine.
16179 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16180 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16181 (TARGET_MODES_TIEABLE_P): Redefine.
16182 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16183 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16184 (arc_modes_tieable_p): New function.
16185 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16186 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16187 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16188 (arm_modes_tieable_p): Make static.
16189 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16190 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16191 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16192 (TARGET_MODES_TIEABLE_P): Redefine.
16193 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16194 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16195 (TARGET_MODES_TIEABLE_P): Redefine.
16196 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16197 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16198 (cr16_modes_tieable_p): New function.
16199 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16200 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16201 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16202 (TRULY_NOOP_TRUNCATION): Update comment.
16203 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16204 (TRULY_NOOP_TRUNCATION): Update comment.
16205 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16206 (frv_modes_tieable_p): New function.
16207 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16208 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16209 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16210 (TARGET_MODES_TIEABLE_P): Redefine.
16211 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16212 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16213 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16214 (TARGET_MODES_TIEABLE_P): Redefine.
16215 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16216 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16217 (ia64_modes_tieable_p): New function.
16218 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16219 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16220 (iq2000_modes_tieable_p): New function.
16221 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16222 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16223 (lm32_modes_tieable_p): New function.
16224 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16225 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16226 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16227 (TARGET_MODES_TIEABLE_P): Redefine.
16228 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16229 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16230 (m32r_modes_tieable_p): New function.
16231 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16232 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16233 (m68k_modes_tieable_p): New function.
16234 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16235 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16236 (mcore_modes_tieable_p): New function.
16237 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16238 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16240 (TARGET_MODES_TIEABLE_P): Redefine.
16241 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16242 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16243 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16244 (TARGET_MODES_TIEABLE_P): Redefine.
16245 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16246 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16247 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16248 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16249 (mn10300_modes_tieable_p): ...this and make static.
16250 (TARGET_MODES_TIEABLE_P): Redefine.
16251 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16252 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16253 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16254 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16255 (msp430_modes_tieable_p): Make static.
16256 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16257 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16258 (TARGET_MODES_TIEABLE_P): Redefine.
16259 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16260 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16261 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16262 (TARGET_MODES_TIEABLE_P): Redefine.
16263 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16264 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16265 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16266 (TARGET_MODES_TIEABLE_P): Redefine.
16267 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16268 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16269 (pdp11_modes_tieable_p): New function.
16270 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16271 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16272 (rs6000_modes_tieable_p): New function.
16273 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16274 * config/powerpcspe/powerpcspe.md: Update comment.
16275 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16276 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16277 (TARGET_MODES_TIEABLE_P): Redefine.
16278 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16279 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16280 (rl78_modes_tieable_p): New function.
16281 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16282 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16283 (rs6000_modes_tieable_p): New function.
16284 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16285 * config/rs6000/rs6000.md: Update comment.
16286 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16287 * config/rx/rx.c (rx_modes_tieable_p): New function.
16288 (TARGET_MODES_TIEABLE_P): Redefine.
16289 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16290 * config/s390/s390.c (s390_modes_tieable_p): New function.
16291 (TARGET_MODES_TIEABLE_P): Redefine.
16292 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16293 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16294 (sh_modes_tieable_p): New function.
16295 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16296 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16297 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16298 (sparc_modes_tieable_p): Make static.
16299 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16300 * config/spu/spu.c (spu_modes_tieable_p): New function.
16301 (TARGET_MODES_TIEABLE_P): Redefine.
16302 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16303 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16304 (TARGET_MODES_TIEABLE_P): Redefine.
16305 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16306 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16307 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16308 * config/v850/v850.c (v850_modes_tieable_p): New function.
16309 (TARGET_MODES_TIEABLE_P): Redefine.
16310 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16311 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16312 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16313 (visium_modes_tieable_p): New function.
16314 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16315 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16316 (xtensa_modes_tieable_p): New function.
16317 * system.h (MODES_TIEABLE_P): Poison.
16319 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16320 Alan Hayward <alan.hayward@arm.com>
16321 David Sherwood <david.sherwood@arm.com>
16323 * target.def (hard_regno_mode_ok): New hook.
16324 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16325 (TARGET_HARD_REGNO_MODE_OK): ...this.
16326 * doc/tm.texi.in: Regenerate.
16327 * hooks.h (hook_bool_uint_mode_true): Declare.
16328 * hooks.c (hook_bool_uint_mode_true): New function.
16329 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16330 HARD_REGNO_MODE_OK.
16331 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16332 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16333 instead of HARD_REGNO_MODE_OK.
16334 * caller-save.c: Include target.h.
16335 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16336 HARD_REGNO_MODE_OK.
16337 * combine.c (can_combine_p): Likewise.
16338 (combinable_i3pat): Likewise.
16339 (can_change_dest_mode): Likewise.
16340 * expr.c (init_expr_target): Likewise.
16341 (convert_move): Likewise.
16342 (convert_modes): Likewise.
16343 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16344 (setup_prohibited_mode_move_regs): Likewise.
16345 * ira.h (target_ira): Likewise.
16346 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16347 * lra-constraints.c (process_alt_operands): Likewise.
16348 (split_reg): Likewise.
16349 * recog.c (peep2_find_free_register): Likewise.
16350 * ree.c (combine_reaching_defs): Likewise.
16351 * regcprop.c (maybe_mode_change): Likewise.
16352 * reginfo.c (init_reg_sets_1): Likewise.
16353 (choose_hard_reg_mode): Likewise.
16354 (simplifiable_subregs): Likewise.
16355 * regrename.c (check_new_reg_p): Likewise.
16356 * reload.c (find_valid_class): Likewise.
16357 (find_valid_class_1): Likewise.
16358 (reload_inner_reg_of_subreg): Likewise.
16359 (push_reload): Likewise.
16360 (combine_reloads): Likewise.
16361 (find_dummy_reload): Likewise.
16362 (find_reloads): Likewise.
16363 * reload1.c (find_reg): Likewise.
16364 (set_reload_reg): Likewise.
16365 (allocate_reload_reg): Likewise.
16366 (choose_reload_regs): Likewise.
16367 (reload_adjust_reg_for_temp): Likewise.
16368 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16369 (simplify_subreg_regno): Likewise.
16370 * sel-sched.c (init_regs_for_mode): Likewise.
16371 * varasm.c (make_decl_rtl): Likewise.
16372 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16373 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16374 HARD_REGNO_MODE_OK.
16375 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16376 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16377 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16378 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16379 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16380 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16381 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16382 (arc_mode_class): Delete.
16383 (HARD_REGNO_MODE_OK): Delete.
16384 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16385 (arc_hard_regno_mode_ok): Rename old array to...
16386 (arc_hard_regno_mode_ok_modes): ...this.
16387 (arc_conditional_register_usage): Update accordingly.
16388 (arc_mode_class): Make static.
16389 (arc_hard_regno_mode_ok): New function.
16390 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16391 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16392 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16393 (arm_hard_regno_mode_ok): Make static.
16394 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16395 HARD_REGNO_MODE_OK.
16396 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16397 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16398 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16400 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16401 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16402 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16403 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16404 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16405 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16406 * config/bfin/predicates.md (valid_reg_operand): Use
16407 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16408 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16409 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16410 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16411 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16412 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16413 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16414 (cr16_hard_regno_mode_ok): Make static and return a bool.
16415 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16416 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16417 (cris_hard_regno_mode_ok): New function.
16418 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16419 (epiphany_mode_class): Delete.
16420 (HARD_REGNO_MODE_OK): Delete.
16421 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16422 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16423 (hard_regno_mode_ok): Rename to...
16424 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16425 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16426 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16427 HARD_REGNO_MODE_OK.
16428 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16429 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16430 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16431 (frv_hard_regno_mode_ok): Make static and return a bool.
16432 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16433 HARD_REGNO_MODE_OK.
16434 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16435 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16436 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16437 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16439 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16440 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16441 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16442 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16444 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16445 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16446 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16447 (ia64_hard_regno_mode_ok): New function.
16448 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16449 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16450 (iq2000_hard_regno_mode_ok): New function.
16451 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16452 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16453 (lm32_hard_regno_mode_ok): New function.
16454 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16455 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16456 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16457 instead of HARD_REGNO_MODE_OK.
16458 (m32c_hard_regno_ok): Rename to...
16459 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16460 (m32c_cannot_change_mode_class): Update accordingly.
16461 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16462 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16463 (m32r_mode_class): Delete.
16464 (HARD_REGNO_MODE_OK): Delete.
16465 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16466 (m32r_hard_regno_mode_ok): Rename to...
16467 (m32r_hard_regno_modes): ...this.
16468 (m32r_mode_class): Make static.
16469 (m32r_hard_regno_mode_ok): New function.
16470 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16471 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16472 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16473 (m68k_hard_regno_mode_ok): Make static.
16474 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16475 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16476 (mcore_hard_regno_mode_ok): New function.
16477 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16478 (HARD_REGNO_MODE_OK): Delete.
16479 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16481 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16482 (microblaze_hard_regno_mode_ok): New function.
16483 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16484 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16485 (mips_hard_regno_mode_ok): Delete.
16486 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16487 (mips_hard_regno_mode_ok_p): ...this and make static.
16488 (mips_hard_regno_mode_ok_p): Rename to...
16489 (mips_hard_regno_mode_ok_uncached): ...this.
16490 (mips_hard_regno_mode_ok): New function.
16491 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16492 of HARD_REGNO_MODE_OK.
16493 (mips_option_override): Update after above name changes.
16494 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16495 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16496 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16497 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16498 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16499 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16500 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16501 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16502 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16503 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16504 (msp430_hard_regno_mode_ok): Make static and return a bool.
16505 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16506 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16507 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16509 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16510 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16511 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16512 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16513 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16514 (PA_HARD_REGNO_MODE_OK): ...this
16515 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16516 (PA_HARD_REGNO_MODE_OK): ...this.
16517 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16518 (pa_hard_regno_mode_ok): New function.
16519 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16520 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16521 (pdp11_hard_regno_mode_ok): New function.
16522 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16523 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16525 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16527 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16528 (rs6000_hard_regno_mode_ok): Rename to...
16529 (rs6000_hard_regno_mode_ok_uncached): ...this.
16530 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16531 (rs6000_hard_regno_mode_ok): New function.
16532 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16533 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16534 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16535 (riscv_hard_regno_mode_ok): ...this and make static.
16536 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16537 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16538 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16539 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16540 (rl78_hard_regno_mode_ok): Make static and return bool.
16541 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16542 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16544 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16545 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16546 (rs6000_hard_regno_mode_ok): Rename to...
16547 (rs6000_hard_regno_mode_ok_uncached): ...this.
16548 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16549 (rs6000_hard_regno_mode_ok): New function.
16550 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16551 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16552 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16553 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16554 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16555 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16556 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16557 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16558 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16559 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16560 (sh_hard_regno_mode_ok): Make static.
16561 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16562 instead of HARD_REGNO_MODE_OK.
16563 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16564 (sparc_mode_class): Delete.
16565 (HARD_REGNO_MODE_OK): Delete.
16566 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16567 (hard_regno_mode_classes): Make static.
16568 (sparc_mode_class): Likewise.
16569 (sparc_hard_regno_mode_ok): New function.
16570 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16571 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16572 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16574 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16575 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16576 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16577 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16578 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16579 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16580 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16581 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16582 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16583 (visium_hard_regno_mode_ok): New function.
16584 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16585 instead of HARD_REGNO_MODE_OK.
16586 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16587 (HARD_REGNO_MODE_OK): Delete.
16588 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16589 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16590 (xtensa_option_override): Update accordingly.
16591 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16592 (xtensa_hard_regno_mode_ok): New function.
16593 * system.h (HARD_REGNO_MODE_OK): Poison.
16595 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16596 Alan Hayward <alan.hayward@arm.com>
16597 David Sherwood <david.sherwood@arm.com>
16599 * target.def (hard_regno_call_part_clobbered): New hook.
16600 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16601 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16602 * doc/tm.texi: Regenerate.
16603 * hooks.h (hook_bool_uint_mode_false): Declare.
16604 * hooks.c (hook_bool_uint_mode_false): New function.
16605 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16606 * cselib.c (cselib_process_insn): Use
16607 targetm.hard_regno_call_part_clobbered instead of
16608 HARD_REGNO_CALL_PART_CLOBBERED.
16609 * ira-conflicts.c (ira_build_conflicts): Likewise.
16610 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16611 * lra-constraints.c (need_for_call_save_p): Likewise.
16612 * lra-lives.c: Include target.h.
16613 (check_pseudos_live_through_calls): Use
16614 targetm.hard_regno_call_part_clobbered instead of
16615 HARD_REGNO_CALL_PART_CLOBBERED.
16616 * regcprop.c: Include target.h.
16617 (copyprop_hardreg_forward_1): Use
16618 targetm.hard_regno_call_part_clobbered instead of
16619 HARD_REGNO_CALL_PART_CLOBBERED.
16620 * reginfo.c (choose_hard_reg_mode): Likewise.
16621 * regrename.c (check_new_reg_p): Likewise.
16622 * reload.c (find_equiv_reg): Likewise.
16623 * reload1.c (emit_reload_insns): Likewise.
16624 * sched-deps.c (deps_analyze_insn): Likewise.
16625 * sel-sched.c (init_regs_for_mode): Likewise.
16626 (mark_unavailable_hard_regs): Likewise.
16627 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16628 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16629 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16631 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16632 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16633 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16635 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16637 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16638 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16639 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16641 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16642 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16643 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16645 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16646 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16648 * config/powerpcspe/powerpcspe.c
16649 (rs6000_hard_regno_call_part_clobbered): New function.
16650 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16651 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16652 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16654 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16655 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16656 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16658 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16659 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16660 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16662 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16663 Alan Hayward <alan.hayward@arm.com>
16664 David Sherwood <david.sherwood@arm.com>
16666 * rtl.h (subreg_memory_offset): Declare.
16667 * emit-rtl.c (subreg_memory_offset): New function.
16668 * expmed.c (store_bit_field_1): Use it.
16669 * expr.c (undefined_operand_subword_p): Likewise.
16670 * simplify-rtx.c (simplify_subreg): Likewise.
16672 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16674 PR rtl-optimization/57448
16677 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16678 using atomic_load pattern.
16679 (expand_atomic_store): Likewise.
16681 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16684 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16685 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16686 handling. Use replace_call_with_value with NULL instead of
16687 gsi_replace, unlink_stmt_vdef and release_defs.
16689 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16692 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16694 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16697 * rtl.h (emit_library_call_value_1): Declare.
16698 (emit_library_call): Replace declaration with a series of overloads.
16699 Remove the parameter count argument.
16700 (emit_library_call_value): Likewise.
16701 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16702 with an "rtx_mode_t *".
16703 (emit_library_call_value): Delete.
16704 (emit_library_call): Likewise.
16705 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16706 (asan_emit_allocas_unpoison): Likewise.
16707 * builtins.c (expand_builtin_powi): Likewise.
16708 (expand_asan_emit_allocas_unpoison): Likewise.
16709 * cfgexpand.c (expand_main_function): Likewise.
16710 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16711 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16712 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16713 * config/arm/arm.c (arm_trampoline_init): Likewise.
16714 (arm_call_tls_get_addr): Likewise.
16715 (arm_expand_divmod_libfunc): Likewise.
16716 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16717 (smulsi3_highpart): Likewise.
16718 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16719 (c6x_expand_compare): Likewise.
16720 (c6x_expand_movmem): Likewise.
16721 * config/frv/frv.c (frv_trampoline_init): Likewise.
16722 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16723 (ix86_expand_divmod_libfunc): Likewise.
16724 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16725 (ia64_expand_compare): Likewise.
16726 (ia64_profile_hook): Likewise.
16727 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16728 (nonlocal_goto): Likewise.
16729 (restore_stack_nonlocal): Likewise.
16730 * config/m32r/m32r.c (block_move_call): Likewise.
16731 (m32r_trampoline_init): Likewise.
16732 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16733 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16734 (m68k_call_m68k_read_tp): Likewise.
16735 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16736 (microblaze_expand_divide): Likewise.
16737 * config/mips/mips.h (mips_args): Likewise.
16738 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16739 (MIPS_ICACHE_SYNC): Likewise.
16740 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16741 (nios2_trampoline_init): Likewise.
16742 * config/pa/pa.c (hppa_tls_call): Likewise.
16743 (pa_trampoline_init): Likewise.
16744 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16745 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16746 (expand_strn_compare): Likewise.
16747 (rs6000_generate_compare): Likewise.
16748 (rs6000_expand_float128_convert): Likewise.
16749 (output_profile_hook): Likewise.
16750 (rs6000_trampoline_init): Likewise.
16751 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16752 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16753 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16754 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16755 (rs6000_generate_compare): Likewise.
16756 (rs6000_expand_float128_convert): Likewise.
16757 (output_profile_hook): Likewise.
16758 (rs6000_trampoline_init): Likewise.
16759 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16760 * config/sh/sh.c (sh_trampoline_init): Likewise.
16761 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16762 (sparc_emit_float_lib_cmp): Likewise.
16763 (sparc32_initialize_trampoline): Likewise.
16764 (sparc64_initialize_trampoline): Likewise.
16765 (sparc_profile_hook): Likewise.
16766 * config/spu/spu.c (ea_load_store): Likewise.
16767 * config/spu/spu.md (floatunssidf2): Likewise.
16768 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16769 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16770 * config/visium/visium.c (expand_block_move_4): Likewise.
16771 (expand_block_move_2): Likewise.
16772 (expand_block_move_1): Likewise.
16773 (expand_block_set_4): Likewise.
16774 (expand_block_set_2): Likewise.
16775 (expand_block_set_1): Likewise.
16776 (visium_trampoline_init): Likewise.
16777 (visium_profile_hook): Likewise.
16778 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16779 (xtensa_setup_frame_addresses): Likewise.
16780 (xtensa_trampoline_init): Likewise.
16781 * except.c (sjlj_emit_function_enter): Likewise.
16782 (sjlj_emit_function_exit): Likewise.
16783 * explow.c (allocate_dynamic_stack_space): Likewise.
16784 (probe_stack_range): Likewise.
16785 * expr.c (convert_mode_scalar): Likewise.
16786 * optabs.c (expand_binop): Likewise.
16787 (expand_twoval_binop_libfunc): Likewise.
16788 (expand_unop): Likewise.
16789 (prepare_cmp_insn): Likewise.
16790 (prepare_float_lib_cmp): Likewise.
16791 (expand_float): Likewise.
16792 (expand_fix): Likewise.
16793 (expand_fixed_convert): Likewise.
16794 (maybe_emit_sync_lock_test_and_set): Likewise.
16795 (expand_atomic_compare_and_swap): Likewise.
16796 (expand_mem_thread_fence): Likewise.
16797 (expand_atomic_fetch_op): Likewise.
16799 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16801 * doc/generic.texi (OpenACC): Adjust URL.
16802 * doc/invoke.texi (C Dialect Options): Ditto.
16804 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16806 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16807 predicate for operand 1. Add (m,<S>) constraint.
16808 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16809 Prevent memory operand 1 with register operand 2.
16811 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16813 PR rtl-optimization/82024
16814 * combine.c (try_combine): If the combination result is a PARALLEL,
16815 and we only need to retain the SET in there that would be placed
16816 at I2, check that we can place that at I3 instead, before doing so.
16818 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16821 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16823 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16824 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16827 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16828 Jeff Law <law@redhat.com>
16830 * varasm.c (bss_initializer_p): Do not put constants into .bss
16831 (categorize_decl_for_section): Handle bss_initializer_p returning
16832 false when DECL_INITIAL is NULL.
16834 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16837 * config/s390/s390.c (s390_can_inline_p): New function.
16839 2017-09-01 Jeff Law <law@redhat.com>
16841 PR tree-optimization/82052
16842 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16843 Always initialize the returned slot after a hash table miss
16844 when INSERT is true.
16846 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16848 * config/s390/s390.md (mem_signal_fence): Remove.
16849 * doc/md.texi (mem_signal_fence): Remove.
16850 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
16852 * target-insns.def (mem_signal_fence): Remove.
16854 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16857 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
16860 * asan.c (create_odr_indicator): Strip name encoding from assembler
16861 name before appending it after __odr_asan_.
16863 2017-09-01 Martin Liska <mliska@suse.cz>
16865 PR tree-optimization/82059
16866 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
16867 frequency only when an edge is redirected.
16869 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16871 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
16872 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
16873 (arc_conditional_register_usage): Remove ARC600 lp_count
16875 (arc_file_start): Emit Tag_ARC_CPU_variation.
16876 (arc_can_use_doloop_p): New conditions to use ZOLs.
16877 (hwloop_fail): New function.
16878 (hwloop_optimize): Likewise.
16879 (hwloop_pattern_reg): Likewise.
16880 (arc_doloop_hooks): New struct, to be used with reorg_loops.
16881 (arc_reorg_loops): New function, calls reorg_loops.
16882 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
16883 (arc600_corereg_hazard): Remove ZOL checking, case handled by
16885 (arc_loop_hazard): Remove function, functionality moved into
16887 (arc_hazard): Remove arc_loop_hazard call.
16888 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
16889 into hwloop_optimize.
16890 (arc_label_align): Remove ZOL handling.
16891 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
16892 * config/arc/arc.md (doloop_begin): Remove pattern.
16893 (doloop_begin_i): Likewise.
16894 (doloop_end_i): Likewise.
16895 (doloop_fallback): Likewise.
16896 (doloop_fallback_m): Likewise.
16897 (doloop_end): Reimplement expand.
16898 (arc_lp): New pattern for LP instruction.
16899 (loop_end): New pattern.
16900 (loop_fail): Likewise.
16901 (decrement_and_branch_until_zero): Likewise.
16902 * config/arc/arc.opt (mlpc-width): New option.
16903 * doc/invoke.texi (mlpc-width): Document option.
16905 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16907 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
16908 (arc_ccfsm_advance): Fix checking for delay slots.
16909 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
16911 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16913 * config/arc/arc.md (movqi_insn): Add stores to save constant long
16915 (movhi_insn): Update store instruction constraint which are saving
16916 6-bit short immediates.
16917 (movsi_insn): Consider also short scaled load operations.
16918 (zero_extendhisi2_i): Use Usd constraint instead of T.
16919 (extendhisi2_i): Add q constraint.
16920 (arc_clzsi2): Add type and length attributes.
16921 (arc_ctzsi2): Likewise.
16922 * config/arc/constraints.md (Usc): Update constraint, the
16923 assembler can parse two relocations for a single instruction.
16925 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16927 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
16928 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
16930 2017-08-31 Olivier Hainque <hainque@adacore.com>
16932 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
16933 match as powerpc-wrs-vxworks*.
16935 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
16937 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
16938 register constraint for by-element operand.
16939 (aarch64_mls_elt_merge<mode>): Likewise.
16941 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16943 * config/arc/arc.c (arc_can_follow_jump): Check for short
16946 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16948 * config.gcc: Use g.opt for arc.
16949 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
16950 functionality moved to ...
16951 (legitimate_scaled_address_p): New function, ...here.
16952 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
16953 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
16954 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
16956 (arc_override_options): Handle G option.
16957 (arc_output_pic_addr_const): Correct function definition.
16958 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
16959 (arc_decl_anon_ns_mem_p): Delete.
16960 (arc_in_small_data_p): Overhaul this function to take into
16961 consideration the value given via G option.
16962 (arc_rewrite_small_data_1): Renamed and corrected old
16963 arc_rewrite_small_data function.
16964 (arc_rewrite_small_data): New function.
16965 (small_data_pattern): Don't use pic_offset_table_rtx.
16966 * config/arc/arc.h (CC1_SPEC): Recognize G option.
16967 * config/arc/simdext.md (movmisalignv2hi): Use
16968 prepare_move_operands function.
16970 (movmisalign*): Likewise.
16971 * doc/invoke.texi (ARC options): Document -G option.
16973 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16975 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
16977 * config/arc/arc.c (arc_print_operand): Output scalled address for
16978 sdata whenever is possible.
16979 (arc_in_small_data_p): Allow sdata for 64bit datum when double
16980 load/stores are available.
16981 (compact_sda_memory_operand): Check for the alignment required by
16982 code density instructions.
16983 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
16985 * config/arc/constraints.md (Usd): Update constraint.
16986 (Us0): New constraint.
16987 (Usc): Update constraint.
16989 2017-08-31 Richard Biener <rguenther@suse.de>
16991 PR middle-end/82054
16992 * dwarf2out.c (dwarf2out_early_global_decl): Process each
16993 function only once.
16995 2017-08-31 Tamar Christina <tamar.christina@arm.com>
16997 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
16998 Resize type_signature.
17000 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17001 Alan Hayward <alan.hayward@arm.com>
17002 David Sherwood <david.sherwood@arm.com>
17004 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17005 subregs whose inner modes can be stored in GPRs.
17006 (aarch64_classify_index): Likewise.
17008 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17009 Alan Hayward <alan.hayward@arm.com>
17010 David Sherwood <david.sherwood@arm.com>
17012 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17013 (V_INT_EQUIV): ...this.
17014 (v_cmp_result): Rename to...
17015 (v_int_equiv): ...this.
17016 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17017 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17018 (copysign<mode>3): Likewise.
17019 (aarch64_simd_bsl<mode>_internal): Likewise.
17020 (aarch64_simd_bsl<mode>): Likewise.
17021 (vec_cmp<mode><mode>): Likewise.
17022 (vcond<mode><mode>): Likewise.
17023 (vcond<v_cmp_mixed><mode>): Likewise.
17024 (vcondu<mode><v_cmp_mixed>): Likewise.
17025 (aarch64_cm<optab><mode>): Likewise.
17026 (aarch64_cmtst<mode>): Likewise.
17027 (aarch64_fac<optab><mode>): Likewise.
17028 (vec_perm_const<mode>): Likewise.
17029 (vcond_mask_<mode><v_cmp_result>): Rename to...
17030 (vcond_mask_<mode><v_int_equiv>): ...this.
17031 (vec_cmp<mode><v_cmp_result>): Rename to...
17032 (vec_cmp<mode><v_int_equiv>): ...this.
17034 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17035 Alan Hayward <alan.hayward@arm.com>
17036 David Sherwood <david.sherwood@arm.com>
17038 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17040 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17041 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17042 (UNSPEC_LD4_DREG): New unspecs.
17043 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17044 (aarch64_ld2<mode>_dreg_be): Replace with...
17045 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17047 (aarch64_ld3<mode>_dreg_le)
17048 (aarch64_ld3<mode>_dreg_be): Replace with...
17049 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17051 (aarch64_ld4<mode>_dreg_le)
17052 (aarch64_ld4<mode>_dreg_be): Replace with...
17053 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17056 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17058 PR tree-optimization/81987
17059 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17060 insert an initializer in a location not dominated by the stride
17063 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17065 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17066 on the entire header of the finally block in the fallthru case.
17068 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17070 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17072 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17074 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17075 rs6000_emit_move_from_cr and call renamed function.
17076 (rs6000_emit_prologue): Call renamed functions.
17077 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17078 movesi_from_cr, remove volatile CRs.
17080 2017-08-30 Jon Beniston <jon@beniston.com>
17081 Richard Biener <rguenther@suse.de>
17083 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17084 of VECTOR_MODE_P check.
17085 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17086 element vector types.
17088 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17090 * df.h (df_read_modify_subreg_p): Remove in favor of...
17091 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17092 const_rtx instead of an rtx.
17093 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17094 * df-problems.c (df_word_lr_mark_ref): Likewise.
17095 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17096 (mark_pseudo_reg_dead): Likewise.
17097 (mark_ref_dead): Likewise.
17098 * reginfo.c (init_subregs_of_mode): Likewise.
17099 * sched-deps.c (sched_analyze_1): Likewise.
17100 * df-scan.c (df_def_record_1): Likewise.
17101 (df_uses_record): Likewise.
17102 (df_read_modify_subreg_p): Remove in favor of...
17103 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17104 const_rtx instead of an rtx.
17106 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17107 Alan Hayward <alan.hayward@arm.com>
17108 David Sherwood <david.sherwood@arm.com>
17110 * rtl.h (partial_subreg_p): New function.
17111 * caller-save.c (save_call_clobbered_regs): Use it.
17112 * calls.c (expand_call): Likewise.
17113 * combine.c (combinable_i3pat): Likewise.
17114 (simplify_set): Likewise.
17115 (make_extraction): Likewise.
17116 (make_compound_operation_int): Likewise.
17117 (gen_lowpart_or_truncate): Likewise.
17118 (force_to_mode): Likewise.
17119 (make_field_assignment): Likewise.
17120 (reg_truncated_to_mode): Likewise.
17121 (record_truncated_value): Likewise.
17122 (move_deaths): Likewise.
17123 * cse.c (record_jump_cond): Likewise.
17124 (cse_insn): Likewise.
17125 * cselib.c (cselib_lookup_1): Likewise.
17126 * expmed.c (extract_bit_field_using_extv): Likewise.
17127 * function.c (assign_parm_setup_reg): Likewise.
17128 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17129 * ira-build.c (create_insn_allocnos): Likewise.
17130 * lra-coalesce.c (merge_pseudos): Likewise.
17131 * lra-constraints.c (match_reload): Likewise.
17132 (simplify_operand_subreg): Likewise.
17133 (curr_insn_transform): Likewise.
17134 * lra-lives.c (process_bb_lives): Likewise.
17135 * lra.c (new_insn_reg): Likewise.
17136 (lra_substitute_pseudo): Likewise.
17137 * regcprop.c (mode_change_ok): Likewise.
17138 (maybe_mode_change): Likewise.
17139 (copyprop_hardreg_forward_1): Likewise.
17140 * reload.c (push_reload): Likewise.
17141 (find_reloads): Likewise.
17142 (find_reloads_subreg_address): Likewise.
17143 * reload1.c (alter_reg): Likewise.
17144 (eliminate_regs_1): Likewise.
17145 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17147 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17149 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17150 back to if statements, including unpack.
17152 2017-08-30 Martin Liska <mliska@suse.cz>
17154 PR inline-asm/82001
17155 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17157 (func_checker::compare_asm_inputs_outputs): ... this function.
17158 (func_checker::compare_gimple_asm): Use the function to compare
17159 also ASM constrains.
17160 * ipa-icf-gimple.h: Rename the function.
17162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17163 Alan Hayward <alan.hayward@arm.com>
17164 David Sherwood <david.sherwood@arm.com>
17166 * coretypes.h (complex_mode): New type.
17167 * gdbhooks.py (build_pretty_printer): Handle it.
17168 * machmode.h (complex_mode): New class.
17169 (complex_mode::includes_p): New function.
17170 (is_complex_int_mode): Likewise.
17171 (is_complex_float_mode): Likewise.
17172 * genmodes.c (get_mode_class): Handle complex mode classes.
17173 * function.c (expand_function_end): Use is_complex_int_mode.
17175 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17176 Alan Hayward <alan.hayward@arm.com>
17177 David Sherwood <david.sherwood@arm.com>
17179 * coretypes.h (scalar_mode_pod): New typedef.
17180 * gdbhooks.py (build_pretty_printer): Handle it.
17181 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17182 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17183 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17184 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17185 as_a <scalar_mode>.
17187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17188 Alan Hayward <alan.hayward@arm.com>
17189 David Sherwood <david.sherwood@arm.com>
17191 * machmode.h (mode_for_vector): Take a scalar_mode instead
17193 * stor-layout.c (mode_for_vector): Likewise.
17194 * explow.c (promote_mode): Use as_a <scalar_mode>.
17195 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17198 Alan Hayward <alan.hayward@arm.com>
17199 David Sherwood <david.sherwood@arm.com>
17201 * target.def (preferred_simd_mode): Take a scalar_mode
17202 instead of a machine_mode.
17203 * targhooks.h (default_preferred_simd_mode): Likewise.
17204 * targhooks.c (default_preferred_simd_mode): Likewise.
17205 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17206 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17207 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17208 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17209 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17210 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17211 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17212 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17214 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17215 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17216 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17217 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17218 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17219 * doc/tm.texi: Regenerate.
17220 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17224 Alan Hayward <alan.hayward@arm.com>
17225 David Sherwood <david.sherwood@arm.com>
17227 * target.def (scalar_mode_supported_p): Take a scalar_mode
17228 instead of a machine_mode.
17229 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17230 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17231 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17232 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17233 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17234 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17235 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17236 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17237 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17238 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17239 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17240 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17241 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17242 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17244 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17245 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17246 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17247 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17248 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17250 * doc/tm.texi: Regenerate.
17252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17253 Alan Hayward <alan.hayward@arm.com>
17254 David Sherwood <david.sherwood@arm.com>
17256 * coretypes.h (opt_scalar_mode): New typedef.
17257 * gdbhooks.py (build_pretty_printers): Handle it.
17258 * machmode.h (mode_iterator::get_2xwider): Add overload for
17260 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17262 * expr.c (convert_mode_scalar): Likewise.
17263 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17264 * optabs.c (expand_float): Likewise.
17265 (expand_fix): Likewise.
17266 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17269 Alan Hayward <alan.hayward@arm.com>
17270 David Sherwood <david.sherwood@arm.com>
17272 * optabs.c (expand_float): Explicitly check for scalars before
17273 using a branching expansion.
17274 (expand_fix): Likewise.
17276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17277 Alan Hayward <alan.hayward@arm.com>
17278 David Sherwood <david.sherwood@arm.com>
17280 * expr.c (convert_mode): Split scalar handling out into...
17281 (convert_mode_scalar): ...this new function. Treat the modes
17284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17285 Alan Hayward <alan.hayward@arm.com>
17286 David Sherwood <david.sherwood@arm.com>
17288 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17290 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17293 Alan Hayward <alan.hayward@arm.com>
17294 David Sherwood <david.sherwood@arm.com>
17296 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17297 rather than a machine_mode.
17298 (fixed_from_string): Likewise.
17299 (fixed_convert): Likewise.
17300 (fixed_convert_from_int): Likewise.
17301 (fixed_convert_from_real): Likewise.
17302 (real_convert_from_fixed): Likewise.
17303 * fixed-value.c (fixed_from_double_int): Likewise.
17304 (fixed_from_string): Likewise.
17305 (fixed_convert): Likewise.
17306 (fixed_convert_from_int): Likewise.
17307 (fixed_convert_from_real): Likewise.
17308 (real_convert_from_fixed): Likewise.
17309 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17312 Alan Hayward <alan.hayward@arm.com>
17313 David Sherwood <david.sherwood@arm.com>
17315 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17316 of separate mode class checks. Do not allow vector modes here.
17317 (immed_wide_int_const): Use as_a <scalar_mode>.
17318 * explow.c (trunc_int_for_mode): Likewise.
17319 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17320 (wi::shwi): Likewise.
17321 (wi::min_value): Likewise.
17322 (wi::max_value): Likewise.
17323 * dwarf2out.c (loc_descriptor): Likewise.
17324 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17325 for CONST_WIDE_INT.
17327 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17328 Alan Hayward <alan.hayward@arm.com>
17329 David Sherwood <david.sherwood@arm.com>
17331 * tree.h (SCALAR_TYPE_MODE): New macro.
17332 * expr.c (expand_expr_addr_expr_1): Use it.
17333 (expand_expr_real_2): Likewise.
17334 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17335 (fold_convert_const_fixed_from_int): Likewise.
17336 (fold_convert_const_fixed_from_real): Likewise.
17337 (native_encode_fixed): Likewise
17338 (native_encode_complex): Likewise
17339 (native_encode_vector): Likewise.
17340 (native_interpret_fixed): Likewise.
17341 (native_interpret_real): Likewise.
17342 (native_interpret_complex): Likewise.
17343 (native_interpret_vector): Likewise.
17344 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17345 (simd_clone_adjust_argument_types): Likewise.
17346 (simd_clone_init_simd_arrays): Likewise.
17347 (simd_clone_adjust): Likewise.
17348 * stor-layout.c (layout_type): Likewise.
17349 * tree.c (build_minus_one_cst): Likewise.
17350 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17351 * tree-inline.c (estimate_move_cost): Likewise.
17352 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17353 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17354 (vectorizable_reduction): Likewise.
17355 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17356 (vect_recog_mixed_size_cond_pattern): Likewise.
17357 (check_bool_pattern): Likewise.
17358 (adjust_bool_pattern): Likewise.
17359 (search_type_for_mask_1): Likewise.
17360 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17361 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17362 (vectorizable_load): Likewise.
17363 (vectorizable_store): Likewise.
17364 * ubsan.c (ubsan_encode_value): Likewise.
17365 * varasm.c (output_constant): Likewise.
17367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17368 Alan Hayward <alan.hayward@arm.com>
17369 David Sherwood <david.sherwood@arm.com>
17371 * coretypes.h (scalar_mode): New class.
17372 * machmode.h (scalar_mode): Likewise.
17373 (scalar_mode::includes_p): New function.
17374 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17375 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17376 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17377 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17378 * expmed.c (store_bit_field_1): Likewise.
17379 (extract_bit_field_1): Likewise.
17380 * expr.c (write_complex_part): Likewise.
17381 (read_complex_part): Likewise.
17382 (emit_move_complex_push): Likewise.
17383 (expand_expr_real_2): Likewise.
17384 * function.c (assign_parm_setup_reg): Likewise.
17385 (assign_parms_unsplit_complex): Likewise.
17386 * optabs.c (expand_binop): Likewise.
17387 * rtlanal.c (subreg_get_info): Likewise.
17388 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17389 * varasm.c (output_constant_pool_2): Likewise.
17391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17392 Alan Hayward <alan.hayward@arm.com>
17393 David Sherwood <david.sherwood@arm.com>
17395 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17397 (expmed_mult_highpart_optab): Likewise.
17398 (expmed_mult_highpart): Likewise.
17400 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17401 Alan Hayward <alan.hayward@arm.com>
17402 David Sherwood <david.sherwood@arm.com>
17404 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17405 instead of a machine_mode.
17406 (builtin_memset_read_str): Likewise.
17407 * builtins.c (c_readstr): Likewise.
17408 (builtin_memcpy_read_str): Likewise.
17409 (builtin_strncpy_read_str): Likewise.
17410 (builtin_memset_read_str): Likewise.
17411 (builtin_memset_gen_str): Likewise.
17412 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17413 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17414 instead of a machine_mode.
17415 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17417 (make_extraction): Likewise.
17418 (try_widen_shift_mode): Take and return scalar_int_modes instead
17420 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17421 a scalar_int_mode instead of a machine_mode.
17422 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17423 (avr_addr_space_pointer_mode): Likewise.
17424 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17425 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17426 (msp430_unwind_word_mode): Likewise.
17427 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17428 (spu_addr_space_pointer_mode): Likewise.
17429 (spu_addr_space_address_mode): Likewise.
17430 (spu_libgcc_cmp_return_mode): Likewise.
17431 (spu_libgcc_shift_count_mode): Likewise.
17432 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17433 (rl78_addr_space_pointer_mode): Likewise.
17434 (fl78_unwind_word_mode): Likewise.
17435 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17437 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17438 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17439 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17440 (mips_valid_pointer_mode): Likewise.
17441 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17442 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17443 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17445 (ft32_addr_space_address_mode): Likewise.
17446 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17447 scalar_int_mode instead of a machine_mode.
17448 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17450 (m32c_addr_space_address_mode): Likewise.
17451 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17452 (rs6000_eh_return_filter_mode): Likewise.
17453 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17454 (rs6000_eh_return_filter_mode): Likewise.
17455 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17456 (s390_libgcc_shift_count_mode): Likewise.
17457 (s390_unwind_word_mode): Likewise.
17458 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17460 * target.def (mode_rep_extended): Likewise.
17461 (valid_pointer_mode): Likewise.
17462 (addr_space.valid_pointer_mode): Likewise.
17463 (eh_return_filter_mode): Return a scalar_int_mode rather than
17465 (libgcc_cmp_return_mode): Likewise.
17466 (libgcc_shift_count_mode): Likewise.
17467 (unwind_word_mode): Likewise.
17468 (addr_space.pointer_mode): Likewise.
17469 (addr_space.address_mode): Likewise.
17470 * doc/tm.texi: Regenerate.
17471 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17473 (do_jump): Use scalar_int_mode for local variables.
17474 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17475 rather than a machine_mode.
17476 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17477 (scompare_loc_descriptor_wide): Likewise.
17478 (scompare_loc_descriptor_narrow): Likewise.
17479 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17481 * except.c (sjlj_emit_dispatch_table): Likewise.
17482 (expand_builtin_eh_copy_values): Likewise.
17483 * explow.c (convert_memory_address_addr_space_1): Likewise.
17484 Take a scalar_int_mode rather than a machine_mode.
17485 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17486 than a machine_mode.
17487 (memory_address_addr_space): Use scalar_int_mode for local variables.
17488 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17489 rather than a machine_mode.
17490 * expmed.c (mask_rtx): Likewise.
17491 (init_expmed_one_conv): Likewise.
17492 (expand_mult_highpart_adjust): Likewise.
17493 (extract_high_half): Likewise.
17494 (expmed_mult_highpart_optab): Likewise.
17495 (expmed_mult_highpart): Likewise.
17496 (expand_smod_pow2): Likewise.
17497 (expand_sdiv_pow2): Likewise.
17498 (emit_store_flag_int): Likewise.
17499 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17501 (extract_low_bits): Likewise.
17502 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17504 * expr.c (pieces_addr::adjust): Likewise.
17505 (can_store_by_pieces): Likewise.
17506 (store_by_pieces): Likewise.
17507 (clear_by_pieces_1): Likewise.
17508 (expand_expr_addr_expr_1): Likewise.
17509 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17510 (expand_expr_real_1): Likewise.
17511 (try_casesi): Likewise.
17512 * final.c (shorten_branches): Likewise.
17513 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17514 type of "mode" to machine_mode.
17515 * internal-fn.c (expand_arith_overflow_result_store): Take a
17516 scalar_int_mode rather than a machine_mode.
17517 (expand_mul_overflow): Use scalar_int_mode for local variables.
17518 * loop-doloop.c (doloop_modify): Likewise.
17519 (doloop_optimize): Likewise.
17520 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17521 than a machine_mode.
17522 (expand_doubleword_shift_condmove): Likewise.
17523 (expand_doubleword_shift): Likewise.
17524 (expand_doubleword_clz): Likewise.
17525 (expand_doubleword_popcount): Likewise.
17526 (expand_doubleword_parity): Likewise.
17527 (expand_absneg_bit): Use scalar_int_mode for local variables.
17528 (prepare_float_lib_cmp): Likewise.
17529 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17530 rather than a machine_mode.
17531 (convert_memory_address_addr_space): Likewise.
17532 (get_mode_bounds): Likewise.
17533 (get_address_mode): Return a scalar_int_mode rather than a
17535 * rtlanal.c (get_address_mode): Likewise.
17536 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17537 than a machine_mode.
17538 * targhooks.c (default_mode_rep_extended): Likewise.
17539 (default_valid_pointer_mode): Likewise.
17540 (default_addr_space_valid_pointer_mode): Likewise.
17541 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17542 than a machine_mode.
17543 (default_libgcc_cmp_return_mode): Likewise.
17544 (default_libgcc_shift_count_mode): Likewise.
17545 (default_unwind_word_mode): Likewise.
17546 (default_addr_space_pointer_mode): Likewise.
17547 (default_addr_space_address_mode): Likewise.
17548 * targhooks.h (default_eh_return_filter_mode): Likewise.
17549 (default_libgcc_cmp_return_mode): Likewise.
17550 (default_libgcc_shift_count_mode): Likewise.
17551 (default_unwind_word_mode): Likewise.
17552 (default_addr_space_pointer_mode): Likewise.
17553 (default_addr_space_address_mode): Likewise.
17554 (default_mode_rep_extended): Take a scalar_int_mode rather than
17556 (default_valid_pointer_mode): Likewise.
17557 (default_addr_space_valid_pointer_mode): Likewise.
17558 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17560 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17561 rather than a machine_mode.
17562 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17563 for local variables.
17564 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17565 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17566 than a machine_mode.
17568 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17569 Alan Hayward <alan.hayward@arm.com>
17570 David Sherwood <david.sherwood@arm.com>
17572 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17573 the mode argument to scalar_int_mode.
17574 (do_jump_by_parts_zero_rtx): Likewise.
17575 (do_jump_by_parts_equality_rtx): Likewise.
17576 (do_jump_by_parts_greater): Take a mode argument.
17577 (do_jump_by_parts_equality): Likewise.
17578 (do_jump_1): Update calls accordingly.
17580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17581 Alan Hayward <alan.hayward@arm.com>
17582 David Sherwood <david.sherwood@arm.com>
17584 * is-a.h (safe_dyn_cast): New function.
17585 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17586 (jump_table_for_label): Likewise.
17587 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17588 instead of an rtx_insn *.
17589 (shorten_branches): Use dyn_cast instead of LABEL_P and
17590 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17591 rtx_jump_table_data::get_data_mode.
17592 (final_scan_insn): Likewise.
17594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17595 Alan Hayward <alan.hayward@arm.com>
17596 David Sherwood <david.sherwood@arm.com>
17598 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17599 trying to combine a full-register integer set with a subreg
17602 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17603 Alan Hayward <alan.hayward@arm.com>
17604 David Sherwood <david.sherwood@arm.com>
17606 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17607 that is always either address_mode or pointer_mode.
17609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17610 Alan Hayward <alan.hayward@arm.com>
17611 David Sherwood <david.sherwood@arm.com>
17613 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17614 when the two are known to be equal.
17616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17617 Alan Hayward <alan.hayward@arm.com>
17618 David Sherwood <david.sherwood@arm.com>
17620 * simplify-rtx.c (simplify_const_unary_operation): Use
17621 is_a <scalar_int_mode> instead of checking for a nonzero
17622 precision. Forcibly convert op_mode to a scalar_int_mode
17623 in that case. More clearly differentiate the operand and
17624 result modes and use the former when deciding what the value
17625 of a count-bits operation should be. Use is_int_mode instead
17626 of checking for a MODE_INT. Remove redundant check for whether
17627 this mode has a zero precision.
17629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17630 Alan Hayward <alan.hayward@arm.com>
17631 David Sherwood <david.sherwood@arm.com>
17633 * optabs.c (widen_leading): Change the type of the mode argument
17634 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17635 (widen_bswap): Likewise.
17636 (expand_parity): Likewise.
17637 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17638 (expand_ffs): Likewise.
17639 (epand_unop): Check for scalar integer modes before calling the
17642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17643 Alan Hayward <alan.hayward@arm.com>
17644 David Sherwood <david.sherwood@arm.com>
17646 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17648 (expand_expr_real_1): Update call accordingly.
17650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17651 Alan Hayward <alan.hayward@arm.com>
17652 David Sherwood <david.sherwood@arm.com>
17654 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17655 value_mode arguments. Use scalar_int_mode internally.
17656 (store_bit_field_1): Rename the new integer mode from imode
17657 to op0_mode and use it instead of GET_MODE (op0). Update calls
17658 to store_split_bit_field, store_bit_field_using_insv and
17659 store_fixed_bit_field.
17660 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17661 Use scalar_int_mode internally. Use a bit count rather than a mode
17662 when calculating the largest bit size for get_best_mode.
17663 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17664 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17665 Remove assertion that OP0 has a scalar integer mode.
17666 (store_split_bit_field): Add op0_mode and value_mode arguments.
17667 Update calls to extract_fixed_bit_field.
17668 (extract_bit_field_using_extv): Add an op0_mode argument.
17669 Use scalar_int_mode internally.
17670 (extract_bit_field_1): Rename the new integer mode from imode to
17671 op0_mode and use it instead of GET_MODE (op0). Update calls to
17672 extract_split_bit_field, extract_bit_field_using_extv and
17673 extract_fixed_bit_field.
17674 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17675 to extract_split_bit_field and extract_fixed_bit_field_1.
17676 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17677 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17678 on the target mode.
17679 (extract_split_bit_field): Add an op0_mode argument. Update call
17680 to extract_fixed_bit_field.
17682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17683 Alan Hayward <alan.hayward@arm.com>
17684 David Sherwood <david.sherwood@arm.com>
17686 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17687 * explow.c (hard_function_value): Likewise.
17688 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17689 convert_to_mode call outside the loop.
17690 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17691 for the mode iterator. Require the mode specified by max_pieces
17693 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17695 (copy_blkmode_to_reg): Likewise.
17696 (set_storage_via_setmem): Likewise.
17697 * optabs.c (prepare_cmp_insn): Likewise.
17698 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17699 * stor-layout.c (finish_bitfield_representative): Likewise.
17701 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17702 Alan Hayward <alan.hayward@arm.com>
17703 David Sherwood <david.sherwood@arm.com>
17705 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17706 * expr.c (convert_move): Use them.
17707 (convert_modes): Likewise.
17708 (store_expr_with_bounds): Likewise.
17710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17711 Alan Hayward <alan.hayward@arm.com>
17712 David Sherwood <david.sherwood@arm.com>
17714 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17715 parameter for the mode of "x". Remove the "known_x", "known_mode"
17716 and "known_ret" arguments. Change the type of the mode argument
17717 to scalar_int_mode.
17718 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17719 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17720 (reg_num_sign_bit_copies_for_combine): Likewise.
17721 * rtlanal.c (nonzero_bits1): Likewise.
17722 (num_sign_bit_copies1): Likewise.
17723 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17724 (reg_num_sign_bit_copies_general): Likewise.
17725 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17726 (reg_nonzero_bits_general): Likewise.
17728 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17729 Alan Hayward <alan.hayward@arm.com>
17730 David Sherwood <david.sherwood@arm.com>
17732 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17733 than in subroutines. Return 1 for non-integer modes.
17734 (cached_num_sign_bit_copies): Change the type of the mode parameter
17735 to scalar_int_mode.
17736 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17737 classes. Handle CONST_INT_P first and then check whether X also
17738 has a scalar integer mode. Check the same thing for inner registers
17739 of a SUBREG and for values that are being extended or truncated.
17741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17742 Alan Hayward <alan.hayward@arm.com>
17743 David Sherwood <david.sherwood@arm.com>
17745 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17746 in subroutines. Return the mode mask for non-integer modes.
17747 (cached_nonzero_bits): Change the type of the mode parameter
17748 to scalar_int_mode.
17749 (nonzero_bits1): Likewise. Remove early exit for other mode
17750 classes. Handle CONST_INT_P first and then check whether X
17751 also has a scalar integer mode.
17753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17754 Alan Hayward <alan.hayward@arm.com>
17755 David Sherwood <david.sherwood@arm.com>
17757 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17758 Return a scalar_int_mode and assert that the size is greater than
17760 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17761 (op_by_pieces_d::op_by_pieces_d): Likewise.
17762 (op_by_pieces_d::run): Likewise.
17763 (can_store_by_pieces): Likewise.
17765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17766 Alan Hayward <alan.hayward@arm.com>
17767 David Sherwood <david.sherwood@arm.com>
17769 * combine.c (extract_left_shift): Add a mode argument and update
17771 (make_compound_operation_int): Change the type of the mode parameter
17772 to scalar_int_mode and update the call to extract_left_shift.
17774 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17775 Alan Hayward <alan.hayward@arm.com>
17776 David Sherwood <david.sherwood@arm.com>
17778 * combine.c (simplify_and_const_int): Change the type of the mode
17779 parameter to scalar_int_mode.
17780 (simplify_and_const_int_1): Likewise. Update recursive call.
17782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17783 Alan Hayward <alan.hayward@arm.com>
17784 David Sherwood <david.sherwood@arm.com>
17786 * combine.c (simplify_compare_const): Check that the mode is a
17787 scalar_int_mode (rather than VOIDmode) before testing its
17789 (simplify_comparison): Move COMPARISON_P handling out of the
17790 loop and restrict the latter part of the loop to scalar_int_modes.
17791 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17792 and when considering SUBREG_REGs. Use is_int_mode instead of
17793 checking GET_MODE_CLASS against MODE_INT.
17795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17796 Alan Hayward <alan.hayward@arm.com>
17797 David Sherwood <david.sherwood@arm.com>
17799 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17800 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17801 shift_unit_mode and for modes involved in scalar shifts.
17803 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17804 Alan Hayward <alan.hayward@arm.com>
17805 David Sherwood <david.sherwood@arm.com>
17807 * combine.c (force_int_to_mode): New function, split out from...
17808 (force_to_mode): ...here. Keep xmode up-to-date and use it
17809 instead of GET_MODE (x).
17811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17812 Alan Hayward <alan.hayward@arm.com>
17813 David Sherwood <david.sherwood@arm.com>
17815 * optabs-query.h (extraction_insn::struct_mode): Change type to
17816 opt_scalar_int_mode and update comment.
17817 (extraction_insn::field_mode): Change type to scalar_int_mode.
17818 (extraction_insn::pos_mode): Likewise.
17819 * combine.c (make_extraction): Update accordingly.
17820 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17821 (get_optab_extraction_insn): Likewise.
17822 * recog.c (simplify_while_replacing): Likewise.
17823 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17824 parameter to opt_scalar_int_mode.
17826 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17827 Alan Hayward <alan.hayward@arm.com>
17828 David Sherwood <david.sherwood@arm.com>
17830 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17831 to a scalar_int_mode instead of a machine_mode.
17832 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17833 (get_best_mode): Return a boolean and use a pointer argument to store
17834 the selected mode. Replace the limit mode parameter with a bit limit.
17835 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17836 for the values returned by bit_field_mode_iterator::next_mode.
17837 (store_bit_field): Update call to get_best_mode.
17838 (store_fixed_bit_field): Likewise.
17839 (extract_fixed_bit_field): Likewise.
17840 * expr.c (optimize_bitfield_assignment_op): Likewise.
17841 * fold-const.c (optimize_bit_field_compare): Likewise.
17842 (fold_truth_andor_1): Likewise.
17843 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17844 Update for new type of m_mode.
17845 (get_best_mode): As above.
17847 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17848 Alan Hayward <alan.hayward@arm.com>
17849 David Sherwood <david.sherwood@arm.com>
17851 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
17852 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
17853 (store_bit_field): Check is_a <scalar_int_mode> before calling
17854 strict_volatile_bitfield_p.
17855 (extract_bit_field): Likewise.
17857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17858 Alan Hayward <alan.hayward@arm.com>
17859 David Sherwood <david.sherwood@arm.com>
17861 * target.def (cstore_mode): Return a scalar_int_mode.
17862 * doc/tm.texi: Regenerate.
17863 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
17864 * targhooks.h (default_cstore_mode): Likewise.
17865 * targhooks.c (default_cstore_mode): Likewise, using a forced
17867 * expmed.c (emit_cstore): Expect the target of the cstore to be
17870 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17871 Alan Hayward <alan.hayward@arm.com>
17872 David Sherwood <david.sherwood@arm.com>
17874 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
17876 (niter_desc): Likewise mode.
17877 (iv_analyze): Add a mode parameter.
17879 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
17880 and change its type to scalar_int_mode.
17881 * loop-iv.c: Update commentary at head of file.
17882 (iv_constant): Pass the mode paraeter before the rtx it describes
17883 and change its type to scalar_int_mode. Remove VOIDmode handling.
17884 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
17885 (iv_extend): Likewise.
17886 (shorten_into_mode): Likewise.
17887 (iv_add): Use scalar_int_mode.
17888 (iv_mult): Likewise.
17889 (iv_shift): Likewise.
17890 (canonicalize_iv_subregs): Likewise.
17891 (get_biv_step_1): Pass the outer_mode parameter before the rtx
17892 it describes and change its mode to scalar_int_mode. Also change
17893 the type of the returned inner_mode to scalar_int_mode.
17894 (get_biv_step): Likewise, turning outer_mode from a pointer
17895 into a direct parameter. Update call to get_biv_step_1.
17896 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
17897 iv_constant and get_biv_step.
17898 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
17899 and change its type to scalar_int_mode. Don't initialise iv->mode
17900 to VOIDmode and remove later checks for its still being VOIDmode.
17901 Update calls to iv_analyze_op and iv_analyze_expr. Check
17902 is_a <scalar_int_mode> when changing the mode under consideration.
17903 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
17904 Update call to iv_analyze_expr.
17905 (iv_analyze_op): Add a mode parameter. Reject subregs whose
17906 inner register is not also a scalar_int_mode. Update call to
17908 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
17909 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
17910 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
17911 separate mode class checks. Update calls to iv_analyze. Remove
17912 fix-up of VOIDmodes after iv_analyze_biv.
17913 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
17914 don't have a scalar_int_mode. Update call to biv_p.
17916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17917 Alan Hayward <alan.hayward@arm.com>
17918 David Sherwood <david.sherwood@arm.com>
17920 * cfgexpand.c (convert_debug_memory_address): Use
17921 as_a <scalar_int_mode>.
17922 * combine.c (expand_compound_operation): Likewise.
17923 (make_extraction): Likewise.
17924 (change_zero_ext): Likewise.
17925 (simplify_comparison): Likewise.
17926 * cse.c (cse_insn): Likewise.
17927 * dwarf2out.c (minmax_loc_descriptor): Likewise.
17928 (mem_loc_descriptor): Likewise.
17929 (loc_descriptor): Likewise.
17930 * expmed.c (init_expmed_one_mode): Likewise.
17931 (synth_mult): Likewise.
17932 (emit_store_flag_1): Likewise.
17933 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
17934 of a comparison with size.
17935 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
17936 (reduce_to_bit_field_precision): Likewise.
17937 * function.c (expand_function_end): Likewise.
17938 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
17939 * loop-doloop.c (doloop_modify): Likewise.
17940 * optabs.c (expand_binop): Likewise.
17941 (expand_unop): Likewise.
17942 (expand_copysign_absneg): Likewise.
17943 (prepare_cmp_insn): Likewise.
17944 (maybe_legitimize_operand): Likewise.
17945 * recog.c (const_scalar_int_operand): Likewise.
17946 * rtlanal.c (get_address_mode): Likewise.
17947 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17948 (simplify_cond_clz_ctz): Likewise.
17949 * tree-nested.c (get_nl_goto_field): Likewise.
17950 * tree.c (build_vector_type_for_mode): Likewise.
17951 * var-tracking.c (use_narrower_mode): Likewise.
17953 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17954 Alan Hayward <alan.hayward@arm.com>
17955 David Sherwood <david.sherwood@arm.com>
17957 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
17958 * builtins.c (expand_builtin_signbit): Use it.
17959 * cfgexpand.c (expand_debug_expr): Likewise.
17960 * dojump.c (do_jump): Likewise.
17961 (do_compare_and_jump): Likewise.
17962 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
17963 * expmed.c (make_tree): Likewise.
17964 * expr.c (expand_expr_real_2): Likewise.
17965 (expand_expr_real_1): Likewise.
17966 (try_casesi): Likewise.
17967 * fold-const-call.c (fold_const_call_ss): Likewise.
17968 * fold-const.c (unextend): Likewise.
17969 (extract_muldiv_1): Likewise.
17970 (fold_single_bit_test): Likewise.
17971 (native_encode_int): Likewise.
17972 (native_encode_string): Likewise.
17973 (native_interpret_int): Likewise.
17974 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
17975 * internal-fn.c (expand_addsub_overflow): Likewise.
17976 (expand_neg_overflow): Likewise.
17977 (expand_mul_overflow): Likewise.
17978 (expand_arith_overflow): Likewise.
17979 * match.pd: Likewise.
17980 * stor-layout.c (layout_type): Likewise.
17981 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17982 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
17983 * tree-ssanames.c (get_range_info): Likewise.
17984 * tree-switch-conversion.c (array_value_type) Likewise.
17985 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
17986 (vect_recog_divmod_pattern): Likewise.
17987 (vect_recog_mixed_size_cond_pattern): Likewise.
17988 * tree-vrp.c (extract_range_basic): Likewise.
17989 (simplify_float_conversion_using_ranges): Likewise.
17990 * tree.c (int_fits_type_p): Likewise.
17991 * ubsan.c (instrument_bool_enum_load): Likewise.
17992 * varasm.c (mergeable_string_section): Likewise.
17993 (narrowing_initializer_constant_valid_p): Likewise.
17994 (output_constant): Likewise.
17996 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17997 Alan Hayward <alan.hayward@arm.com>
17998 David Sherwood <david.sherwood@arm.com>
18000 * machmode.h (NARROWEST_INT_MODE): New macro.
18001 * expr.c (alignment_for_piecewise_move): Use it instead of
18002 GET_CLASS_NARROWEST_MODE (MODE_INT).
18003 (push_block): Likewise.
18004 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18006 * tree-vrp.c (simplify_float_conversion_using_ranges): 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 * postreload.c (move2add_valid_value_p): Change the type of the
18013 mode parameter to scalar_int_mode.
18014 (move2add_use_add2_insn): Add a mode parameter and use it instead
18016 (move2add_use_add3_insn): Likewise.
18017 (reload_cse_move2add): Update accordingly.
18019 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18020 Alan Hayward <alan.hayward@arm.com>
18021 David Sherwood <david.sherwood@arm.com>
18023 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18025 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18026 * optabs.c (expand_unop): Likewise.
18028 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18029 Alan Hayward <alan.hayward@arm.com>
18030 David Sherwood <david.sherwood@arm.com>
18032 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18033 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18034 (popcount_loc_descriptor): Likewise.
18035 (bswap_loc_descriptor): Likewise.
18036 (rotate_loc_descriptor): Likewise.
18037 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18038 calling the functions above.
18040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18041 Alan Hayward <alan.hayward@arm.com>
18042 David Sherwood <david.sherwood@arm.com>
18044 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18046 (try_combine): Likewise.
18047 (simplify_if_then_else): Likewise.
18048 * cse.c (cse_insn): Likewise.
18049 * dwarf2out.c (mem_loc_descriptor): Likewise.
18050 * emit-rtl.c (gen_lowpart_common): Likewise.
18051 * simplify-rtx.c (simplify_truncation): Likewise.
18052 (simplify_binary_operation_1): Likewise.
18053 (simplify_const_relational_operation): Likewise.
18054 (simplify_ternary_operation): Likewise.
18055 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18058 Alan Hayward <alan.hayward@arm.com>
18059 David Sherwood <david.sherwood@arm.com>
18061 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18062 * reload.c (push_reload): Likewise.
18063 (find_reloads): Likewise.
18065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18066 Alan Hayward <alan.hayward@arm.com>
18067 David Sherwood <david.sherwood@arm.com>
18069 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18070 (make_compound_operation_int): Likewise.
18071 (change_zero_ext): Likewise.
18072 * expr.c (convert_move): Likewise.
18073 (convert_modes): Likewise.
18074 * fwprop.c (forward_propagate_subreg): Likewise.
18075 * loop-iv.c (get_biv_step_1): Likewise.
18076 * optabs.c (widen_operand): Likewise.
18077 * postreload.c (move2add_valid_value_p): Likewise.
18078 * recog.c (simplify_while_replacing): Likewise.
18079 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18080 (simplify_binary_operation_1): Likewise. Remove redundant
18081 mode equality check.
18083 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18084 Alan Hayward <alan.hayward@arm.com>
18085 David Sherwood <david.sherwood@arm.com>
18087 * combine.c (combine_simplify_rtx): Add checks for
18088 is_a <scalar_int_mode>.
18089 (simplify_if_then_else): Likewise.
18090 (make_field_assignment): Likewise.
18091 (simplify_comparison): Likewise.
18092 * ifcvt.c (noce_try_bitop): Likewise.
18093 * loop-invariant.c (canonicalize_address_mult): Likewise.
18094 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18096 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18097 Alan Hayward <alan.hayward@arm.com>
18098 David Sherwood <david.sherwood@arm.com>
18100 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18101 is_a <scalar_int_mode> instead of != BLKmode.
18103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18104 Alan Hayward <alan.hayward@arm.com>
18105 David Sherwood <david.sherwood@arm.com>
18107 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18108 instead of != VOIDmode.
18109 * combine.c (if_then_else_cond): Likewise.
18110 (change_zero_ext): Likewise.
18111 * dwarf2out.c (mem_loc_descriptor): Likewise.
18112 (loc_descriptor): Likewise.
18113 * rtlanal.c (canonicalize_condition): Likewise.
18114 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18117 Alan Hayward <alan.hayward@arm.com>
18118 David Sherwood <david.sherwood@arm.com>
18120 * simplify-rtx.c (simplify_binary_operation_1): Use
18121 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18124 Alan Hayward <alan.hayward@arm.com>
18125 David Sherwood <david.sherwood@arm.com>
18127 * wide-int.h (int_traits<unsigned char>) New class.
18128 (int_traits<unsigned short>) Likewise.
18129 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18130 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18132 * combine.c (set_nonzero_bits_and_sign_copies): Use
18133 is_a <scalar_int_mode>.
18134 (find_split_point): Likewise.
18135 (combine_simplify_rtx): Likewise.
18136 (simplify_logical): Likewise.
18137 (expand_compound_operation): Likewise.
18138 (expand_field_assignment): Likewise.
18139 (make_compound_operation): Likewise.
18140 (extended_count): Likewise.
18141 (change_zero_ext): Likewise.
18142 (simplify_comparison): Likewise.
18143 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18144 (ucompare_loc_descriptor): Likewise.
18145 (minmax_loc_descriptor): Likewise.
18146 (mem_loc_descriptor): Likewise.
18147 (loc_descriptor): Likewise.
18148 * expmed.c (init_expmed_one_mode): Likewise.
18149 * lra-constraints.c (lra_constraint_offset): Likewise.
18150 * optabs.c (prepare_libcall_arg): Likewise.
18151 * postreload.c (move2add_note_store): Likewise.
18152 * reload.c (operands_match_p): Likewise.
18153 * rtl.h (load_extend_op): Likewise.
18154 * rtlhooks.c (gen_lowpart_general): Likewise.
18155 * simplify-rtx.c (simplify_truncation): Likewise.
18156 (simplify_unary_operation_1): Likewise.
18157 (simplify_binary_operation_1): Likewise.
18158 (simplify_const_binary_operation): Likewise.
18159 (simplify_const_relational_operation): Likewise.
18160 (simplify_subreg): Likewise.
18161 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18162 * var-tracking.c (adjust_mems): Likewise.
18163 (prepare_call_arguments): Likewise.
18165 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18166 Alan Hayward <alan.hayward@arm.com>
18167 David Sherwood <david.sherwood@arm.com>
18169 * machmode.h (is_int_mode): New fuction.
18170 * combine.c (find_split_point): Use it.
18171 (combine_simplify_rtx): Likewise.
18172 (simplify_if_then_else): Likewise.
18173 (simplify_set): Likewise.
18174 (simplify_shift_const_1): Likewise.
18175 (simplify_comparison): Likewise.
18176 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18177 * cse.c (notreg_cost): Likewise.
18178 (cse_insn): Likewise.
18179 * cselib.c (cselib_lookup_1): Likewise.
18180 * dojump.c (do_jump_1): Likewise.
18181 (do_compare_rtx_and_jump): Likewise.
18182 * dse.c (get_call_args): Likewise.
18183 * dwarf2out.c (rtl_for_decl_init): Likewise.
18184 (native_encode_initializer): Likewise.
18185 * expmed.c (emit_store_flag_1): Likewise.
18186 (emit_store_flag): Likewise.
18187 * expr.c (convert_modes): Likewise.
18188 (store_field): Likewise.
18189 (expand_expr_real_1): Likewise.
18190 * fold-const.c (fold_read_from_constant_string): Likewise.
18191 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18192 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18193 * optabs.c (expand_binop): Likewise.
18194 (expand_unop): Likewise.
18195 (expand_abs_nojump): Likewise.
18196 (expand_one_cmpl_abs_nojump): Likewise.
18197 * simplify-rtx.c (mode_signbit_p): Likewise.
18198 (val_signbit_p): Likewise.
18199 (val_signbit_known_set_p): Likewise.
18200 (val_signbit_known_clear_p): Likewise.
18201 (simplify_relational_operation_1): Likewise.
18202 * tree.c (vector_type_mode): Likewise.
18204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18205 Alan Hayward <alan.hayward@arm.com>
18206 David Sherwood <david.sherwood@arm.com>
18208 * machmode.h (smallest_mode_for_size): Fix formatting.
18209 (smallest_int_mode_for_size): New function.
18210 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18211 instead of smallest_mode_for_size.
18212 * combine.c (make_extraction): Likewise.
18213 * config/arc/arc.c (arc_expand_movmem): Likewise.
18214 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18215 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18216 * config/s390/s390.c (s390_expand_insv): Likewise.
18217 * config/sparc/sparc.c (assign_int_registers): Likewise.
18218 * config/spu/spu.c (spu_function_value): Likewise.
18219 (spu_function_arg): Likewise.
18220 * coverage.c (get_gcov_type): Likewise.
18221 (get_gcov_unsigned_t): Likewise.
18222 * dse.c (find_shift_sequence): Likewise.
18223 * expmed.c (store_bit_field_1): Likewise.
18224 * expr.c (convert_move): Likewise.
18225 (store_field): Likewise.
18226 * internal-fn.c (expand_arith_overflow): Likewise.
18227 * optabs-query.c (get_best_extraction_insn): Likewise.
18228 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18229 * stor-layout.c (layout_type): Likewise.
18230 (initialize_sizetypes): Likewise.
18231 * targhooks.c (default_get_mask_mode): Likewise.
18232 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18235 Alan Hayward <alan.hayward@arm.com>
18236 David Sherwood <david.sherwood@arm.com>
18238 * machmode.h (opt_mode::else_blk): New function.
18239 (int_mode_for_mode): Declare.
18240 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18241 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18243 * cfgexpand.c (expand_debug_expr): Likewise.
18244 * combine.c (gen_lowpart_or_truncate): Likewise.
18245 (gen_lowpart_for_combine): Likewise.
18246 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18247 * config/avr/avr.c (avr_to_int_mode): Likewise.
18248 (avr_out_plus_1): Likewise.
18249 (avr_out_plus): Likewise.
18250 (avr_out_round): Likewise.
18251 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18252 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18253 (s390_expand_vcond): Likewise.
18254 * config/spu/spu.c (spu_split_immediate): Likewise.
18255 (spu_expand_mov): Likewise.
18256 * dse.c (get_stored_val): Likewise.
18257 * expmed.c (store_bit_field_1): Likewise.
18258 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18260 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18261 (extract_low_bits): Likewise.
18262 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18263 handling rather than repeating the check.
18264 (emit_group_store): Likewise.
18265 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18266 * optabs.c (expand_absneg_bit): Likewise.
18267 (expand_copysign_absneg): Likewise.
18268 (expand_copysign_bit): Likewise.
18269 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18270 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18271 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18272 * var-tracking.c (prepare_call_arguments): Likewise.
18273 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18274 int_mode_for_mode instead of mode_for_size.
18275 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18278 Alan Hayward <alan.hayward@arm.com>
18279 David Sherwood <david.sherwood@arm.com>
18281 * machmode.h (int_mode_for_size): New function.
18282 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18283 instead of mode_for_size.
18284 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18286 * combine.c (expand_field_assignment): Use int_mode_for_size
18287 instead of mode_for_size.
18288 (make_extraction): Likewise.
18289 (simplify_shift_const_1): Likewise.
18290 (simplify_comparison): Likewise.
18291 * dojump.c (do_jump): Likewise.
18292 * dwarf2out.c (mem_loc_descriptor): Likewise.
18293 * emit-rtl.c (init_derived_machine_modes): Likewise.
18294 * expmed.c (flip_storage_order): Likewise.
18295 (convert_extracted_bit_field): Likewise.
18296 * expr.c (copy_blkmode_from_reg): Likewise.
18297 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18298 * internal-fn.c (expand_mul_overflow): Likewise.
18299 * lower-subreg.c (simple_move): Likewise.
18300 * optabs-libfuncs.c (init_optabs): Likewise.
18301 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18302 * tree.c (vector_type_mode): Likewise.
18303 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18304 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18305 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18306 * tree-vect-stmts.c (vectorizable_load): Likewise.
18307 (vectorizable_store): Likewise.
18309 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18310 Alan Hayward <alan.hayward@arm.com>
18311 David Sherwood <david.sherwood@arm.com>
18313 * coretypes.h (pod_mode): New type.
18314 (scalar_int_mode_pod): New typedef.
18315 * machmode.h (pod_mode): New class.
18316 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18317 * genmodes.c (emit_mode_int_n): Update accordingly.
18318 * lower-subreg.h (target_lower_subreg): Change type to
18319 scalar_int_mode_pod.
18320 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18321 scalar_int_mode_pod.
18323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18324 Alan Hayward <alan.hayward@arm.com>
18325 David Sherwood <david.sherwood@arm.com>
18327 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18328 machine_mode to scalar_int_mode.
18329 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18330 (rs6000_option_override_internal): Remove cast to int.
18331 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18332 machine_mode to scalar_int_mode.
18333 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18334 (rs6000_option_override_internal): Remove cast to int.
18335 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18336 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18338 * config/s390/s390.c (s390_expand_builtin): Likewise.
18339 * coretypes.h (scalar_int_mode): New type.
18340 (opt_scalar_int_mode): New typedef.
18341 * machmode.h (scalar_int_mode): New class.
18342 (scalar_int_mode::includes_p): New function.
18343 (byte_mode): Change type to scalar_int_mode.
18344 (word_mode): Likewise.
18345 (ptr_mode): Likewise.
18346 * emit-rtl.c (byte_mode): Likewise.
18347 (word_mode): Likewise.
18348 (ptr_mode): Likewise.
18349 (init_derived_machine_modes): Update accordingly.
18350 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18351 and MODE_PARTIAL_INT.
18352 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18353 opt_scalar_int_mode.
18355 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18356 Alan Hayward <alan.hayward@arm.com>
18357 David Sherwood <david.sherwood@arm.com>
18359 * target.def (libgcc_floating_mode_supported_p): Take a
18361 * doc/tm.texi: Regenerate.
18362 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18364 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18365 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18368 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18369 Alan Hayward <alan.hayward@arm.com>
18370 David Sherwood <david.sherwood@arm.com>
18372 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18373 * doc/tm.texi: Regenerate.
18374 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18375 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18376 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18377 * targhooks.h (default_floatn_mode): Likewise.
18378 * targhooks.c (default_floatn_mode): Likewise.
18379 * tree.c (build_common_tree_nodes): Update accordingly.
18381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18382 Alan Hayward <alan.hayward@arm.com>
18383 David Sherwood <david.sherwood@arm.com>
18385 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18386 (mode_iterator::iterate_p): Likewise.
18387 (mode_iterator::get_wider): Likewise.
18388 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18391 Alan Hayward <alan.hayward@arm.com>
18392 David Sherwood <david.sherwood@arm.com>
18394 * coretypes.h (opt_scalar_float_mode): New typedef.
18395 * machmode.h (float_mode_for_size): New function.
18396 * emit-rtl.c (double_mode): Delete.
18397 (init_emit_once): Use float_mode_for_size.
18398 * stor-layout.c (layout_type): Likewise.
18399 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18402 Alan Hayward <alan.hayward@arm.com>
18403 David Sherwood <david.sherwood@arm.com>
18405 * output.h (assemble_real): Take a scalar_float_mode.
18406 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18407 * config/arm/arm.md (consttable_4): Likewise.
18408 (consttable_8): Likewise.
18409 (consttable_16): Likewise.
18410 * config/mips/mips.md (consttable_float): Likewise.
18411 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18412 * varasm.c (assemble_real): Take a scalar_float_mode.
18413 (output_constant_pool_2): Update accordingly.
18414 (output_constant): Likewise.
18416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18417 Alan Hayward <alan.hayward@arm.com>
18418 David Sherwood <david.sherwood@arm.com>
18420 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18421 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18422 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18423 (native_encode_real): Likewise.
18424 (native_interpret_real): Likewise.
18425 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18426 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18429 Alan Hayward <alan.hayward@arm.com>
18430 David Sherwood <david.sherwood@arm.com>
18432 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18433 <scalar_float_mode>. Simplify.
18434 (gen_extend_conv_libfunc): Likewise.
18436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18437 Alan Hayward <alan.hayward@arm.com>
18438 David Sherwood <david.sherwood@arm.com>
18440 * coretypes.h (scalar_float_mode): New type.
18441 * machmode.h (mode_traits::from_int): Use machine_mode if
18442 USE_ENUM_MODES is defined.
18443 (is_a): New function.
18445 (dyn_cast): Likewise.
18446 (scalar_float_mode): New class.
18447 (scalar_float_mode::includes_p): New function.
18448 (is_float_mode): Likewise.
18449 * gdbhooks.py (MachineModePrinter): New class.
18450 (build_pretty_printer): Use it for scalar_float_mode.
18451 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18452 (format_helper::format_helper): Turn into a template.
18453 * genmodes.c (get_mode_class): New function.
18454 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18455 or machine_mode if none.
18456 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18457 as_a <scalar_float_mode>.
18458 * dwarf2out.c (mem_loc_descriptor): Likewise.
18459 (insert_float): Likewise.
18460 (add_const_value_attribute): Likewise.
18461 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18462 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18463 (expand_unop): Update accordingly.
18464 (expand_abs_nojump): Likewise.
18465 (expand_copysign_absneg): Take a scalar_float_mode.
18466 (expand_copysign_bit): Likewise.
18467 (expand_copysign): Update accordingly.
18469 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18470 Alan Hayward <alan.hayward@arm.com>
18471 David Sherwood <david.sherwood@arm.com>
18473 * coretypes.h (opt_mode): New class.
18474 * machmode.h (opt_mode): Likewise.
18475 (opt_mode::else_void): New function.
18476 (opt_mode::require): Likewise.
18477 (opt_mode::exists): Likewise.
18478 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18479 (GET_MODE_2XWIDER_MODE): Likewise.
18480 (mode_iterator::get_wider): Update accordingly.
18481 (mode_iterator::get_2xwider): Likewise.
18482 (mode_iterator::get_known_wider): Likewise, turning into a template.
18483 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18484 forcing a wider mode to exist.
18485 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18486 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18487 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18488 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18489 * lower-subreg.c (init_lower_subreg): Likewise.
18490 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18491 on the final iteration.
18492 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18493 a wider mode exists before asking for a move pattern.
18494 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18495 forcing a wider mode to exist.
18496 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18497 returning false if no such mode exists.
18498 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18499 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18500 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18501 Avoid checking for a MODE_INT if we already know the mode is not a
18503 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18504 forcing a wider mode to exist.
18505 (expmed_mult_highpart_optab): Likewise.
18506 (expmed_mult_highpart): Likewise.
18507 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18509 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18510 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18511 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18512 * internal-fn.c (expand_mul_overflow): Update use of
18513 GET_MODE_2XWIDER_MODE.
18514 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18515 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18516 GET_MODE_WIDER_MODE.
18517 (convert_plusminus_to_widen): Likewise.
18518 * tree-switch-conversion.c (array_value_type): Likewise.
18519 * var-tracking.c (emit_note_insn_var_location): Likewise.
18520 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18521 Return false inside rather than outside the loop if no wider mode
18523 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18524 and GET_MODE_2XWIDER_MODE
18525 (can_compare_p): Use else_void.
18526 * gdbhooks.py (OptMachineModePrinter): New class.
18527 (build_pretty_printer): Use it for opt_mode.
18529 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18530 Alan Hayward <alan.hayward@arm.com>
18531 David Sherwood <david.sherwood@arm.com>
18533 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18534 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18537 Alan Hayward <alan.hayward@arm.com>
18538 David Sherwood <david.sherwood@arm.com>
18540 * machmode.h (mode_traits): New structure.
18541 (get_narrowest_mode): New function.
18542 (mode_iterator::start): Likewise.
18543 (mode_iterator::iterate_p): Likewise.
18544 (mode_iterator::get_wider): Likewise.
18545 (mode_iterator::get_known_wider): Likewise.
18546 (mode_iterator::get_2xwider): Likewise.
18547 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18548 (FOR_EACH_MODE): Likewise.
18549 (FOR_EACH_MODE_FROM): Likewise.
18550 (FOR_EACH_MODE_UNTIL): Likewise.
18551 (FOR_EACH_WIDER_MODE): Likewise.
18552 (FOR_EACH_2XWIDER_MODE): Likewise.
18553 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18554 * combine.c (simplify_comparison): Likewise
18555 * config/i386/i386.c (type_natural_mode): Likewise.
18556 * cse.c (cse_insn): Likewise.
18557 * dse.c (find_shift_sequence): Likewise.
18558 * emit-rtl.c (init_derived_machine_modes): Likewise.
18559 (init_emit_once): Likewise.
18560 * explow.c (hard_function_value): Likewise.
18561 * expmed.c (extract_fixed_bit_field_1): Likewise.
18562 (extract_bit_field_1): Likewise.
18563 (expand_divmod): Likewise.
18564 (emit_store_flag_1): Likewise.
18565 * expr.c (init_expr_target): Likewise.
18566 (convert_move): Likewise.
18567 (alignment_for_piecewise_move): Likewise.
18568 (widest_int_mode_for_size): Likewise.
18569 (emit_block_move_via_movmem): Likewise.
18570 (copy_blkmode_to_reg): Likewise.
18571 (set_storage_via_setmem): Likewise.
18572 (compress_float_constant): Likewise.
18573 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18574 * optabs-query.c (get_best_extraction_insn): Likewise.
18575 * optabs.c (expand_binop): Likewise.
18576 (expand_twoval_unop): Likewise.
18577 (expand_twoval_binop): Likewise.
18578 (widen_leading): Likewise.
18579 (widen_bswap): Likewise.
18580 (expand_parity): Likewise.
18581 (expand_unop): Likewise.
18582 (prepare_cmp_insn): Likewise.
18583 (prepare_float_lib_cmp): Likewise.
18584 (expand_float): Likewise.
18585 (expand_fix): Likewise.
18586 (expand_sfix_optab): Likewise.
18587 * postreload.c (move2add_use_add2_insn): Likewise.
18588 * reg-stack.c (reg_to_stack): Likewise.
18589 * reginfo.c (choose_hard_reg_mode): Likewise.
18590 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18591 * stor-layout.c (mode_for_size): Likewise.
18592 (smallest_mode_for_size): Likewise.
18593 (mode_for_vector): Likewise.
18594 (finish_bitfield_representative): Likewise.
18595 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18596 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18597 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18598 * var-tracking.c (prepare_call_arguments): Likewise.
18600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18601 Alan Hayward <alan.hayward@arm.com>
18602 David Sherwood <david.sherwood@arm.com>
18604 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18605 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18606 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18607 * machmode.h (mode_size): Move earlier in file.
18608 (mode_precision): Likewise.
18609 (mode_inner): Likewise.
18610 (mode_nunits): Likewise.
18611 (mode_unit_size): Likewise.
18612 (unit_unit_precision): Likewise.
18613 (mode_wider): Likewise.
18614 (mode_2xwider): Likewise.
18615 (machine_mode): New class.
18616 (mode_to_bytes): New function.
18617 (mode_to_bits): Likewise.
18618 (mode_to_precision): Likewise.
18619 (mode_to_inner): Likewise.
18620 (mode_to_unit_size): Likewise.
18621 (mode_to_unit_precision): Likewise.
18622 (mode_to_nunits): Likewise.
18623 (GET_MODE_SIZE): Use mode_to_bytes.
18624 (GET_MODE_BITSIZE): Use mode_to_bits.
18625 (GET_MODE_PRECISION): Use mode_to_precision.
18626 (GET_MODE_INNER): Use mode_to_inner.
18627 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18628 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18629 (GET_MODE_NUNITS): Use mode_to_nunits.
18630 * system.h (ALWAYS_INLINE): New macro.
18631 * config/powerpcspe/powerpcspe-c.c
18632 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18633 int for arg1_mode and arg2_mode.
18635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18636 Alan Hayward <alan.hayward@arm.com>
18637 David Sherwood <david.sherwood@arm.com>
18639 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18640 Prefix mode names with E_ in case statements.
18641 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18642 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18643 (aarch64_split_simd_move): Likewise.
18644 (aarch64_gen_storewb_pair): Likewise.
18645 (aarch64_gen_loadwb_pair): Likewise.
18646 (aarch64_gen_store_pair): Likewise.
18647 (aarch64_gen_load_pair): Likewise.
18648 (aarch64_get_condition_code_1): Likewise.
18649 (aarch64_constant_pool_reload_icode): Likewise.
18650 (get_rsqrte_type): Likewise.
18651 (get_rsqrts_type): Likewise.
18652 (get_recpe_type): Likewise.
18653 (get_recps_type): Likewise.
18654 (aarch64_gimplify_va_arg_expr): Likewise.
18655 (aarch64_simd_container_mode): Likewise.
18656 (aarch64_emit_load_exclusive): Likewise.
18657 (aarch64_emit_store_exclusive): Likewise.
18658 (aarch64_expand_compare_and_swap): Likewise.
18659 (aarch64_gen_atomic_cas): Likewise.
18660 (aarch64_emit_bic): Likewise.
18661 (aarch64_emit_atomic_swap): Likewise.
18662 (aarch64_emit_atomic_load_op): Likewise.
18663 (aarch64_evpc_trn): Likewise.
18664 (aarch64_evpc_uzp): Likewise.
18665 (aarch64_evpc_zip): Likewise.
18666 (aarch64_evpc_ext): Likewise.
18667 (aarch64_evpc_rev): Likewise.
18668 (aarch64_evpc_dup): Likewise.
18669 (aarch64_gen_ccmp_first): Likewise.
18670 (aarch64_gen_ccmp_next): Likewise.
18671 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18672 (alpha_emit_xfloating_libcall): Likewise.
18673 (emit_insxl): Likewise.
18674 (alpha_arg_type): Likewise.
18675 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18676 (arc_preferred_simd_mode): Likewise.
18677 (arc_secondary_reload): Likewise.
18678 (get_arc_condition_code): Likewise.
18679 (arc_print_operand): Likewise.
18680 (arc_legitimate_constant_p): Likewise.
18681 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18682 * config/arc/arc.md (casesi_load): Likewise.
18683 (casesi_compact_jump): Likewise.
18684 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18685 (cc_use_register): Likewise.
18686 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18687 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18688 (arm_init_iwmmxt_builtins): Likewise.
18689 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18690 (neon_expand_vector_init): Likewise.
18691 (arm_attr_length_move_neon): Likewise.
18692 (maybe_get_arm_condition_code): Likewise.
18693 (arm_emit_vector_const): Likewise.
18694 (arm_preferred_simd_mode): Likewise.
18695 (arm_output_iwmmxt_tinsr): Likewise.
18696 (thumb1_output_casesi): Likewise.
18697 (thumb2_output_casesi): Likewise.
18698 (arm_emit_load_exclusive): Likewise.
18699 (arm_emit_store_exclusive): Likewise.
18700 (arm_expand_compare_and_swap): Likewise.
18701 (arm_evpc_neon_vuzp): Likewise.
18702 (arm_evpc_neon_vzip): Likewise.
18703 (arm_evpc_neon_vrev): Likewise.
18704 (arm_evpc_neon_vtrn): Likewise.
18705 (arm_evpc_neon_vext): Likewise.
18706 (arm_validize_comparison): Likewise.
18707 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18708 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18709 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18710 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18711 (c6x_preferred_simd_mode): Likewise.
18712 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18713 (epiphany_rtx_costs): Likewise.
18714 * config/epiphany/predicates.md (proper_comparison_operator):
18716 * config/frv/frv.c (condexec_memory_operand): Likewise.
18717 (frv_emit_move): Likewise.
18718 (output_move_single): Likewise.
18719 (output_condmove_single): Likewise.
18720 (frv_hard_regno_mode_ok): Likewise.
18721 (frv_matching_accg_mode): Likewise.
18722 * config/h8300/h8300.c (split_adds_subs): Likewise.
18723 (h8300_rtx_costs): Likewise.
18724 (h8300_print_operand): Likewise.
18725 (compute_mov_length): Likewise.
18726 (output_logical_op): Likewise.
18727 (compute_logical_op_length): Likewise.
18728 (compute_logical_op_cc): Likewise.
18729 (h8300_shift_needs_scratch_p): Likewise.
18730 (output_a_shift): Likewise.
18731 (compute_a_shift_length): Likewise.
18732 (compute_a_shift_cc): Likewise.
18733 (expand_a_rotate): Likewise.
18734 (output_a_rotate): Likewise.
18735 * config/i386/i386.c (classify_argument): Likewise.
18736 (function_arg_advance_32): Likewise.
18737 (function_arg_32): Likewise.
18738 (function_arg_64): Likewise.
18739 (function_value_64): Likewise.
18740 (ix86_gimplify_va_arg): Likewise.
18741 (ix86_legitimate_constant_p): Likewise.
18742 (put_condition_code): Likewise.
18743 (split_double_mode): Likewise.
18744 (ix86_avx256_split_vector_move_misalign): Likewise.
18745 (ix86_expand_vector_logical_operator): Likewise.
18746 (ix86_split_idivmod): Likewise.
18747 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18748 (ix86_build_const_vector): Likewise.
18749 (ix86_build_signbit_mask): Likewise.
18750 (ix86_match_ccmode): Likewise.
18751 (ix86_cc_modes_compatible): Likewise.
18752 (ix86_expand_branch): Likewise.
18753 (ix86_expand_sse_cmp): Likewise.
18754 (ix86_expand_sse_movcc): Likewise.
18755 (ix86_expand_int_sse_cmp): Likewise.
18756 (ix86_expand_vec_perm_vpermi2): Likewise.
18757 (ix86_expand_vec_perm): Likewise.
18758 (ix86_expand_sse_unpack): Likewise.
18759 (ix86_expand_int_addcc): Likewise.
18760 (ix86_split_to_parts): Likewise.
18761 (ix86_vectorize_builtin_gather): Likewise.
18762 (ix86_vectorize_builtin_scatter): Likewise.
18763 (avx_vpermilp_parallel): Likewise.
18764 (inline_memory_move_cost): Likewise.
18765 (ix86_tieable_integer_mode_p): Likewise.
18766 (x86_maybe_negate_const_int): Likewise.
18767 (ix86_expand_vector_init_duplicate): Likewise.
18768 (ix86_expand_vector_init_one_nonzero): Likewise.
18769 (ix86_expand_vector_init_one_var): Likewise.
18770 (ix86_expand_vector_init_concat): Likewise.
18771 (ix86_expand_vector_init_interleave): Likewise.
18772 (ix86_expand_vector_init_general): Likewise.
18773 (ix86_expand_vector_set): Likewise.
18774 (ix86_expand_vector_extract): Likewise.
18775 (emit_reduc_half): Likewise.
18776 (ix86_emit_i387_round): Likewise.
18777 (ix86_mangle_type): Likewise.
18778 (ix86_expand_round_sse4): Likewise.
18779 (expand_vec_perm_blend): Likewise.
18780 (canonicalize_vector_int_perm): Likewise.
18781 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18782 (expand_vec_perm_1): Likewise.
18783 (expand_vec_perm_interleave3): Likewise.
18784 (expand_vec_perm_even_odd_pack): Likewise.
18785 (expand_vec_perm_even_odd_1): Likewise.
18786 (expand_vec_perm_broadcast_1): Likewise.
18787 (ix86_vectorize_vec_perm_const_ok): Likewise.
18788 (ix86_expand_vecop_qihi): Likewise.
18789 (ix86_expand_mul_widen_hilo): Likewise.
18790 (ix86_expand_sse2_abs): Likewise.
18791 (ix86_expand_pextr): Likewise.
18792 (ix86_expand_pinsr): Likewise.
18793 (ix86_preferred_simd_mode): Likewise.
18794 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18795 * config/i386/sse.md (*andnot<mode>3): Likewise.
18796 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18797 (*<code><mode>3): Likewise.
18798 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18799 (ia64_expand_atomic_op): Likewise.
18800 (ia64_arg_type): Likewise.
18801 (ia64_mode_to_int): Likewise.
18802 (ia64_scalar_mode_supported_p): Likewise.
18803 (ia64_vector_mode_supported_p): Likewise.
18804 (expand_vec_perm_broadcast): Likewise.
18805 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18806 (iq2000_function_arg_advance): Likewise.
18807 (iq2000_function_arg): Likewise.
18808 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18809 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18810 (m68k_libcall_value): Likewise.
18811 (m68k_function_value): Likewise.
18812 (sched_attr_op_type): Likewise.
18813 * config/mcore/mcore.c (mcore_output_move): Likewise.
18814 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18816 (microblaze_function_arg): Likewise.
18817 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18818 (mips_print_operand): Likewise.
18819 (mips_mode_ok_for_mov_fmt_p): Likewise.
18820 (mips_vector_mode_supported_p): Likewise.
18821 (mips_preferred_simd_mode): Likewise.
18822 (mips_expand_vpc_loongson_even_odd): Likewise.
18823 (mips_expand_vec_unpack): Likewise.
18824 (mips_expand_vi_broadcast): Likewise.
18825 (mips_expand_vector_init): Likewise.
18826 (mips_expand_vec_reduc): Likewise.
18827 (mips_expand_msa_cmp): Likewise.
18828 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18829 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18830 (cc_flags_for_mode): Likewise.
18831 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18832 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18833 (nds32_output_casesi_pc_relative): Likewise.
18834 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18835 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18836 (nvptx_gen_unpack): Likewise.
18837 (nvptx_gen_pack): Likewise.
18838 (nvptx_gen_shuffle): Likewise.
18839 (nvptx_gen_wcast): Likewise.
18840 (nvptx_preferred_simd_mode): Likewise.
18841 * config/pa/pa.c (pa_secondary_reload): Likewise.
18842 * config/pa/predicates.md (base14_operand): Likewise.
18843 * config/powerpcspe/powerpcspe-c.c
18844 (altivec_resolve_overloaded_builtin): Likewise.
18845 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18847 (rs6000_preferred_simd_mode): Likewise.
18848 (output_vec_const_move): Likewise.
18849 (rs6000_expand_vector_extract): Likewise.
18850 (rs6000_split_vec_extract_var): Likewise.
18851 (reg_offset_addressing_ok_p): Likewise.
18852 (rs6000_legitimate_offset_address_p): Likewise.
18853 (rs6000_legitimize_address): Likewise.
18854 (rs6000_emit_set_const): Likewise.
18855 (rs6000_const_vec): Likewise.
18856 (rs6000_emit_move): Likewise.
18857 (spe_build_register_parallel): Likewise.
18858 (rs6000_darwin64_record_arg_recurse): Likewise.
18859 (swap_selector_for_mode): Likewise.
18860 (spe_init_builtins): Likewise.
18861 (paired_init_builtins): Likewise.
18862 (altivec_init_builtins): Likewise.
18863 (do_load_for_compare): Likewise.
18864 (rs6000_generate_compare): Likewise.
18865 (rs6000_expand_float128_convert): Likewise.
18866 (emit_load_locked): Likewise.
18867 (emit_store_conditional): Likewise.
18868 (rs6000_output_function_epilogue): Likewise.
18869 (rs6000_handle_altivec_attribute): Likewise.
18870 (rs6000_function_value): Likewise.
18871 (emit_fusion_gpr_load): Likewise.
18872 (emit_fusion_p9_load): Likewise.
18873 (emit_fusion_p9_store): Likewise.
18874 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
18875 (fusion_gpr_mem_load): Likewise.
18876 (fusion_addis_mem_combo_load): Likewise.
18877 (fusion_addis_mem_combo_store): Likewise.
18878 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
18879 (fusion_gpr_mem_load): Likewise.
18880 (fusion_addis_mem_combo_load): Likewise.
18881 (fusion_addis_mem_combo_store): Likewise.
18882 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18884 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
18885 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18886 (rs6000_preferred_simd_mode): Likewise.
18887 (output_vec_const_move): Likewise.
18888 (rs6000_expand_vector_extract): Likewise.
18889 (rs6000_split_vec_extract_var): Likewise.
18890 (reg_offset_addressing_ok_p): Likewise.
18891 (rs6000_legitimate_offset_address_p): Likewise.
18892 (rs6000_legitimize_address): Likewise.
18893 (rs6000_emit_set_const): Likewise.
18894 (rs6000_const_vec): Likewise.
18895 (rs6000_emit_move): Likewise.
18896 (rs6000_darwin64_record_arg_recurse): Likewise.
18897 (swap_selector_for_mode): Likewise.
18898 (paired_init_builtins): Likewise.
18899 (altivec_init_builtins): Likewise.
18900 (rs6000_expand_float128_convert): Likewise.
18901 (emit_load_locked): Likewise.
18902 (emit_store_conditional): Likewise.
18903 (rs6000_output_function_epilogue): Likewise.
18904 (rs6000_handle_altivec_attribute): Likewise.
18905 (rs6000_function_value): Likewise.
18906 (emit_fusion_gpr_load): Likewise.
18907 (emit_fusion_p9_load): Likewise.
18908 (emit_fusion_p9_store): Likewise.
18909 * config/rx/rx.c (rx_gen_move_template): Likewise.
18910 (flags_from_mode): Likewise.
18911 * config/s390/predicates.md (s390_alc_comparison): Likewise.
18912 (s390_slb_comparison): Likewise.
18913 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
18914 (s390_vector_mode_supported_p): Likewise.
18915 (s390_cc_modes_compatible): Likewise.
18916 (s390_match_ccmode_set): Likewise.
18917 (s390_canonicalize_comparison): Likewise.
18918 (s390_emit_compare_and_swap): Likewise.
18919 (s390_branch_condition_mask): Likewise.
18920 (s390_rtx_costs): Likewise.
18921 (s390_secondary_reload): Likewise.
18922 (__SECONDARY_RELOAD_CASE): Likewise.
18923 (s390_expand_cs): Likewise.
18924 (s390_preferred_simd_mode): Likewise.
18925 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
18926 * config/sh/sh.c (sh_print_operand): Likewise.
18927 (dump_table): Likewise.
18928 (sh_secondary_reload): Likewise.
18929 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18930 * config/sh/sh.md (casesi_worker_1): Likewise.
18931 (casesi_worker_2): Likewise.
18932 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
18933 (fcc_comparison_operator): Likewise.
18934 * config/sparc/sparc.c (sparc_expand_move): Likewise.
18935 (emit_soft_tfmode_cvt): Likewise.
18936 (sparc_preferred_simd_mode): Likewise.
18937 (output_cbranch): Likewise.
18938 (sparc_print_operand): Likewise.
18939 (sparc_expand_vec_perm_bmask): Likewise.
18940 (vector_init_bshuffle): Likewise.
18941 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18942 (spu_vector_mode_supported_p): Likewise.
18943 (spu_expand_insv): Likewise.
18944 (spu_emit_branch_or_set): Likewise.
18945 (spu_handle_vector_attribute): Likewise.
18946 (spu_builtin_splats): Likewise.
18947 (spu_builtin_extract): Likewise.
18948 (spu_builtin_promote): Likewise.
18949 (spu_expand_sign_extend): Likewise.
18950 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18951 (tilegx_simd_int): Likewise.
18952 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
18953 (tilepro_simd_int): Likewise.
18954 * config/v850/v850.c (const_double_split): Likewise.
18955 (v850_print_operand): Likewise.
18956 (ep_memory_offset): Likewise.
18957 * config/vax/vax.c (vax_rtx_costs): Likewise.
18958 (vax_output_int_move): Likewise.
18959 (vax_output_int_add): Likewise.
18960 (vax_output_int_subtract): Likewise.
18961 * config/visium/predicates.md (visium_branch_operator): Likewise.
18962 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
18963 (visium_print_operand_address): Likewise.
18964 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18965 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
18966 (xtensa_expand_conditional_branch): Likewise.
18967 (xtensa_copy_incoming_a7): Likewise.
18968 (xtensa_output_literal): Likewise.
18969 * dfp.c (decimal_real_maxval): Likewise.
18970 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18973 Alan Hayward <alan.hayward@arm.com>
18974 David Sherwood <david.sherwood@arm.com>
18976 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
18977 (mode_nunits_inline): Likewise.
18978 (mode_inner_inline): Likewise.
18979 (mode_unit_size_inline): Likewise.
18980 (mode_unit_precision_inline): Likewise.
18981 (emit_insn_modes_h): Likewise. Also emit a #define of the
18983 (emit_mode_wider): Add an E_ prefix to mode names.
18984 (emit_mode_complex): Likewise.
18985 (emit_mode_inner): Likewise.
18986 (emit_mode_adjustments): Likewise.
18987 (emit_mode_int_n): Likewise.
18988 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
18989 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
18990 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
18991 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
18992 (CRC32_BUILTIN, ENTRY): Likewise.
18993 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
18994 (aarch64_pop_regs): Likewise.
18995 (aarch64_process_components): Likewise.
18996 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
18997 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
18998 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
18999 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19000 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19001 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19003 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19004 (pcmpistr): Likewise.
19005 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19006 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19007 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19009 * config/rl78/rl78.c (mduc_regs): Likewise.
19010 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19011 (htm_expand_builtin): Likewise.
19012 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19013 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19014 * config/xtensa/xtensa.c (print_operand): Likewise.
19015 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19016 (NUM_MODE_VECTOR_INT): Likewise.
19017 * genoutput.c (null_operand): Likewise.
19018 (output_operand_data): Likewise.
19019 * genrecog.c (print_parameter_value): Likewise.
19020 * lra.c (debug_operand_data): Likewise.
19022 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19024 * dwarf2out.c (scompare_loc_descriptor_wide)
19025 (scompare_loc_descriptor_narrow): New functions, split out from...
19026 (scompare_loc_descriptor): ...here.
19027 * expmed.c (emit_store_flag_int): New function, split out from...
19028 (emit_store_flag): ...here.
19030 2017-08-30 Richard Biener <rguenther@suse.de>
19032 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19033 (dwarf2out_early_finish): Move setting of AT_pubnames from
19034 early debug output to early finish.
19036 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19038 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19039 and -mdata-region to the assembler.
19041 2017-08-30 Richard Biener <rguenther@suse.de>
19043 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19045 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19046 (dwarf2out_early_global_decl): Only generate a DIE for the
19047 abstract origin if it doesn't already exist or is a declaration DIE.
19048 (resolve_addr): Do not add the linkage name twice when
19049 generating a stub DIE for the DW_TAG_GNU_call_site target.
19051 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19053 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19054 Use machine_mode rather than int for arg1_mode.
19056 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19059 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19060 that the second argument of the built-in functions to unpack
19061 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19062 switch statement instead a lot of if statements.
19063 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19064 Allow 64-bit values to be in Altivec registers as well as
19065 traditional floating point registers.
19066 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19068 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19070 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19071 MAX_REGS_PER_ADDRESS == 1.
19073 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19075 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19076 * config/i386/i386.c (ix86_option_override_internal): Simplify
19077 setting of opts->x_flag_entry.
19079 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19080 Jakub Jelinek <jakub@redhat.com>
19081 Richard Biener <rguenther@suse.de>
19083 PR tree-optimization/81503
19084 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19085 folded constant fits in the target type; reorder tests for clarity.
19087 2017-08-29 Martin Liska <mliska@suse.cz>
19089 * passes.def: Include pass_lower_switch.
19090 * stmt.c (dump_case_nodes): Remove and move to
19091 tree-switch-conversion.
19092 (case_values_threshold): Likewise.
19093 (expand_switch_as_decision_tree_p): Likewise.
19094 (emit_case_decision_tree): Likewise.
19095 (expand_case): Likewise.
19096 (balance_case_nodes): Likewise.
19097 (node_has_low_bound): Likewise.
19098 (node_has_high_bound): Likewise.
19099 (node_is_bounded): Likewise.
19100 (emit_case_nodes): Likewise.
19101 (struct simple_case_node): New struct.
19102 (add_case_node): Remove.
19103 (emit_case_dispatch_table): Use vector instead of case_list.
19104 (reset_out_edges_aux): Remove.
19105 (compute_cases_per_edge): Likewise.
19106 (expand_case): Build list of simple_case_node.
19107 (expand_sjlj_dispatch_table): Use it.
19108 * tree-switch-conversion.c (struct case_node): Moved from
19109 stmt.c and adjusted.
19110 (emit_case_nodes): Likewise.
19111 (node_has_low_bound): Likewise.
19112 (node_has_high_bound): Likewise.
19113 (node_is_bounded): Likewise.
19114 (case_values_threshold): Likewise.
19115 (reset_out_edges_aux): Likewise.
19116 (compute_cases_per_edge): Likewise.
19117 (add_case_node): Likewise.
19118 (dump_case_nodes): Likewise.
19119 (balance_case_nodes): Likewise.
19120 (expand_switch_as_decision_tree_p): Likewise.
19121 (emit_jump): Likewise.
19122 (emit_case_decision_tree): Likewise.
19123 (try_switch_expansion): Likewise.
19124 (do_jump_if_equal): Likewise.
19125 (emit_cmp_and_jump_insns): Likewise.
19126 (fix_phi_operands_for_edge): New function.
19127 (record_phi_operand_mapping): Likewise.
19128 (class pass_lower_switch): New pass.
19129 (pass_lower_switch::execute): New function.
19130 (make_pass_lower_switch): Likewise.
19131 (conditional_probability):
19132 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19133 * tree-pass.h: Add make_pass_lower_switch.
19135 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19138 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19141 2017-08-29 Richard Biener <rguenther@suse.de>
19143 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19144 we do not add a DW_AT_inline attribute twice.
19145 (gen_subprogram_die): Remove code setting DW_AT_inline on
19146 DECL_ABSTRACT_P nodes.
19148 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19150 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19151 calls to internal functions.
19152 (gimplify_modify_expr): Likewise.
19153 * tree-call-cdce.c (use_internal_fn): Likewise.
19154 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19155 (convert_to_divmod): Set the nothrow flag.
19156 * tree-if-conv.c (predicate_mem_writes): Likewise.
19157 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19158 (vectorizable_call): Likewise.
19159 (vectorizable_store): Likewise.
19160 (vectorizable_load): Likewise.
19161 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19162 (vect_recog_mask_conversion_pattern): Likewise.
19164 2017-08-29 Martin Liska <mliska@suse.cz>
19167 * gcc.c (driver_handle_option): Add new argument.
19168 * opts-common.c (handle_option): Pass
19169 target_option_override_hook.
19170 * opts-global.c (lang_handle_option): Add new option.
19171 (set_default_handlers): Add new argument.
19172 (decode_options): Likewise.
19173 * opts.c (target_handle_option): Likewise.
19174 (common_handle_option): Call target_option_override_hook.
19175 * opts.h (struct cl_option_handler_func): Add hook for
19176 target option override.
19177 (struct cl_option_handlers): Likewise.
19178 (set_default_handlers): Add new argument.
19179 (decode_options): Likewise.
19180 (common_handle_option): Likewise.
19181 (target_handle_option): Likewise.
19182 * toplev.c (toplev::main): Pass targetm.target_option.override
19185 2017-08-29 Richard Biener <rguenther@suse.de>
19186 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19188 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19189 life to the active subtree.
19191 2017-08-28 Jeff Law <law@redhat.com>
19193 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19194 derive_equivalences.
19195 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19196 Code moved into....
19197 (edge_info::derive_equivalences): New private member function
19199 * tree-ssa-dom.c (class edge_info): Changed from a struct
19200 to a class. Add ctor/dtor, methods and data members.
19201 (edge_info::edge_info): Renamed from allocate_edge_info.
19202 Initialize additional members.
19203 (edge_info::~edge_info): New.
19204 (free_dom_edge_info): Delete the edge info.
19205 (record_edge_info): Use new class & associated member functions.
19206 Tighten forms for testing for edge equivalences.
19207 (record_temporary_equivalences): Iterate over the simple
19208 equivalences rather than assuming there's only one per edge.
19209 (cprop_into_successor_phis): Iterate over the simple
19210 equivalences rather than assuming there's only one per edge.
19211 (optimize_stmt): Use operand_equal_p rather than pointer
19212 equality for mini-DSE code.
19214 2017-08-28 Nathan Sidwell <nathan@acm.org>
19216 * gcc.c (execute): Fold SIGPIPE handling into switch
19217 statement. Adjust internal error message.
19219 2017-08-28 Richard Biener <rguenther@suse.de>
19222 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19223 Do nothing for removed DIEs.
19225 2017-08-28 Richard Biener <rguenther@suse.de>
19227 PR tree-optimization/81977
19228 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19231 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19234 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19235 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19236 using targetm.gen_mem_thread_fence.
19238 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19241 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19242 predicate to register_operand. Reorder operands.
19243 (*btr<mode>): Ditto.
19244 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19245 (*btr<mode>_mask): Ditto.
19247 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19249 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19250 * config/rs6000/xmmintrin.h: New file.
19251 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19253 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19256 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19257 parameter and_insn and return it.
19258 (recombine_lvx_pattern): Insert a copy to ensure availability of
19259 the base register of the copied masking operation at the point of
19260 the instruction replacement.
19261 (recombine_stvx_pattern): Likewise.
19263 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19265 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19266 undocumented switches.
19267 (-mpower9-dform-vector): Likewise.
19268 (-mpower9-dform): Likewise.
19269 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19270 comments to delete references to -mpower9-dform* switches.
19271 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19272 Delete reference to -mpower9-dform* switches, test for
19273 -mpower9-vector instead.
19274 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19275 (OTHER_P9_VECTOR_MASKS): Likewise.
19276 (POWERPC_MASKS): Likewise.
19277 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19278 tests against -mpower9-dform* to -mpower9-vector. Delete code
19279 that checked for -mpower9-dform* consistancy with other options.
19280 Add test for -mpower9-misc to enable other power9 switches.
19281 (rs6000_init_hard_regno_mode_ok): Likewise.
19282 (rs6000_option_override_internal): Likewise.
19283 (rs6000_emit_prologue): Likewise.
19284 (rs6000_emit_epilogue): Likewise.
19285 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19286 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19287 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19289 (emit_fusion_p9_store): Likewise.
19290 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19291 resetting these macros if the assembler does not support ISA 3.0
19293 (TARGET_P9_DFORM_VECTOR): Likewise.
19294 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19295 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19297 2017-08-25 Alan Modra <amodra@gmail.com>
19299 PR rtl-optimization/81747
19300 * cse.c (cse_extended_basic_block): Don't attempt to record
19301 equivalences for degenerate conditional branches that branch
19302 to their fall-through.
19304 2017-08-24 Martin Sebor <msebor@redhat.com>
19306 PR middle-end/81908
19307 * gimple-fold.c (size_must_be_zero_p): New function.
19308 (gimple_fold_builtin_memory_op): Call it.
19310 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19312 * config/rs6000/mm_malloc.h: New file.
19314 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19316 PR tree-optimization/81913
19317 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19318 analysis when either IVs in condition can wrap.
19320 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19322 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19323 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19325 2017-08-24 Richard Biener <rguenther@suse.de>
19328 * targhooks.c (default_target_can_inline_p): Properly
19329 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19330 is present and always compare.
19331 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19332 infer -mfpmath=sse from TARGET_SSE_P.
19333 (ix86_can_inline_p): Properly use target_option_default_node when
19334 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19336 2017-08-24 Richard Biener <rguenther@suse.de>
19339 * dwarf2out.c (output_die): Handle flag_generate_offload like
19341 (output_comp_unit): Likewise.
19342 (gen_array_type_die): Likewise.
19343 (dwarf2out_early_finish): Likewise.
19344 (note_variable_value_in_expr): Likewise.
19345 (dwarf2out_finish): Likewise. Adjust assert.
19346 * cgraphunit.c (symbol_table::compile): Move setting of
19347 flag_generate_offload earlier ...
19348 (symbol_table::finalize_compilation_unit): ... here, before
19349 early debug finalization.
19351 2017-08-24 Richard Biener <rguenther@suse.de>
19353 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19354 and ipa-fnsummary.h.
19355 (ix86_can_inline_p): When ix86_fpmath flags do not match
19356 check whether the callee uses FP math at all.
19358 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19360 PR middle-end/81931
19361 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19362 instead of TYPE_PRECISION.
19364 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19365 Alan Hayward <alan.hayward@arm.com>
19366 David Sherwood <david.sherwood@arm.com>
19368 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19370 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19372 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19374 (ix86_frame::stack_realign_allocate): New field.
19375 (struct machine_frame_state): Modify comments.
19376 (machine_frame_state::sp_realigned_fp_end): New field.
19377 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19378 layout calculation.
19379 (sp_valid_at): Add assertion to assure no attempt to access invalid
19380 offset of a realigned stack.
19381 (fp_valid_at): Likewise.
19382 (choose_baseaddr): Modify comments.
19383 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19384 ix86_expand_prologue.
19385 (ix86_expand_prologue): Modify stack realignment and allocation.
19386 (ix86_expand_epilogue): Modify comments.
19387 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19388 avx2_runtime, avx512f, and avx512f_runtime.
19390 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19392 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19393 (mstackrealign): Do not init to -1.
19394 * config/i386/i386.c (ix86_option_override_internal):
19395 Check opts_set, not opts when setting default value of
19396 opts->x_ix86_force_align_arg_pointer.
19398 2017-08-23 Richard Biener <rguenther@suse.de>
19400 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19401 lang_hooks.decl_printable_name.
19402 * print-rtl-function.c (print_rtx_function): Likewise.
19403 * tree-pretty-print.c (dump_function_header): Likewise.
19405 2017-08-23 Richard Biener <rguenther@suse.de>
19408 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19409 -g0 at compile-time.
19411 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19413 PR middle-end/19706
19414 * doc/sourcebuild.texi (Other hardware attributes):
19417 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19419 PR middle-end/19706
19420 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19421 Fix single-use check.
19423 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19425 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19427 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19429 * doc/install.texi: Modify to add more details on running selected
19432 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19434 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19435 is combined with -mabi=ms.
19436 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19439 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19441 PR tree-optimization/81488
19442 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19443 and cached_basis fields.
19444 (MAX_SPREAD): New constant.
19445 (alloc_cand_and_find_basis): Initialize new fields.
19446 (clear_visited): New function.
19447 (create_phi_basis_1): Rename from create_phi_basis, set visited
19448 and cached_basis fields.
19449 (create_phi_basis): New wrapper function.
19450 (phi_add_costs_1): Rename from phi_add_costs, add spread
19451 parameter, set visited field, short-circuit when limits reached.
19452 (phi_add_costs): New wrapper function.
19453 (record_phi_increments_1): Rename from record_phi_increments, set
19455 (record_phi_increments): New wrapper function.
19456 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19457 (phi_incr_cost): New wrapper function.
19458 (all_phi_incrs_profitable_1): Rename from
19459 all_phi_incrs_profitable, set visited field.
19460 (all_phi_incrs_profitable): New wrapper function.
19462 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19463 Alan Hayward <alan.hayward@arm.com>
19464 David Sherwood <david.sherwood@arm.com>
19466 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19467 that takes the outer and inner modes.
19468 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19469 comparison as the canonical test for a paradoxical subreg.
19470 * combine.c (simplify_set): Use paradoxical_subreg_p.
19471 (make_extraction): Likewise.
19472 (force_to_mode): Likewise.
19473 (rtx_equal_for_field_assignment_p): Likewise.
19474 (gen_lowpart_for_combine): Likewise.
19475 (simplify_comparison): Likewise.
19476 * cse.c (equiv_constant): Likewise.
19477 * expmed.c (store_bit_field_1): Likewise.
19478 * final.c (alter_subreg): Likewise.
19479 * fwprop.c (propagate_rtx): Likewise.
19480 (forward_propagate_subreg): Likewise.
19481 * ira-conflicts.c (ira_build_conflicts): Likewise.
19482 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19483 * lra-constraints.c (curr_insn_transform): Likewise.
19484 (split_reg): Likewise.
19485 * lra-eliminations.c (move_plus_up): Likewise.
19486 (lra_eliminate_regs_1): Likewise.
19487 * recog.c (general_operand): Likewise.
19488 * ree.c (combine_reaching_defs): Likewise.
19489 * reload.c (push_reload): Likewise.
19490 (find_reloads): Likewise.
19491 * reload1.c (elimination_effects): Likewise.
19492 (compute_reload_subreg_offset): Likewise.
19493 (choose_reload_regs): Likewise.
19494 * rtlanal.c (subreg_lsb_1): Likewise.
19495 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19496 (simplify_subreg): Likewise.
19497 * var-tracking.c (track_loc_p): Likewise.
19498 * emit-rtl.c (byte_lowpart_offset): Likewise.
19499 (paradoxical_subreg_p): Delete out-of-line definition.
19501 2017-08-22 Jeff Law <law@redhat.com>
19503 PR tree-optimization/81741
19504 PR tree-optimization/71947
19505 * tree-ssa-dom.c: Include tree-inline.h.
19506 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19507 equivalences if one is more expensive to compute than the other.
19508 * tree-ssa-scopedtables.h (class const_or_copies): Make
19509 record_const_or_copy_raw method private.
19510 (class avail_exprs_stack): New method simplify_binary_operation.
19511 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19512 avail_exprs_stack::simplify_binary_operation as needed.
19513 (avail_exprs_stack::simplify_binary_operation): New function.
19515 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19517 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19518 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19519 (DOT_SYMBOLS): Likewise.
19520 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19521 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19522 (RS6000_ABI_NAME): Likewise.
19523 (TARGET_CMODEL): Likewise.
19524 (TOC_SECTION_ASM_OP): Likewise.
19525 (SET_CMODEL): New macro.
19526 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19528 2017-08-22 Richard Biener <rguenther@suse.de>
19530 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19531 to allow for free-lang-data replacements similar to verify_type_variant.
19533 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19534 Alan Hayward <alan.hayward@arm.com>
19535 David Sherwood <david.sherwood@arm.com>
19537 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19538 VOIDmode for the LABEL_REF.
19540 2017-08-22 Richard Biener <rguenther@suse.de>
19542 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19544 2017-08-22 Richard Biener <rguenther@suse.de>
19546 * common.opt (feliminate-dwarf2-dups): Ignore.
19547 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19548 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19549 same_die_p_wrap, compute_section_prefix,
19550 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19551 (comdat_symbol_id, comdat_symbol_number): Likewise.
19552 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19554 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19555 (output_die): Mark unreachable path unreachable.
19556 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19557 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19558 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19559 (dwarf2out_early_finish): Likewise.
19561 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19563 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19565 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19568 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19569 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19570 (avr_attribute_table) <io, io_low, address>: Initialize
19571 .decl_required with true.
19573 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19575 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19576 undocumented debugging options.
19577 (-mvsx-scalar-double): Likewise.
19578 (-mallow-df-permute): Likewise.
19579 (-mvectorize-builtins): Likewise.
19580 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19581 (rs6000_builtin_vectorized_function): Likewise.
19582 (rs6000_builtin_md_vectorized_function): Likewise.
19583 (rs6000_opt_vars): Likewise.
19585 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19588 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19589 (*btrq_imm): Rename from *btrq.
19590 (*btcq_imm): Rename from *btcq.
19591 (btsc): New code attribute.
19592 (*<btsc><mode>): New insn pattern.
19593 (*btr<mode>): Ditto.
19594 (*<btsc><mode>_mask): New insn_and_split pattern.
19595 (*btr<mode>_mask): Ditto.
19597 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19598 Alan Hayward <alan.hayward@arm.com>
19599 David Sherwood <david.sherwood@arm.com>
19601 * function.c (pad_below): Simplify padding calculation.
19603 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19604 Alan Hayward <alan.hayward@arm.com>
19605 David Sherwood <david.sherwood@arm.com>
19607 * target.def (function_prologue): Remove frame size argument.
19608 (function_epilogue): Likewise.
19609 * doc/tm.texi: Regenerate.
19610 * final.c (final_start_function): Update call to function_prologue.
19611 (final_end_function): Update call to function_epilogue.
19612 (default_function_pro_epilogue): Remove frame size argument.
19613 * output.h (default_function_pro_epilogue): Likewise.
19614 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19615 (arm_output_function_prologue): Likewise.
19616 * config/frv/frv.c (frv_function_prologue): Likewise.
19617 (frv_function_epilogue): Likewise.
19618 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19619 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19620 (ia64_output_function_epilogue): Likewise.
19621 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19622 (m32r_output_function_epilogue): Likewise.
19623 * config/microblaze/microblaze.c (microblaze_function_prologue)
19624 (microblaze_function_epilogue): Likewise.
19625 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19626 (mips_output_function_epilogue): Likewise.
19627 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19628 (mmix_target_asm_function_epilogue): Likewise.
19629 * config/msp430/msp430.c (msp430_start_function): Likewise.
19630 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19631 (nds32_asm_function_epilogue): Likewise.
19632 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19633 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19634 (pa_output_function_epilogue): Likewise.
19635 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19636 (rs6000_output_function_epilogue): Likewise.
19637 * config/rl78/rl78.c (rl78_start_function): Likewise.
19638 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19639 (rs6000_output_function_epilogue): Likewise.
19640 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19641 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19642 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19643 (sparc_asm_function_epilogue): Likewise.
19645 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19647 * tree.h (type_has_mode_precision_p): New function.
19648 * convert.c (convert_to_integer_1): Use it.
19649 * expr.c (expand_expr_real_2): Likewise.
19650 (expand_expr_real_1): Likewise.
19651 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19652 * match.pd: Likewise.
19653 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19654 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19655 * tree-tailcall.c (process_assignment): Likewise.
19656 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19657 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19658 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19659 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19660 (vectorizable_assignment): Likewise.
19661 (vectorizable_shift): Likewise.
19662 (vectorizable_operation): Likewise.
19663 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19665 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19667 * match.pd: Add pow (C, x) simplification.
19669 2017-08-21 Richard Biener <rguenther@suse.de>
19671 PR tree-optimization/81900
19672 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19673 for blocks with abnormal predecessors.
19674 (compute_antic): Do not set visited flag prematurely.
19676 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19679 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19681 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19683 * stor-layout.h (vector_type_mode): Move to...
19684 * tree.h (vector_type_mode): ...here.
19685 * stor-layout.c (vector_type_mode): Move to...
19686 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19688 2017-08-21 Richard Biener <rguenther@suse.de>
19690 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19691 register_external_die hooks.
19692 (debug_false_tree_charstarstar_uhwistar): Declare.
19693 (debug_nothing_tree_charstar_uhwi): Likewise.
19694 * debug.c (do_nothing_debug_hooks): Adjust.
19695 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19696 (debug_nothing_tree_charstar_uhwi): Likewise.
19697 * dbxout.c (dbx_debug_hooks): Adjust.
19698 (xcoff_debug_hooks): Likewise.
19699 * sdbout.c (sdb_debug_hooks): Likewise.
19700 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19701 * dwarf2out.c (macinfo_label_base): New global.
19702 (dwarf2out_register_external_die): New function for the
19703 register_external_die hook.
19704 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19705 (dwarf2_debug_hooks): Use them.
19706 (dwarf2_lineno_debug_hooks): Adjust.
19707 (struct die_struct): Add with_offset flag.
19708 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19709 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19710 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19711 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19712 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19713 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19714 defining section names for the early LTO debug variants.
19715 (reset_indirect_string): New helper.
19716 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19717 (print_dw_val): Add support for offsetted symbol references.
19718 (get_ultimate_context): Split out from is_cxx.
19719 (is_cxx): Use get_ultimate_context.
19720 (is_fortran): Add decl overload.
19721 (compute_comp_unit_symbol): Split out worker from
19722 compute_section_prefix.
19723 (compute_section_prefix): Call compute_comp_unit_symbol and
19724 set comdat_type_p here.
19725 (output_die): Skip DIE symbol output for the LTO added one.
19726 Handle DIE symbol references with offset.
19727 (output_comp_unit): Guard section name mangling properly.
19728 For LTO debug sections emit a symbol at the section beginning
19729 which we use to refer to its DIEs.
19730 (add_abstract_origin_attribute): For DIEs registered via
19731 dwarf2out_register_external_die directly refer to the early
19732 DIE rather than indirectly through the shadow one we created.
19733 Remove obsolete call to dwarf2out_abstract_function for
19734 non-function/block origins.
19735 (gen_array_type_die): When generating early LTO debug do
19736 not emit DW_AT_string_length.
19737 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19738 late when in LTO. As suggested place a gcc_unreachable for
19739 the DECL_ABSTRACT_P case.
19740 (gen_subprogram_die): Avoid another specification DIE
19741 for early built declarations/definitions for the late LTO case.
19742 (gen_variable_die): Add type references for late duplicated VLA dies
19744 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19745 we have the abstract instance already.
19746 (process_scope_var): Adjust decl DIE contexts in LTO which
19747 first puts them in limbo.
19748 (gen_decl_die): Do not generate type DIEs late apart from
19749 types for VLAs or for decls we do not yet have a DIE. Do not
19750 call dwarf2out_abstract_function late.
19751 (dwarf2out_early_global_decl): Make sure to create DIEs
19752 for abstract instances of a decl first.
19753 (dwarf2out_late_global_decl): Adjust comment.
19754 (output_macinfo_op): With multiple macro sections use
19755 macinfo_label_base to distinguish labels.
19756 (output_macinfo): Likewise. Update macinfo_label_base.
19757 Pass in the line info label.
19758 (note_variable_value_in_expr): When generating LTO resolve
19759 all variable values here by generating DIEs as needed.
19760 (init_sections_and_labels): Add early LTO debug flag parameter
19761 and generate different sections and names if set. Add generation
19762 counter for the labels so we can have multiple of them.
19763 (reset_dies): Helper to allow DIEs to be output multiple times.
19764 (dwarf2out_finish): When outputting DIEs to the fat part of an
19765 LTO object first reset DIEs.
19766 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19767 (modified_type_die): Check for decl_ultimate_origin being self
19769 (gen_type_die_with_usage): Likewise.
19770 (gen_typedef_die): Allow decl_ultimate_origin being self.
19771 (set_decl_abstract_flags): Remove.
19772 (set_block_abstract_flags): Likewise.
19773 (dwarf2out_abstract_function): Treat the early generated DIEs
19774 as the abstract copy and only add DW_AT_inline and
19775 DW_AT_artificial here and call set_decl_origin_self.
19776 If the DIE has an abstract origin don't do anything.
19777 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19778 if we have none yet (Go fails to build one, PR78628).
19779 (variably_modified_type_p): Prevent endless recursion for Ada
19780 cyclic pointer types.
19781 * lto-streamer-in.c: Include debug.h.
19782 (dref_queue): New global.
19783 (lto_read_tree_1): Stream in DIE references.
19784 (lto_input_tree): Register DIE references.
19785 (input_function): Stream DECL_DEBUG_ARGS.
19786 * lto-streamer-out.c: Include debug.h.
19787 (lto_write_tree_1): Output DIE references.
19788 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19789 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19790 (output_function): Stream DECL_DEBUG_ARGS.
19791 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19792 Stream DECL_ABSTRACT_ORIGIN.
19793 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19794 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19795 DECL_CONTEXT for file-scope decls.
19796 * lto-streamer.h (struct dref_entry): Declare.
19797 (dref_queue): Likewise.
19798 * cfgexpand.c (pass_expand::execute): Do not call the
19799 outlining_inline_function hook here.
19800 * lto-wrapper.c (debug_obj): New global.
19801 (tool_cleanup): Unlink it if required.
19802 (debug_objcopy): New function.
19803 (run_gcc): Handle early debug sections in the IL files by
19804 extracting them to separate files, partially linkin them and
19805 feeding the result back as result to the linker.
19806 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19807 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19808 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19809 sections into a separate segment.
19810 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19812 (darwin_asm_dwarf_section): Likewise.
19813 (darwin_asm_output_dwarf_offset): Likewise.
19814 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19816 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19817 Alan Hayward <alan.hayward@arm.com>
19818 David Sherwood <david.sherwood@arm.com>
19820 * read-md.h (md_reader::record_potential_iterator_use): Replace
19821 pointer argument with an rtx and an index.
19822 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19823 (apply_mode_iterator): Likewise.
19824 (apply_code_iterator): Likewise.
19825 (apply_int_iterator): Likewise.
19826 (apply_subst_iterator): Likewise.
19827 (record_iterator_use): Likewise.
19828 (record_attribute_use): Likewise.
19829 (md_reader::record_potential_iterator_use): Likewise. Update calls
19830 to record_iterator_use and apply_iterator.
19831 (iterator_use): Replace ptr with x and index.
19832 (attribute_use): Likewise.
19833 (apply_attribute_uses): Update calls to apply_iterator.
19834 (apply_iterators): Likewise. Update initialization of iterator_use.
19835 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19836 and record_potential_iterator_use.
19837 (rtx_reader::read_rtx_operand): Likewise.
19839 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19840 Alan Hayward <alan.hayward@arm.com>
19841 David Sherwood <david.sherwood@arm.com>
19843 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19846 2017-08-21 Richard Biener <rguenther@suse.de>
19848 PR middle-end/81884
19849 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19850 at struct end conservatively when comparing common bases.
19852 2017-08-21 Richard Biener <rguenther@suse.de>
19854 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
19855 (mem_ref_in_stmt): Remove.
19856 (determine_max_movement): Use ref index to get at the reference.
19857 (invariantness_dom_walker::before_dom_children): Deal with
19858 lim data already initialized.
19859 (gather_mem_refs_stmt): Initialize lim data and record ref index.
19861 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19863 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
19864 (TARGET_ISA_ROUND): Ditto.
19865 (TARGET_ROUND): Ditto.
19866 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
19867 * config/i386/i386.md: Ditto.
19868 * config/i386/sse.md: Ditto.
19869 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
19870 with OPTION_MASK_ISA_SSE4_1.
19872 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19875 * doc/extend.texi (x86 Built-in Functions): Correct the name of
19876 __builtin_ia32_lzcnt_u16.
19878 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
19881 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
19882 (rs6000_set_current_function): Rewrite function to use it.
19884 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
19887 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
19888 and TYPE_WARN_IF_NOT_ALIGN.
19889 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
19890 (handle_warn_if_not_align): New.
19891 (place_union_field): Call handle_warn_if_not_align.
19892 (place_field): Call handle_warn_if_not_align.
19893 Copy TYPE_WARN_IF_NOT_ALIGN.
19894 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
19895 (layout_type): Likewise.
19896 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
19898 (tree_decl_common): Add warn_if_not_align.
19899 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
19900 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
19901 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
19902 (DECL_WARN_IF_NOT_ALIGN): Likewise.
19903 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
19904 * doc/extend.texi: Document warn_if_not_aligned attribute.
19905 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
19907 2017-08-17 Martin Liska <mliska@suse.cz>
19910 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
19911 (get_data_dependence): Use it as pointer type.
19912 (distribute_loop): Likewise.
19914 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19916 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
19917 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
19918 (p8_vmrgow_<mode>_direct): New define_insn.
19919 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
19920 handle endianness for vmrgew and vmrgow permute patterns.
19922 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19924 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
19925 * config/rs6000/rs6000-cpus.def: Remove comment.
19926 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
19927 (POWERPC_MASKS): Likewise.
19928 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
19929 use of TARGET_VSX_TIMODE.
19930 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
19931 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
19932 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
19933 (rs6000_option_override_internal): Remove dead code.
19934 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
19935 (rs6000_legitimize_reload_address): Likewise.
19936 (rs6000_legitimate_address_p): Likewise.
19937 (rs6000_opt_masks): Delete "vsx-timode".
19938 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
19939 from function comment.
19940 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
19941 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
19942 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
19944 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
19945 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
19946 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
19947 (VSX_TI): Likewise.
19949 (define_peephole2): Likewise.
19951 2017-08-17 Martin Sebor <msebor@redhat.com>
19954 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
19955 past the end of an array.
19956 (test_pp_format): Add test cases.
19958 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
19960 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
19961 missing ECF_NOTHROW flags.
19963 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19966 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
19967 operands residing in integer registers.
19968 (*vsx_le_perm_load_<mode>): Likewise.
19969 (*vsx_le_perm_store_<mode>): Likewise.
19970 (define_peephole2): Add peepholes to optimize the above.
19972 2017-08-17 Marek Polacek <polacek@redhat.com>
19974 PR middle-end/81814
19975 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
19976 to mimic what shorten_compare did. Change the return type to bool.
19977 (fold_cond_expr_with_comparison): Update call to
19978 operand_equal_for_comparison_p.
19979 (fold_ternary_loc): Likewise.
19981 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
19983 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
19985 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
19987 2017-08-17 Richard Biener <rguenther@suse.de>
19989 * tree-ssa-structalias.c (solve_graph): When propagating
19990 to successors update the graphs succ edges and avoid duplicate work.
19992 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
19995 * config/i386/i386.c (ix86_option_override_internal): Save target
19996 specific options after ix86_stack_protector_guard_reg was changed.
19998 2017-08-17 Richard Biener <rguenther@suse.de>
20000 PR tree-optimization/81827
20001 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20002 (new_var_info): Initialize it conservatively.
20003 (get_call_vi): Mark register vars.
20004 (new_scalar_tmp_constraint_exp): Likewise.
20005 (handle_rhs_call): Likewise.
20006 (handle_const_call): Likewise.
20007 (create_function_info_for): Likewise.
20008 (solve_constraints): Sort varinfos to separate register from
20009 non-register vars to pack points-to solution bitmaps during
20012 2017-08-17 Marek Polacek <polacek@redhat.com>
20014 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20016 2017-08-17 Richard Biener <rguenther@suse.de>
20018 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20019 to true when overflow is undefined and we saturated the result.
20021 2017-08-17 Alan Modra <amodra@gmail.com>
20024 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20025 Don't use store multiple if only one reg needs saving.
20026 (interesting_frame_related_regno): New function.
20027 (rs6000_frame_related): Don't emit frame info for regs that
20029 (rs6000_emit_epilogue): Likewise.
20031 2017-08-16 Nathan Sidwell <nathan@acm.org>
20033 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20034 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20035 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20036 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20037 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20038 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20039 (verify_type): Adjust for TYPE_BINFO move.
20040 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20041 process TYPE_BINFO directly.
20042 (hash_tree): Likewise.
20043 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20045 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20048 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20050 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20052 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20055 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20056 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20057 Add ix86_binary_operator_ok to insn constraint.
20058 (*iordi_1_bts): Ditto.
20059 (*xordi_1_btc): Ditto.
20060 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20061 Update corresponding peephole2 pattern.
20065 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20067 PR tree-optimization/81832
20068 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20069 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20071 2017-08-16 Marek Polacek <polacek@redhat.com>
20074 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20075 regarding size_zero_node.
20077 2017-08-16 Martin Liska <mliska@suse.cz>
20080 * config.gcc: Respect previously set extra_objs in case
20083 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20085 PR tree-optimization/81835
20086 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20087 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20088 not depend on the phi.
20090 2017-08-16 Alan Modra <amodra@gmail.com>
20092 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20095 2017-08-16 Alan Modra <amodra@gmail.com>
20097 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20098 (save_reg_p): ..into this. Update all callers.
20099 (first_reg_to_save): Simplify.
20101 2017-08-16 Alan Modra <amodra@gmail.com>
20103 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20106 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20110 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20111 constant count if that count is less than 32.
20113 2017-08-15 Nathan Sidwell <nathan@acm.org>
20115 * gcc.c (execute): Emit friendlier message if inferior is killed
20116 by an external cause.
20118 2017-08-15 Richard Biener <rguenther@suse.de>
20120 PR tree-optimization/81790
20121 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20122 CONSTRUCTORs from simplifying and VN.
20124 2017-08-14 Martin Sebor <msebor@redhat.com>
20126 * builtin-attrs.def: Add comments.
20128 2017-08-14 Martin Sebor <msebor@redhat.com>
20131 * doc/extend.texi (attribute nonstring): Document new attribute.
20133 2017-08-14 Martin Sebor <msebor@redhat.com>
20136 * tree-diagnostic.c (default_tree_printer): Handle %G.
20137 * gimple-pretty-print.h (percent_G_format): Declare new function.
20138 * gimple-pretty-print.c (percent_G_format): Define.
20139 * tree-pretty-print.c (percent_K_format): Add argument.
20141 2017-08-14 Martin Sebor <msebor@redhat.com>
20143 PR translation/79998
20144 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20145 Remove a stray space.
20147 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20150 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20151 (*iordi_1_bts): Ditto.
20152 (*xordi_1_btc): Ditto.
20154 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20157 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20158 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20160 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20161 quoted strings, and make more translator-friendly.
20162 (darwin_rs6000_override_options): Likewise.
20163 (rs6000_option_override_internal): Likewise.
20164 (rs6000_return_in_memory): Fix overlong line.
20165 (init_cmulative_args): Use quoted strings, and make more
20166 translator-friendly.
20167 (rs6000_pass_by_reference): Fix overlong line.
20168 (def_builtin): Use quoted strings.
20169 (altivec_expand_predicate_builtin): Use quoted strings, and make
20170 more translator-friendly.
20171 (htm_expand_builtin): Use quoted strings.
20172 (cpu_expand_builtin): Use quoted strings, and make more
20173 translator-friendly.
20174 (altivec_expand_builtin): Likewise.
20175 (paired_expand_predicate_builtin): Likewise.
20176 (rs6000_invalid_builtin): Likewise.
20177 (builtin_function_type): Use quoted strings.
20178 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20179 more translator-friendly.
20180 (rs6000_trampoline_init): Likewise.
20181 (rs6000_handle_altivec_attribute): Likewise.
20182 (rs6000_inner_target_options): Use quoted strings.
20183 (rs6000_disable_incompatible_switches): Likewise.
20184 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20185 strings, and make more translator-friendly.
20186 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20188 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20190 PR tree-optimization/81799
20191 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20192 cond_expr to simple gimple operand.
20194 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20196 PR middle-end/46932
20197 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20199 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20203 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20204 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20205 TARGET_GASISR_PROLOGUES.
20206 * config/avr/avr.c (avr_option_override): Same.
20207 (avr_pass_pre_proep::execute): Same.
20209 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20212 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20213 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20215 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20217 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20218 define_insn_and_split. Split to a memory load from 0 in
20219 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20220 using PTR mode iterator.
20221 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20222 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20223 (*add_tp_<mode>): Redefine as define_insn_and_split.
20224 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20225 address space. Merge with *add_tp_x32 using PTR mode iterator.
20226 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20227 Split to an add with a memory load from 0 in
20228 DEFAULT_TLS_SEG_REG address space.
20230 2017-08-12 Andrew Pinski <apinski@cavium.com>
20232 * config/aarch64/aarch64-option-extensions.def (rdma):
20233 Fix feature string to what Linux prints out in /proc/cpuinfo.
20235 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20238 * dwarf2out.c (modified_type_die): For C typedef types that have
20239 an ultimate origin, process the ultimate origin instead of the
20241 (gen_typedef_die): Assert that input DECLs have no ultimate
20243 (gen_type_die_with_usage): For typedef variants that have an
20244 ultimate origin, just call gen_decl_die on the original DECL.
20245 (process_scope_var): Avoid creating DIEs for local typedefs and
20246 concrete static variables.
20248 2017-08-12 Alan Modra <amodra@gmail.com>
20252 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20253 match gnu-user.h startfile.
20254 (ENDFILE_LINUX_SPEC): Similarly.
20256 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20259 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20261 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20263 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20264 * config/aarch64/aarch64.md (mov<mode>): Change.
20265 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20266 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20267 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20269 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20271 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20272 for storage order barriers.
20274 2017-08-11 Martin Liska <mliska@suse.cz>
20277 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20278 variables of void type.
20280 2017-08-11 Martin Liska <mliska@suse.cz>
20282 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20283 TARGET_SUPPORTS_ALIASES.
20284 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20285 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20286 (optimize_weakref): Likewise.
20287 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20288 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20289 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20291 2017-08-11 Martin Liska <mliska@suse.cz>
20294 * config/i386/i386.c (make_resolver_func): Do complete
20295 refactoring of the function.
20297 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20300 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20301 * config/i386/i386.c (ix86_stack_protect_guard): Use
20302 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20303 * doc/invoke.texi (x86 Options): Document
20304 -mstack-protector-guard-symbol= option.
20306 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20308 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20309 * config/i386/i386.c (ix86_split_stack_guard): New function.
20310 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20311 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20312 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20313 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20314 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20315 (split_stack_space_check): Call ix86_split_stack_guard.
20317 2017-08-10 Martin Sebor <msebor@redhat.com>
20319 * print-tree.c (print_node): Print location using the established
20321 Replace spaces with colons.
20322 (debug_raw, debug): Ditto.
20324 2017-08-10 Martin Sebor <msebor@redhat.com>
20327 * pretty-print.c (pp_format): Correct the handling of %s precision.
20329 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20332 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20334 (ix86_finalize_stack_frame_flags): This. Also clear
20335 frame_pointer_needed if -fno-omit-frame-pointer is used without
20337 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20338 with ix86_finalize_stack_frame_flags.
20339 (ix86_expand_epilogue): Likewise.
20340 (ix86_expand_split_stack_prologue): Likewise.
20341 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20343 2017-08-10 Martin Liska <mliska@suse.cz>
20346 * c-attribs.c (handle_target_attribute):
20347 Report warning for an empty string argument of target attribute.
20349 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20352 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20354 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20355 or DECL_NONLOCAL labels.
20356 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20357 or DECL_NONLOCAL labels here.
20359 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20361 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20362 to indicate when early gimple folding has been disabled.
20363 (rs6000_gimple_fold_builtin): Add debug content.
20364 (rs6000_invalid_builtin): Fix whitespace.
20365 (rs6000_expand_builtin): Fix whitespace.
20366 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20368 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20371 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20372 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20374 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20376 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20377 (qdf24xx): Likewise.
20378 * config/aarch64/aarch64.md: Include falkor.md.
20379 * config/aarch64/falkor.md: New.
20381 2017-08-09 Marek Polacek <polacek@redhat.com>
20384 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20385 * diagnostic.c (emit_diagnostic): Add a comment.
20386 (emit_diagnostic_valist): New function.
20388 2017-08-09 Marek Polacek <polacek@redhat.com>
20391 * input.c (make_location): New overload.
20392 * input.h (make_location): Declare.
20394 2017-08-08 Alan Modra <amodra@gmail.com>
20395 H.J. Lu <hongjiu.lu@intel.com>
20398 * gcc.c (NO_PIE_SPEC): Delete.
20399 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20401 (LINK_PIE_SPEC): ..to here.
20402 (LINK_COMMAND_SPEC): Support -no-pie.
20403 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20404 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20406 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20407 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20408 (ENDFILE_CRTEND_SPEC): Similarly.
20410 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20413 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20414 (mstack-protector-guard-offset=): Ditto.
20415 * config/i386/i386.c (ix86_option_override): Handle
20416 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20418 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20419 ix86_stack_protect_guard_offset variables.
20420 (TARGET_STACK_PROTECT_GUARD): Always define.
20421 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20422 and -mstack-protector-guard-offset= options.
20424 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20426 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20427 boundary case for the last candidate.
20429 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20431 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20432 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20434 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20436 PR middle-end/19706
20437 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20438 * config/aarch64/aarch64-builtins.c
20439 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20440 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20441 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20443 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20444 Andrew Pinski <pinskia@gmail.com>
20446 PR middle-end/19706
20447 * internal-fn.def (XORSIGN): New.
20448 * optabs.def (xorsign_optab): New.
20449 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20450 (convert_expand_mult_copysign): New.
20451 (pass_optimize_widening_mul::execute): Call
20452 convert_expand_mult_copysign.
20454 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20456 PR tree-optimization/81354
20457 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20458 Insert on edges rather than explicitly creating landing pads.
20459 (analyze_candidates_and_replace): Commit edge inserts.
20461 2017-08-08 Richard Biener <rguenther@suse.de>
20463 PR middle-end/81719
20464 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20465 (expand_simple_operations): Also look through ADDR_EXPRs with
20466 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20468 2017-08-08 Richard Biener <rguenther@suse.de>
20470 PR tree-optimization/81723
20471 * tree-vect-slp.c (struct bst_traits): New hash traits.
20472 (bst_fail): New global.
20473 (vect_build_slp_tree_2): New worker, split out from ...
20474 (vect_build_slp_tree): ... this now wrapping it with using
20475 bst_fail set to cache SLP tree build fails. Properly handle
20477 (vect_analyze_slp_instance): Allocate and free bst_fail.
20479 2017-08-08 Martin Liska <mliska@suse.cz>
20482 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20483 LABEL_DECLs that can be from a different function.
20485 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20487 PR tree-optimization/81744
20488 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20489 loop's number of iterations.
20491 2017-08-08 Martin Liska <mliska@suse.cz>
20493 * asan.c: Include header files.
20494 * attribs.c (build_decl_attribute_variant): New function moved
20496 (build_type_attribute_qual_variant): Likewise.
20497 (cmp_attrib_identifiers): Likewise.
20498 (simple_cst_list_equal): Likewise.
20499 (omp_declare_simd_clauses_equal): Likewise.
20500 (attribute_value_equal): Likewise.
20501 (comp_type_attributes): Likewise.
20502 (build_type_attribute_variant): Likewise.
20503 (lookup_ident_attribute): Likewise.
20504 (remove_attribute): Likewise.
20505 (merge_attributes): Likewise.
20506 (merge_type_attributes): Likewise.
20507 (merge_decl_attributes): Likewise.
20508 (merge_dllimport_decl_attributes): Likewise.
20509 (handle_dll_attribute): Likewise.
20510 (attribute_list_equal): Likewise.
20511 (attribute_list_contained): Likewise.
20512 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20513 (lookup_attribute_by_prefix): Likewise.
20514 * bb-reorder.c: Include header files.
20515 * builtins.c: Likewise.
20516 * calls.c: Likewise.
20517 * cfgexpand.c: Likewise.
20518 * cgraph.c: Likewise.
20519 * cgraphunit.c: Likewise.
20520 * convert.c: Likewise.
20521 * dwarf2out.c: Likewise.
20522 * final.c: Likewise.
20523 * fold-const.c: Likewise.
20524 * function.c: Likewise.
20525 * gimple-expr.c: Likewise.
20526 * gimple-fold.c: Likewise.
20527 * gimple-pretty-print.c: Likewise.
20528 * gimple.c: Likewise.
20529 * gimplify.c: Likewise.
20530 * hsa-common.c: Likewise.
20531 * hsa-gen.c: Likewise.
20532 * internal-fn.c: Likewise.
20533 * ipa-chkp.c: Likewise.
20534 * ipa-cp.c: Likewise.
20535 * ipa-devirt.c: Likewise.
20536 * ipa-fnsummary.c: Likewise.
20537 * ipa-inline.c: Likewise.
20538 * ipa-visibility.c: Likewise.
20540 * lto-cgraph.c: Likewise.
20541 * omp-expand.c: Likewise.
20542 * omp-general.c: Likewise.
20543 * omp-low.c: Likewise.
20544 * omp-offload.c: Likewise.
20545 * omp-simd-clone.c: Likewise.
20546 * opts-global.c: Likewise.
20547 * passes.c: Likewise.
20548 * predict.c: Likewise.
20549 * sancov.c: Likewise.
20550 * sanopt.c: Likewise.
20551 * symtab.c: Likewise.
20552 * toplev.c: Likewise.
20553 * trans-mem.c: Likewise.
20554 * tree-chkp.c: Likewise.
20555 * tree-eh.c: Likewise.
20556 * tree-into-ssa.c: Likewise.
20557 * tree-object-size.c: Likewise.
20558 * tree-parloops.c: Likewise.
20559 * tree-profile.c: Likewise.
20560 * tree-ssa-ccp.c: Likewise.
20561 * tree-ssa-live.c: Likewise.
20562 * tree-ssa-loop.c: Likewise.
20563 * tree-ssa-sccvn.c: Likewise.
20564 * tree-ssa-structalias.c: Likewise.
20565 * tree-ssa.c: Likewise.
20566 * tree-streamer-in.c: Likewise.
20567 * tree-vectorizer.c: Likewise.
20568 * tree-vrp.c: Likewise.
20569 * tsan.c: Likewise.
20570 * ubsan.c: Likewise.
20571 * varasm.c: Likewise.
20572 * varpool.c: Likewise.
20573 * tree.c: Remove functions moved to attribs.[ch].
20574 * tree.h: Likewise.
20575 * config/aarch64/aarch64.c: Add attrs.h header file.
20576 * config/alpha/alpha.c: Likewise.
20577 * config/arc/arc.c: Likewise.
20578 * config/arm/arm.c: Likewise.
20579 * config/avr/avr.c: Likewise.
20580 * config/bfin/bfin.c: Likewise.
20581 * config/c6x/c6x.c: Likewise.
20582 * config/cr16/cr16.c: Likewise.
20583 * config/cris/cris.c: Likewise.
20584 * config/darwin.c: Likewise.
20585 * config/epiphany/epiphany.c: Likewise.
20586 * config/fr30/fr30.c: Likewise.
20587 * config/frv/frv.c: Likewise.
20588 * config/ft32/ft32.c: Likewise.
20589 * config/h8300/h8300.c: Likewise.
20590 * config/i386/winnt.c: Likewise.
20591 * config/ia64/ia64.c: Likewise.
20592 * config/iq2000/iq2000.c: Likewise.
20593 * config/lm32/lm32.c: Likewise.
20594 * config/m32c/m32c.c: Likewise.
20595 * config/m32r/m32r.c: Likewise.
20596 * config/m68k/m68k.c: Likewise.
20597 * config/mcore/mcore.c: Likewise.
20598 * config/microblaze/microblaze.c: Likewise.
20599 * config/mips/mips.c: Likewise.
20600 * config/mmix/mmix.c: Likewise.
20601 * config/mn10300/mn10300.c: Likewise.
20602 * config/moxie/moxie.c: Likewise.
20603 * config/msp430/msp430.c: Likewise.
20604 * config/nds32/nds32-isr.c: Likewise.
20605 * config/nds32/nds32.c: Likewise.
20606 * config/nios2/nios2.c: Likewise.
20607 * config/nvptx/nvptx.c: Likewise.
20608 * config/pa/pa.c: Likewise.
20609 * config/pdp11/pdp11.c: Likewise.
20610 * config/powerpcspe/powerpcspe.c: Likewise.
20611 * config/riscv/riscv.c: Likewise.
20612 * config/rl78/rl78.c: Likewise.
20613 * config/rx/rx.c: Likewise.
20614 * config/s390/s390.c: Likewise.
20615 * config/sh/sh.c: Likewise.
20616 * config/sol2.c: Likewise.
20617 * config/sparc/sparc.c: Likewise.
20618 * config/spu/spu.c: Likewise.
20619 * config/stormy16/stormy16.c: Likewise.
20620 * config/tilegx/tilegx.c: Likewise.
20621 * config/tilepro/tilepro.c: Likewise.
20622 * config/v850/v850.c: Likewise.
20623 * config/vax/vax.c: Likewise.
20624 * config/visium/visium.c: Likewise.
20625 * config/xtensa/xtensa.c: Likewise.
20627 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20630 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20631 constraints since the -mupper-regs-* switches have been
20633 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20634 into a vector from a double word element that was extracted from
20635 another vector, and eliminate extra XXPERMDI instructions.
20636 (vsx_concat_<mode>_2): Likewise.
20637 (vsx_concat_<mode>_3): Likewise.
20638 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20639 concat to allow optimizing inserts from previous extracts.
20641 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20643 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20644 memory reference to a SSP offset in TLS address space.
20645 (ix86_print_operand) <case '@'>: Remove.
20646 (ix86_print_operand_punct_valid_p): Remove '@' code.
20647 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20648 UNSPEC_SP_TLS_TEST.
20649 (stack_tls_protect_set_<mode>): Remove.
20650 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20651 (stack_tls_protect_test_<mode>): Remove.
20652 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20654 2017-08-07 Olivier Hainque <hainque@adacore.com>
20657 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20659 2017-08-07 Douglas Rupp <rupp@adacore.com>
20661 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20662 variable was referenced as multidir in command.
20664 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20667 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20670 2017-08-07 Martin Liska <mliska@suse.cz>
20672 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20673 * config/rl78/rl78.c: Add include of attribs.h.
20674 * config/sh/sh.c: Likewise.
20675 * config/v850/v850.c: Likewise.
20677 2017-08-07 Tom de Vries <tom@codesourcery.com>
20679 PR middle-end/78266
20680 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20682 2017-08-07 Martin Liska <mliska@suse.cz>
20684 * config/mips/mips.c: Include attribs.h.
20686 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20689 * doc/invoke.texi: Document change in behvaior for -Ofast for
20692 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20694 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20696 (aarch64_pop_regs): Likewise.
20697 (aarch64_gen_load_pair): Likewise.
20698 (aarch64_save_callee_saves): Likewise.
20699 (aarch64_restore_callee_saves): Likewise.
20701 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20703 * config/i386/i386.c: Revert the last change.
20705 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20708 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20710 (ix86_finalize_stack_frame_flags): This. Also clear
20711 frame_pointer_needed if -fno-omit-frame-pointer is used without
20713 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20714 with ix86_finalize_stack_frame_flags.
20715 (ix86_expand_epilogue): Likewise.
20716 (ix86_expand_split_stack_prologue): Likewise.
20718 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20721 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20722 to P_AES for Westmere.
20724 2017-08-07 Jonathan Yong <10walls@gmail.com>
20726 * config/i386/mingw.opt (fset-stack-executable): Removed.
20727 * config/i386/cygming.opt (fset-stack-executable): Moved
20729 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20731 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20733 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20735 2017-08-07 Marek Polacek <polacek@redhat.com>
20737 PR middle-end/81737
20738 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20740 2017-08-07 Martin Liska <mliska@suse.cz>
20742 * attribs.h (canonicalize_attr_name): New function.
20743 (cmp_attribs): Move from c-format.c and adjusted.
20744 (is_attribute_p): Moved from tree.h.
20745 * tree-inline.c: Add new includes.
20746 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20747 (private_is_attribute_p): Remove.
20748 (private_lookup_attribute): Likewise.
20749 (private_lookup_attribute_by_prefix): Simplify.
20750 (remove_attribute): Use is_attribute_p.
20751 * tree.h: Remove removed declarations.
20753 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20755 PR middle-end/81698
20756 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20757 instead of computing it in the function. Formatting fix.
20758 (expand_case): Don't rely on default_edge being the first edge,
20759 clear it if removing it, pass default_edge to
20760 emit_case_dispatch_table.
20761 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20764 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20766 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20767 insn in the function, emit NOP after the insn.
20769 2017-08-06 Tom de Vries <tom@codesourcery.com>
20771 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20774 2017-08-06 Tom de Vries <tom@codesourcery.com>
20776 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20779 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20781 PR tree-optimization/57371
20782 * match.pd: New pattern.
20784 2017-08-04 Marek Polacek <polacek@redhat.com>
20786 PR middle-end/81695
20787 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20788 perform the computation in offset_int.
20790 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20792 PR tree-optimization/81136
20793 * tree-vectorizer.h: Include tree-hash-traits.h.
20794 (vec_base_alignments): New typedef.
20795 (vec_info): Add a base_alignments field.
20796 (vect_record_base_alignments): Declare.
20797 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20799 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20800 (create_data_ref): Add an is_conditional_in_stmt argument.
20801 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20802 the is_conditional_in_stmt field.
20803 (data_ref_loc): Add an is_conditional_in_stmt field.
20804 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20805 (find_data_references_in_stmt): Update call to create_data_ref.
20806 (graphite_find_data_references_in_stmt): Likewise.
20807 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20808 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20809 (vect_record_base_alignment): New function.
20810 (vect_record_base_alignments): Likewise.
20811 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20812 for nested statements even if we fail to compute a misalignment.
20813 Use pooled base alignments for unconditional references.
20814 (vect_find_same_alignment_drs): Compare base addresses instead
20816 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20817 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20819 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20821 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20822 Add an explicit name for the enum. Use auto_vec for slp_instances
20823 and grouped_stores.
20824 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20826 (_bb_vec_info): Add a constructor and destructor.
20827 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20828 (destroy_loop_vec_info): Delete.
20829 (vect_destroy_datarefs): Likewise.
20830 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20831 (vec_info::vec_info): New function.
20832 (vec_info::~vec_info): Likewise.
20833 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20834 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20835 destroy_loop_vec_info.
20836 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20837 (_loop_vec_info::_loop_vec_info): ...this.
20838 (destroy_loop_vec_info): Replace with...
20839 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20840 the stmt_vec_infos. Leave handling of vec_info information to its
20841 destructor. Remove explicit vector releases.
20842 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20843 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20844 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20845 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20846 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20847 (destroy_bb_vec_info): Replace with...
20848 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20849 information to its destructor.
20850 (vect_slp_analyze_bb_1): Use new and delete instead of
20851 new_bb_vec_info and destroy_bb_vec_info.
20852 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
20855 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20857 * tree-data-ref.h (subscript): Add access_fn field.
20858 (data_dependence_relation): Add could_be_independent_p.
20859 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
20860 (same_access_functions): Move to tree-data-ref.c.
20861 * tree-data-ref.c (ref_contains_union_access_p): New function.
20862 (access_fn_component_p): Likewise.
20863 (access_fn_components_comparable_p): Likewise.
20864 (dr_analyze_indices): Add a reference to access_fn_component_p.
20865 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
20867 (constant_access_functions): Likewise.
20868 (add_other_self_distances): Likewise.
20869 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
20870 (initialize_data_dependence_relation): Use XCNEW and remove
20871 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
20872 of access functions that have the same type. Allow the
20873 subsequence to end with different bases in some circumstances.
20874 Record the chosen access functions in SUB_ACCESS_FN.
20875 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
20876 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
20877 (subscript_dependence_tester_1): Likewise dra and drb.
20878 (build_classic_dist_vector): Update calls accordingly.
20879 (subscript_dependence_tester): Likewise.
20880 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
20881 DDR_COULD_BE_INDEPENDENT_P.
20882 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
20883 comp_alias_ddrs instead of may_alias_ddrs.
20884 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
20886 (vect_analyze_data_ref_dependence): Use it if
20887 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
20888 distance vectors if that fails.
20889 (dependence_distance_ge_vf): New function.
20890 (vect_prune_runtime_alias_test_list): Use it. Don't clear
20891 LOOP_VINFO_MAY_ALIAS_DDRS.
20893 2017-08-04 Richard Biener <rguenther@suse.de>
20895 PR middle-end/81705
20896 * fold-const.c (fold_binary_loc): Properly restrict
20897 minus_var0 && minus_var1 case when associating undefined overflow
20900 2017-08-04 Tom de Vries <tom@codesourcery.com>
20902 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
20904 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20906 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20907 Don't start diagnostic messages with a capital letter.
20908 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20910 (rs6000_invalid_builtin): Likewise.
20911 (rs6000_trampoline_init): Likewise.
20913 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20916 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
20917 after setting changeable df flags.
20919 2017-08-03 Richard Biener <rguenther@suse.de>
20921 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
20922 up if the use is in USE - X.
20924 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
20926 * toplev.c (dumpfile.h): New include.
20927 (internal_error_reentered): New static function. Use it...
20928 (internal_error_function): ...here to handle reentered internal_error.
20930 2017-08-03 Richard Biener <rguenther@suse.de>
20932 PR middle-end/81148
20933 * fold-const.c (split_tree): Add minus_var and minus_con
20934 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
20935 here but always use minus_*.
20936 (associate_trees): Assert we never associate with MINUS_EXPR
20937 and NULL first operand. Do not recurse for PLUS_EXPR operands
20938 when associating as MINUS_EXPR either.
20939 (fold_binary_loc): Track minus_var and minus_con.
20941 2017-08-03 Tom de Vries <tom@codesourcery.com>
20944 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
20945 ACCEL_COMPILER, apply finish_options on
20946 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
20948 2017-08-03 Tom de Vries <tom@codesourcery.com>
20951 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
20952 function_entry_patch_area_size > 0.
20954 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20957 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
20958 instead of 10??LU, perform unit multiplication in wide_int,
20959 don't change alloc_object_size_limit if the limit is larger
20962 PR tree-optimization/81655
20963 PR tree-optimization/81588
20964 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
20965 the case when ranges[i].low and high are 1 for unsigned type with
20968 PR middle-end/81052
20969 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
20970 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
20972 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20974 * tree-vrp.h: Add include guard.
20976 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
20979 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
20980 (ud2): New insn pattern.
20981 * config/i386/i386.c (ix86_expand_epilogue):
20982 For naked functions, generate ud2 instead of trap insn.
20984 2017-08-02 Marek Polacek <polacek@redhat.com>
20987 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
20989 2017-08-02 Tom de Vries <tom@codesourcery.com>
20990 Cesar Philippidis <cesar@codesourcery.com>
20992 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
20993 Add missing edge probabilities.
20995 2017-08-02 Tamar Christina <tamar.christina@arm.com>
20997 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
20998 Correct endianness.
21000 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21002 PR middle-end/79499
21003 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21004 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21005 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21006 prologue_seq sequences - if any.
21008 2017-08-02 Richard Biener <rguenther@suse.de>
21010 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21011 via vectors if supported, integer extracts via punning if supported
21012 or otherwise vector extracts.
21014 2017-08-02 Richard Biener <rguenther@suse.de>
21016 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21018 (bitmap_insert_into_set): ... this.
21020 2017-08-02 Richard Biener <rguenther@suse.de>
21022 PR tree-optimization/81633
21024 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21026 PR tree-optimization/71752
21027 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21029 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21031 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21032 (machine_function::call_ms2sysv_pad_out): Remove field.
21033 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21034 (ix86_compute_frame_layout): Likewise.
21036 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21039 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21040 attribute with interrupt attribute.
21042 2017-08-01 Andrew Pinski <apinski@cavium.com>
21044 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21045 BIT_INSERT_EXPR's operand 1
21046 to see if the types precision matches.
21048 2017-08-01 Martin Liska <mliska@suse.cz>
21050 PR middle-end/70140
21051 * builtins.c (expand_builtin_memcpy_args): Remove.
21052 (expand_builtin_memcpy): Call newly added function
21053 expand_builtin_memory_copy_args.
21054 (expand_builtin_memcpy_with_bounds): Likewise.
21055 (expand_builtin_mempcpy): Remove last argument.
21056 (expand_builtin_mempcpy_with_bounds): Likewise.
21057 (expand_builtin_memory_copy_args): New function created from
21058 expand_builtin_mempcpy_args with small modifications.
21059 (expand_builtin_mempcpy_args): Remove.
21060 (expand_builtin_stpcpy): Remove unused argument.
21061 (expand_builtin): Likewise.
21062 (expand_builtin_with_bounds): Likewise.
21064 2017-08-01 Martin Liska <mliska@suse.cz>
21067 Make mempcpy more optimal (PR middle-end/70140).
21069 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21072 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21073 __builtin_vec_cmpne verify both arguments are compatible vectors
21074 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21075 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21076 move computation of aligned to after checking the argument types.
21080 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21081 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21084 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21085 Doug Rupp <rupp@adacore.com>
21086 Olivier Hainque <hainque@adacore.com>
21088 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21089 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21091 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21092 for TARGET_OS_CPP_BUILTIN.
21093 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21094 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21096 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21097 passing required abi options to the assembler for EABI configurations.
21098 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21099 of .text.hot and .text.unlikely sections for kernel modules when
21100 using ARM style exceptions.
21101 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21102 options. Add EXTRA_CC1_SPEC.
21103 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21105 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21107 (ARM_TARGET2_DWARF_FORMAT): Define.
21108 * config/arm/t-vxworks: Adjust multilib control to removal of the
21109 Diab command line options.
21111 2017-08-01 Martin Liska <mliska@suse.cz>
21113 PR gcov-profile/81561
21114 * gcov.c (unblock): Make unblocking safe as we need to preserve
21115 index correspondence of blocks and block_lists.
21117 2017-08-01 Richard Biener <rguenther@suse.de>
21119 PR tree-optimization/81181
21120 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21121 (compute_antic): ... end of iteration here.
21123 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21125 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21126 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21127 (ftree-slp-vectorize): Likewise.
21128 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21129 can no longer be set independent of flag_tree_loop_vectorize.
21130 * omp-general.c (emp_max_vf): Likewise.
21131 * opts.c (enable_fdo_optimizations): Remove references to
21132 flag_tree_vectorize, these are now implicit.
21133 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21134 and leave it for the options machinery.
21136 2017-08-01 Martin Liska <mliska@suse.cz>
21138 PR middle-end/70140
21139 * builtins.c (expand_builtin_memcpy_args): Remove.
21140 (expand_builtin_memcpy): Call newly added function
21141 expand_builtin_memory_copy_args.
21142 (expand_builtin_memcpy_with_bounds): Likewise.
21143 (expand_builtin_mempcpy): Remove last argument.
21144 (expand_builtin_mempcpy_with_bounds): Likewise.
21145 (expand_builtin_memory_copy_args): New function created from
21146 expand_builtin_mempcpy_args with small modifications.
21147 (expand_builtin_mempcpy_args): Remove.
21148 (expand_builtin_stpcpy): Remove unused argument.
21149 (expand_builtin): Likewise.
21150 (expand_builtin_with_bounds): Likewise.
21152 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21155 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21156 print "ds:" only for immediates in generic address space.
21158 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21161 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21162 (ix86_function_ok_for_sibcall): Return false for naked functions.
21164 2017-08-01 Richard Biener <rguenther@suse.de>
21166 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21167 (compute_antic): Seed worklist with exit block predecessors.
21168 * cfganal.c (dfs_find_deadend): For a cycle return the source
21169 of the edge closing it.
21171 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21173 * config/aarch64/aarch64.c
21174 (aarch64_can_const_movi_rtx_p): Move 0 check.
21176 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21178 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21179 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21181 * match.pd: Ditto in address comparison pattern.
21183 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21185 PR tree-optimization/81627
21186 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21187 closed ssa form for store-store chain.
21189 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21191 PR tree-optimization/81620
21192 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21193 for store-store chain.
21195 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21197 PR tree-optimization/81588
21198 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21199 ranges[i].in_p, invert comparison code ccode. For >/>=,
21200 swap rhs1 and rhs2 and comparison code unconditionally,
21201 for </<= don't do that. Don't swap rhs1/rhs2 again if
21202 ranges[i].in_p, instead invert comparison code ccode if
21203 opcode or oe->rank is BIT_IOR_EXPR.
21206 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21207 a direct optab to conversion optab.
21208 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21209 with GET_MODE_INNER as last argument instead of optab_handler.
21210 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21211 vector extraction if possible and optab is available.
21212 * expr.c (store_constructor): Use convert_optab_handler instead
21213 of optab_handler. Use vector initialization from smaller
21214 vectors if possible and optab is available.
21215 * tree-vect-stmts.c (vectorizable_load): Likewise.
21216 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21217 now have two modes.
21218 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21219 of vec_init from half-sized vectors with the same element mode.
21220 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21221 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21222 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21223 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21224 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21225 after mode in gen_vec_extract* calls.
21226 (vec_extract<mode>): Renamed to ...
21227 (vec_extract<mode><ssescalarmodelower>): ... this.
21228 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21229 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21230 element mode after mode in gen_vec_init* calls.
21231 (VEC_INIT_HALF_MODE): New mode iterator.
21232 (vec_init<mode>): Renamed to ...
21233 (vec_init<mode><ssescalarmodelower>): ... this.
21234 (vec_init<mode><ssehalfvecmodelower>): New expander.
21235 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21236 (vec_extractv2sfsf): ... this.
21237 (vec_initv2sf): Renamed to ...
21238 (vec_initv2sfsf): ... this.
21239 (vec_extractv2si): Renamed to ...
21240 (vec_extractv2sisi): ... this.
21241 (vec_initv2si): Renamed to ...
21242 (vec_initv2sisi): ... this.
21243 (vec_extractv4hi): Renamed to ...
21244 (vec_extractv4hihi): ... this.
21245 (vec_initv4hi): Renamed to ...
21246 (vec_initv4hihi): ... this.
21247 (vec_extractv8qi): Renamed to ...
21248 (vec_extractv8qiqi): ... this.
21249 (vec_initv8qi): Renamed to ...
21250 (vec_initv8qiqi): ... this.
21251 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21252 (vec_init<mode>): Renamed to ...
21253 (vec_init<mode><VEC_base_l>): ... this.
21254 (vec_extract<mode>): Renamed to ...
21255 (vec_extract<mode><VEC_base_l>): ... this.
21256 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21257 (vec_initv2sfsf): ... this.
21258 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21259 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21260 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21261 element mode after mode in gen_vec_init* calls.
21262 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21263 (vec_init<mode><Vel>): ... this.
21264 (vec_extract<mode>): Renamed to ...
21265 (vec_extract<mode><Vel>): ... this.
21266 * config/aarch64/iterators.md (Vel): New mode attribute.
21267 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21268 Add element mode after mode in gen_vec_extract* calls.
21269 * config/s390/vector.md (non_vec_l): New mode attribute.
21270 (vec_extract<mode>): Renamed to ...
21271 (vec_extract<mode><non_vec_l>): ... this.
21272 (vec_init<mode>): Renamed to ...
21273 (vec_init<mode><non_vec_l>): ... this.
21274 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21275 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21277 * config/arm/iterators.md (V_elem_l): New mode attribute.
21278 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21279 (vec_extract<mode><V_elem_l>): ... this.
21280 (vec_extractv2di): Renamed to ...
21281 (vec_extractv2didi): ... this.
21282 (vec_init<mode>): Renamed to ...
21283 (vec_init<mode><V_elem_l>): ... this.
21284 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21285 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21286 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21287 Add element mode after gen_vec_extract* calls.
21288 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21289 (vec_init<mode><unitmode>): ... this.
21290 (vec_extract<mode>): Renamed to ...
21291 (vec_extract<mode><unitmode>): ... this.
21292 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21293 (vec_init<mode><unitmode>): ... this.
21294 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21295 (vec_initv2sfsf): ... this.
21296 (vec_extractv2sf): Renamed to ...
21297 (vec_extractv2sfsf): ... this.
21298 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21299 Add element mode after gen_vec_extract* calls.
21300 * config/mips/mips.md (unitmode): New mode iterator.
21301 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21302 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21303 * config/spu/spu.md (inner_l): New mode attribute.
21304 (vec_init<mode>): Renamed to ...
21305 (vec_init<mode><inner_l>): ... this.
21306 (vec_extract<mode>): Renamed to ...
21307 (vec_extract<mode><inner_l>): ... this.
21308 * config/sparc/sparc.md (veltmode): New mode iterator.
21309 (vec_init<VMALL:mode>): Renamed to ...
21310 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21311 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21312 (vec_initv2sisi): ... this.
21313 (vec_initv2sf): Renamed to ...
21314 (vec_initv2sfsf): ... this.
21315 (vec_extractv2sf): Renamed to ...
21316 (vec_extractv2sfsf): ... this.
21317 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21318 (vec_init<mode>): Renamed to ...
21319 (vec_init<mode><VEC_base_l>): ... this.
21320 (vec_extract<mode>): Renamed to ...
21321 (vec_extract<mode><VEC_base_l>): ... this.
21322 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21323 (vec_initv2sfsf): ... this.
21324 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21325 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21326 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21327 gen_vec_init* calls.
21329 2017-08-01 Richard Biener <rguenther@suse.de>
21331 PR tree-optimization/81297
21332 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21333 TREE_OVERFLOW from INTEGER_CSTs.
21335 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21337 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21339 2017-07-31 Carl Love <cel@us.ibm.com>
21341 * config/rs6000/rs6000-c: Add support for built-in functions
21342 vector signed char vec_xl_be (signed long long, signed char *);
21343 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21344 vector signed int vec_xl_be (signed long long, signed int *);
21345 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21346 vector signed long long vec_xl_be (signed long long, signed long long *);
21347 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21348 vector signed short vec_xl_be (signed long long, signed short *);
21349 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21350 vector double vec_xl_be (signed long long, double *);
21351 vector float vec_xl_be (signed long long, float *);
21352 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21353 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21354 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21356 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21357 (altivec_expand_builtin): Add switch statement to call
21358 altivec_expand_xl_be for each builtin.
21359 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21360 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21361 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21362 __builtin_vsx_le_be_v16qi.
21363 * doc/extend.texi: Update the built-in documentation file for the
21364 new built-in functions.
21366 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21369 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21371 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21373 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21375 * config.gcc: Add z14.
21376 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21377 CPU model numbers for z13s and z14.
21378 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21380 * config/s390/s390-opts.h (enum processor_type): Rename
21381 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21382 * config/s390/s390.c (processor_table): Add field for CPU name to
21383 be passed to Binutils.
21384 (s390_asm_output_machine_for_arch): Use the new field in
21385 processor_table for Binutils.
21386 (s390_expand_builtin): Replace arch12 with z14.
21387 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21388 (s390_get_sched_attrmask): Likewise.
21389 (s390_get_unit_mask): Likewise.
21390 * config/s390/s390.opt: Add z14 to processor_type enum.
21392 2017-07-31 Martin Jambor <mjambor@suse.cz>
21395 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21396 regardless of optimization level.
21398 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21399 Martin Liska <mliska@suse.cz>
21401 * predict.def: Remove old comment and adjust probability.
21402 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21403 PREDICT statements.
21405 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21408 * config/i386/i386.c (ix86_function_naked): New function.
21409 (ix86_can_use_return_insn_p): Return false for naked functions.
21410 (ix86_expand_prologue): Skip prologue for naked functions.
21411 (ix86_expand_epilogue): Skip epilogue for naked functions
21412 and emit trap instruction.
21413 (ix86_warn_func_return): New function.
21414 (ix86_attribute_table): Add "naked" attribute specification.
21415 (TARGET_WARN_FUNC_RETURN): Define.
21416 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21418 2017-07-31 Martin Liska <mliska@suse.cz>
21420 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21421 (dump_gimple_bb_header): Always dump BB info.
21422 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21424 2017-07-31 Martin Liska <mliska@suse.cz>
21427 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21428 also with current_function_decl non-null equality.
21430 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21433 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21434 change type to the element type, instead add eltype variable and
21435 use it where we are interested in the element type.
21437 PR tree-optimization/81603
21438 * ipa-polymorphic-call.c
21439 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21440 offset arithmetic in offset_int, bail out if the resulting bit offset
21441 doesn't fit into shwi.
21443 2017-07-31 Martin Liska <mliska@suse.cz>
21445 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21446 (gimplify_save_expr): Fix comment.
21448 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21451 * config/i386/i386.c (ix86_function_arg): Update arguments for
21453 (ix86_compute_frame_layout): Set the initial stack offset to
21454 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21455 INCOMING_FRAME_SP_OFFSET.
21456 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21457 stack before exception handler returns.
21458 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21459 the 'ERROR_CODE' for exception handler.
21461 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21463 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21464 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21465 (ASM_OUTPUT_REG_POP): Ditto.
21466 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21467 instead of asm_fprintf to output pure string.
21469 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21471 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21472 to imported_module_or_decl hook.
21473 (debug_nothing_tree_tree_tree_bool): Remove.
21474 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21475 * debug.c (do_nothing_debug_hooks): Use
21476 debug_nothing_tree_tree_tree_bool_bool instead of
21477 debug_nothing_tree_tree_tree_bool.
21478 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21479 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21480 * sdbout.c (sdb_debug_hooks): Likewise.
21481 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21482 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21484 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21485 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21486 attribute, don't add anything.
21488 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21490 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21491 (fold_build2_stat_loc): Likewise.
21492 (fold_build3_stat_loc): Likewise.
21493 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21494 (fold_build1_loc): Remove macro.
21495 (fold_build2_loc): Likewise.
21496 (fold_build3_loc): Likewise.
21498 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21500 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21501 (gimple_build_debug_bind_source_stat): Likewise.
21502 * gimple.h (gimple_build_debug_bind): Remove macro.
21503 (gimple_build_debug_bind_source): Likewise.
21505 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21507 * bitmap.c (bitmap_alloc): Adjust.
21508 (bitmap_gc_alloc): Likewise.
21509 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21511 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21513 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21514 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21515 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21516 (bitmap_gc_alloc_stat): Likewise.
21517 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21519 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21521 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21522 * rtl.h (shallow_copy_rtx): Remove macro.
21524 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21526 * emit-rtl.c (gen_raw_REG): Adjust.
21527 * gengenrtl.c (gendef): Likewise.
21528 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21529 * rtl.h (rtx_alloc): Remove macro.
21531 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21533 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21534 (build_tree_list_stat): Likewise.
21535 * tree.h (build_tree_list): Remove macro.
21536 (build_tree_list_vec): Likewise.
21538 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21540 * tree.c (make_vector_stat): Remove _stat from name.
21541 (build_vector_stat): Likewise.
21542 * tree.h (make_vector_stat): Remove macro.
21543 (build_vector_stat): Likewise.
21545 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21547 * tree.h (build_var_debug_value): Remove prototype.
21549 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21551 * tree.c (tree_cons_stat): Remove _stat from name.
21552 * tree.h (tree_cons): Remove macro.
21554 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21556 * tree.c (build_vl_exp_stat): Remove _stat from name.
21557 * tree.h (build_vl_exp): Remove macro.
21559 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21561 * tree.c (build_decl_stat): Remove _stat from name.
21562 * tree.h (build_decl): Remove macro.
21564 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21566 * gimple.c (gimple_build_with_ops_stat): Adjust.
21567 (gimple_alloc_stat): Remove _stat from name.
21568 * gimple.h (gimple_alloc): Remove macro.
21570 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21572 * tree.c (make_tree_vec_stat): Remove _stat from name.
21573 (grow_tree_vec_stat): Likewise.
21574 * tree.h (make_tree_vec_stat): Adjust prototype.
21575 (grow_tree_vec_stat): Likewise.
21576 (make_tree_vec): Remove macro.
21577 (grow_tree_vec): Likewise.
21579 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21581 * fold-const.c (fold_build1_stat_loc): Adjust.
21582 (fold_build2_stat_loc): Likewise.
21583 (fold_build3_stat_loc): Likewise.
21584 * tree.c (build0_stat): Remove _stat from name.
21585 (build1_stat): Likewise.
21586 (build2_stat): Likewise.
21587 (build3_stat): Likewise.
21588 (build4_stat): Likewise.
21589 (build5_stat): Likewise.
21590 * tree.h (build1_loc): Remove macro, and rename _stat function
21592 (build2_loc): Likewise.
21593 (build3_loc): Likewise.
21594 (build4_loc): Likewise.
21595 (build5_loc): Likewise.
21597 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21599 * tree.c (make_int_cst_stat): Remove _stat from name.
21600 * tree.h (make_int_cst_stat): Adjust prototype.
21601 (make_int_cst): Remove macro.
21603 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21605 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21606 * tree.h (make_tree_binfo_stat): Adjust prototype.
21607 (make_tree_binfo): Remove.
21609 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21611 * tree.c (copy_node_stat): Rename to copy_node.
21612 (build_distinct_type_copy): Adjust.
21613 * tree.h (copy_node_stat): Adjust prototype.
21614 (copy_node): Remove macro.
21616 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21618 * tree.c (make_node_stat): rename to make_node.
21619 (build_tree_list_stat): Adjust.
21620 (build0_stat): Likewise.
21621 (build2_stat): Likewise.
21622 (build3_stat): Likewise.
21623 (build4_stat): Likewise.
21624 (build5_stat): Likewise.
21625 (build_decl_stat): Likewise.
21626 * tree.h (make_node_stat): Adjust prototype.
21627 (make_node): remove macro.
21629 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21631 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21632 (PPC_FEATURE2_SCV): Likewise.
21633 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21635 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21637 * config/aarch64/aarch64.c
21638 (aarch64_internal_mov_immediate): Add new special pattern.
21639 * config/aarch64/aarch64.md (*movdi_aarch64):
21640 Add reg/32bit const mov case.
21642 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21643 Richard Sandiford <richard.sandiford@linaro.org>
21645 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21646 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21647 Add integer and movi cases.
21648 (movi-split-hf-df-sf split, fp16): New.
21649 (enabled): Added TARGET_FP_F16INST.
21650 * config/aarch64/iterators.md (GPF_HF): New.
21651 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21653 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21655 * config/aarch64/aarch64.c
21656 (aarch64_simd_container_mode): Add prototype.
21657 (aarch64_expand_mov_immediate): Add HI support.
21658 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21659 (aarch64_can_const_movi_rtx_p): New.
21660 (aarch64_preferred_reload_class):
21661 Remove restrictions of using FP registers for certain SIMD operations.
21662 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21663 (aarch64_valid_floating_const): Add integer move validation.
21664 (aarch64_simd_imm_scalar_p): Remove.
21665 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21666 (aarch64_legitimate_constant_p): Expand list of supported cases.
21667 * config/aarch64/aarch64-protos.h
21668 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21669 (aarch64_reinterpret_float_as_int): New.
21670 (aarch64_simd_imm_scalar_p): Remove.
21671 * config/aarch64/constraints.md (Uvi): New.
21672 (Dd): Split into Ds and new Dd.
21673 * config/aarch64/aarch64.md (*movsi_aarch64):
21675 (*movdi_aarch64): Add SIMD mov case.
21677 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21679 * tree-predcom.c: (struct chain): Handle store-store chain in which
21680 stores for elimination only store loop invariant values.
21681 (execute_pred_commoning_chain): Ditto.
21682 (prepare_initializers_chain_store_elim): Ditto.
21683 (prepare_finalizers): Ditto.
21684 (is_inv_store_elimination_chain): New function.
21685 (initialize_root_vars_store_elim_1): New function.
21687 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21689 * tree-predcom.c: Revise general description of the pass.
21690 (enum chain_type): New enum type for store elimination.
21691 (struct chain): New field supporting store elimination.
21692 (struct component): Ditto.
21693 (dump_chain): Dump store-stores chain.
21694 (release_chain): Release resources.
21695 (split_data_refs_to_components): Compute and create component
21696 contains only stores for elimination.
21697 (get_chain_last_ref_at): New function.
21698 (make_invariant_chain): Initialization.
21699 (make_rooted_chain): Specify chain type in parameter and record it.
21700 (add_looparound_copies): Skip for store-stores chain.
21701 (determine_roots_comp): Compute type of chain and pass it to
21703 (initialize_root_vars_store_elim_2): New function.
21704 (finalize_eliminated_stores): New function.
21705 (remove_stmt): Handle store for elimination.
21706 (execute_pred_commoning_chain): Execute predictive commoning on
21707 store-store chains.
21708 (determine_unroll_factor): Skip unroll for store-stores chain.
21709 (prepare_initializers_chain_store_elim): New function.
21710 (prepare_initializers_chain): Hanlde store-store chain.
21711 (prepare_finalizers_chain, prepare_finalizers): New function.
21712 (tree_predictive_commoning_loop): Return integer value indicating
21713 if loop is unrolled or lcssa form is corrupted.
21714 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21716 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21718 * tree-predcom.c (initialize_root): Delete.
21719 (execute_pred_commoning_chain): Initialize root vars and replace
21720 reference of non-combined chain directly, rather than call above
21723 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21725 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21726 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21728 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21730 * tree-predcom.c (struct chain): New field init_seq.
21731 (release_chain): Release init_seq.
21732 (prepare_initializers_chain): Record intialization stmts in above
21734 (insert_init_seqs): New function.
21735 (tree_predictive_commoning_loop): Call insert_init_seqs.
21737 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21739 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21741 2017-07-28 Richard Biener <rguenther@suse.de>
21743 * match.pd: Remove superfluous :c.
21744 * genmatch.c (simplify::id): Add member.
21745 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21747 (current_id): New global.
21748 (dt_node::parent): Move from ...
21749 (dt_operand::parent): ... here. Add for_id member.
21750 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21751 (decision_tree::find_node): Relax order requirement when
21752 merging DT_TRUE nodes to ones inbetween the current simplify
21753 and the one we try to merge with. Add diagnostic whenever
21754 we need to enforce pattern order by not merging.
21755 (decision_tree::insert): Set current_id.
21756 (decision_tree::print_node): Dump parent node and for_id.
21757 (parser::last_id): Add member.
21758 (parser::push_simplify): Assign unique id.
21759 (parser::parser): Initialize last_id.
21761 2017-07-28 Martin Liska <mliska@suse.cz>
21764 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21765 gimple_build_debug_bind.
21767 2017-07-28 Richard Biener <rguenther@suse.de>
21769 PR tree-optimization/81502
21770 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21772 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21774 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21775 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21776 for BIT_FIELD_REF args.
21777 * fold-const.c (make_bit_field_ref): Likewise.
21778 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21780 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21783 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21784 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21785 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21786 Or it into SANITIZER_UNDEFINED.
21787 * ubsan.c: Include gimple-fold.h and varasm.h.
21788 (ubsan_expand_ptr_ifn): New function.
21789 (instrument_pointer_overflow): New function.
21790 (maybe_instrument_pointer_overflow): New function.
21791 (instrument_object_size): Formatting fix.
21792 (pass_ubsan::execute): Call instrument_pointer_overflow
21793 and maybe_instrument_pointer_overflow.
21794 * internal-fn.c (expand_UBSAN_PTR): New function.
21795 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21796 * sanitizer.def (__ubsan_handle_pointer_overflow,
21797 __ubsan_handle_pointer_overflow_abort): New builtins.
21798 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21799 * internal-fn.def (UBSAN_PTR): New internal function.
21800 * opts.c (sanitizer_opts): Add pointer-overflow.
21801 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21802 * fold-const.c (build_range_check): Compute pointer range check in
21803 integral type if pointer arithmetics would be needed. Formatting
21806 2017-07-28 Martin Liska <mliska@suse.cz>
21809 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21810 parameters that are of a variable-length.
21812 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21814 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21816 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21817 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21818 (CRT_CALL_STATIC_FUNCTION): Likewise.
21819 (ASM_DEFAULT_SPEC): New define.
21820 (ASM_SPEC32): Likewise.
21821 (ASM_SPEC64): Likewise.
21822 (ASM_SPEC_COMMON): Likewise.
21823 (ASM_SPEC): Likewise.
21824 (INVALID_64BIT): Likewise.
21825 (LINK_OS_DEFAULT_SPEC): Likewise.
21826 (LINK_OS_SPEC32): Likewise.
21827 (LINK_OS_SPEC64): Likewise.
21828 (POWERPC_LINUX): Likewise.
21829 (PTRDIFF_TYPE): Likewise.
21830 (RESTORE_FP_PREFIX): Likewise.
21831 (RESTORE_FP_SUFFIX): Likewise.
21832 (SAVE_FP_PREFIX): Likewise.
21833 (SAVE_FP_SUFFIX): Likewise.
21834 (SIZE_TYPE): Likewise.
21835 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21836 (TARGET_64BIT): Likewise.
21837 (TARGET_64BIT): Likewise.
21838 (TARGET_AIX): Likewise.
21839 (WCHAR_TYPE_SIZE): Likewise.
21840 (WCHAR_TYPE): Undefine.
21841 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21842 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21843 (CPP_OS_RTEMS_SPEC): Delete.
21844 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21845 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21847 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21849 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21851 PR tree-optimization/81578
21852 * tree-parloops.c (build_new_reduction): Bail out if
21853 reduction_code isn't one of the standard OpenMP reductions.
21854 Move the details printing after that decision.
21856 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21858 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
21859 related to reload_in_progress.
21860 (splat_input_operand): Likewise.
21861 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
21863 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
21865 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
21866 (TARGET_INSTANTIATE_DECLS): Likewise.
21867 (legitimate_indexed_address_p): Delete reload_in_progress code.
21868 (rs6000_debug_legitimate_address_p): Likewise.
21869 (rs6000_eliminate_indexed_memrefs): Likewise.
21870 (rs6000_emit_le_vsx_store): Likewise.
21871 (rs6000_emit_move_si_sf_subreg): Likewise.
21872 (rs6000_emit_move): Likewise.
21873 (register_to_reg_type): Likewise.
21874 (rs6000_pre_atomic_barrier): Likewise.
21875 (rs6000_machopic_legitimize_pic_address): Likewise.
21876 (rs6000_allocate_stack_temp): Likewise.
21877 (rs6000_address_for_fpconvert): Likewise.
21878 (rs6000_address_for_altivec): Likewise.
21879 (rs6000_secondary_memory_needed_rtx): Delete function.
21880 (rs6000_check_sdmode): Likewise.
21881 (rs6000_alloc_sdmode_stack_slot): Likewise.
21882 (rs6000_instantiate_decls): Likewise.
21883 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
21884 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
21885 Delete reload_in_progress.
21886 (*vec_reload_and_plus_<mptrsize>): Likewise.
21887 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
21888 (vsx_div_v2di): Likewise.
21889 (vsx_udiv_v2di): Likewise.
21891 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21893 * config/rs6000/rs6000.opt (mlra): Replace with stub.
21894 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
21895 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
21896 (rs6000_debug_reg_global): Delete print of LRA status.
21897 (rs6000_option_override_internal): Delete dead LRA related code.
21898 (rs6000_lra_p): Delete function.
21899 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
21901 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
21903 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
21904 * config/riscv/rtems.h: New file.
21906 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21907 Sudakshina Das <sudi.das@arm.com>
21909 * config/aarch64/aarch64.md
21910 (define_split for and<mode>3nr_compare): Move
21911 non aarch64_logical_operand to a register.
21912 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
21913 register immediate operand to a register.
21914 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
21916 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21918 PR middle-end/81564
21919 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
21921 2017-07-27 Richard Biener <rguenther@suse.de>
21923 PR tree-optimization/81573
21924 PR tree-optimization/81494
21925 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
21926 multi defuse cycle case.
21928 2017-07-27 Richard Biener <rguenther@suse.de>
21930 PR tree-optimization/81571
21931 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
21934 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
21936 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
21937 earlier and only if MASK_FPU is set. Adjust formatting.
21939 2017-07-27 Martin Liska <mliska@suse.cz>
21941 * opt-functions.awk: Add validation of value of Init.
21942 * optc-gen.awk: Pass new argument.
21944 2017-07-27 Martin Liska <mliska@suse.cz>
21946 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
21947 Fix wrong condition.
21949 2017-07-27 Martin Liska <mliska@suse.cz>
21951 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
21952 BBs and edges seen by autoFDO.
21954 2017-07-27 Richard Biener <rguenther@suse.de>
21956 PR tree-optimization/81502
21957 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
21958 with incompatible but same sized type.
21959 (execute_update_addresses_taken): Likewise.
21961 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
21963 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
21964 flag_tree_loop_vectorize rather than flag_tree_vectorize.
21966 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21969 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
21970 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
21971 Change s_operand to memory_operand.
21973 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
21975 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
21976 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
21977 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
21978 Emit instructions rather than returning an expression. Handle TFmode
21979 and KFmode by casting to TImode.
21980 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
21981 (rs6000_emit_le_vsx_store): Likewise.
21982 * config/rs6000/vsx.md (VSX_TI): New iterator.
21983 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
21984 (*vsx_le_undo_permute_<mode>): Likewise.
21985 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
21986 emit the split sequence.
21987 (*vsx_le_perm_store_<mode>): Likewise.
21989 2017-07-27 Jakub Jelinek <jakub@redhat.com>
21991 PR tree-optimization/81555
21992 PR tree-optimization/81556
21993 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
21994 if true, force CHANGED for the recursive invocation.
21995 (reassociate_bb): Remember original length of ops array, pass
21996 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
21998 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
21999 attributes for noipa attribute. For naked attribute use
22000 lookup_attribute first before lookup_attribute_spec.
22001 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22003 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22004 for functions with noipa attribute.
22005 (cgraph_externally_visible_p): Return true for functions with noipa
22007 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22008 for functions with noipa attribute.
22009 * doc/extend.texi: Document noipa function attribute.
22010 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22011 also for functions with noipa attribute.
22012 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22014 2017-07-26 Andrew Pinski <apinski@cavium.com>
22016 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22017 vec_unalign_load_cost and vec_unalign_store_cost.
22019 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22021 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22022 -mvsx-small-integer option.
22023 (ISA_3_0_MASKS_IEEE): Likewise.
22024 (OTHER_VSX_VECTOR_MASKS): Likewise.
22025 (POWERPC_MASKS): Likewise.
22026 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22027 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22028 code, only testing for DImode being allowed in non-VSX floating
22030 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22031 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22033 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22034 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22035 TARGET_P8_VECTOR test.
22036 (rs6000_secondary_reload_simple_move): Likewise.
22037 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22038 since TARGET_P9_VECTOR was already tested.
22039 (rs6000_opt_masks): Remove -mvsx-small-integer.
22040 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22041 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22043 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22044 test for TARGET_VEXTRACTUB was used, and that uses
22046 (p9 extract splitter): Likewise.
22047 (vsx_extract_<mode>_di_p9): Likewise.
22048 (vsx_extract_<mode>_store_p9): Likewise.
22049 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22050 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22051 the elimination of TARGET_VSX_SMALL_INTEGER.
22052 (vsx_extract_<mode>_p8): Likewise.
22053 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22054 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22055 (vsx_set_<mode>_p9): Likewise.
22056 (vsx_set_v4sf_p9): Likewise.
22057 (vsx_set_v4sf_p9_zero): Likewise.
22058 (vsx_insert_extract_v4sf_p9): Likewise.
22059 (vsx_insert_extract_v4sf_p9_2): Likewise.
22060 * config/rs6000/rs6000.md (sign extend splitter): Change
22061 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22062 (floatsi<mode>2_lfiwax_mem): Likewise.
22063 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22064 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22065 since a test for TARGET_P9_VECTOR was used.
22066 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22067 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22068 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22069 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22070 TARGET_P8_VECTOR test.
22071 (fix_trunc<mode>si2_stfiwx): Likewise.
22072 (fix_trunc<mode>si2_internal): Likewise.
22073 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22074 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22076 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22077 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22078 TARGET_P8_VECTOR test.
22079 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22080 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22081 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22083 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22084 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22085 since a test for TARGET_P9_VECTOR was used.
22086 (splitter for loading small constants): Likewise.
22088 2017-07-26 Andrew Pinski <apinski@cavium.com>
22090 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22093 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22096 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22097 (fp_valid_at): Likewise.
22099 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22101 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22102 (qdf24xx_addrcost_table): Likewise.
22103 (cortexa57_tunings): Update to use generic_branch_cost.
22104 (cortexa72_tunings): Likewise.
22105 (cortexa73_tunings): Likewise.
22106 (qdf24xx_tunings): Likewise.
22108 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22110 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22111 (thunderx2t99_branch_cost): Likewise.
22112 (cortexa35_tunings): Update to use generic_branch_cost.
22113 (cortexa53_tunings): Likewise.
22114 (cortexa57_tunings): Likewise.
22115 (cortexa72_tunings): Likewise.
22116 (cortexa73_tunings): Likewise.
22117 (thunderx2t99_tunings): Likewise.
22119 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22121 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22122 (sparc_option_override): Honour MASK_FSMULD.
22123 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22124 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22125 * config/sparc/sparc.opt (mfsmuld): New option.
22126 * doc/invoke.texi (mfsmuld): Document option.
22128 2017-07-26 Marek Polacek <polacek@redhat.com>
22130 PR middle-end/70992
22131 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22133 2017-07-26 Richard Biener <rguenther@suse.de>
22135 * gimple-match-head.c (do_valueize): Return OP if valueize
22137 (get_def): New helper to get at the def stmt of a SSA name
22138 if valueize allows.
22139 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22140 do_valueize to get at the def stmt.
22141 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22143 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22145 PR middle-end/46932
22146 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22148 2017-07-26 Martin Liska <mliska@suse.cz>
22151 * function.c (expand_function_start): Make expansion of
22152 nonlocal_goto_save_area after parm_birth_insn.
22154 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22156 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22157 from all CPU target flags enable members.
22159 2017-07-26 Richard Biener <rguenther@suse.de>
22161 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22162 (decision_tree::gen): Make 'type' const.
22163 (write_predicate): Likewise.
22165 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22167 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22168 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22169 (rs6000_option_override_internal): Likewise.
22170 (rs6000_expand_vector_set): Likewise.
22171 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22172 (TARGET_UPPER_REGS_SF): Likewise.
22173 (TARGET_UPPER_REGS_DI): Likewise.
22174 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22175 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22176 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22177 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22178 (Splitters for DI constants in Altivec registers): Likewise.
22179 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22180 (vsx_set_v4sf_p9): Likewise.
22181 (vsx_set_v4sf_p9_zero): Likewise.
22182 (vsx_insert_extract_v4sf_p9): Likewise.
22183 (vsx_insert_extract_v4sf_p9_2): Likewise.
22185 2017-07-25 Carl Love <cel@us.ibm.com>
22187 * doc/extend.texi: Update the built-in documentation file for the
22188 existing built-in functions
22189 vector signed char vec_cnttz (vector signed char);
22190 vector unsigned char vec_cnttz (vector unsigned char);
22191 vector signed short vec_cnttz (vector signed short);
22192 vector unsigned short vec_cnttz (vector unsigned short);
22193 vector signed int vec_cnttz (vector signed int);
22194 vector unsigned int vec_cnttz (vector unsigned int);
22195 vector signed long long vec_cnttz (vector signed long long);
22196 vector unsigned long long vec_cnttz (vector unsigned long long);
22198 2017-07-25 Andrew Pinski <apinski@cavium.com>
22200 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22201 accesses where the use is for the first operand of a BIT_INSERT.
22203 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22206 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22207 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22209 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22211 * config/i386/gstabs.h: Delete.
22212 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22214 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22216 * config/i386/i386.c (ix86_decompose_address): Do not check for
22217 register RTX when looking at index_reg or base_reg.
22218 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22220 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22222 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22223 to update EH info here.
22225 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22227 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22229 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22231 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22233 2017-07-25 Torsten Duwe <duwe@suse.de>
22235 * common.opt: Introduce -fpatchable-function-entry
22236 command line option, and its variables function_entry_patch_area_size
22237 and function_entry_patch_area_start.
22238 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22239 including a two-value parser.
22240 * target.def (print_patchable_function_entry): New target hook.
22241 * targhooks.h (default_print_patchable_function_entry): New function.
22242 * targhooks.c (default_print_patchable_function_entry): Likewise.
22243 * toplev.c (process_options): Switch off IPA-RA if
22244 patchable function entries are being generated.
22245 * varasm.c (assemble_start_function): Look at the
22246 patchable-function-entry command line switch and current
22247 function attributes and maybe generate NOP instructions by
22248 calling the print_patchable_function_entry hook.
22249 * doc/extend.texi: Document patchable_function_entry attribute.
22250 * doc/invoke.texi: Document -fpatchable_function_entry
22251 command line option.
22252 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22254 * doc/tm.texi: Re-generate.
22256 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22259 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22260 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22262 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22264 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22266 2017-07-25 Richard Biener <rguenther@suse.de>
22268 PR tree-optimization/81455
22269 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22270 not walk in cycles when looking for guards.
22272 2017-07-25 Richard Biener <rguenther@suse.de>
22274 PR tree-optimization/81529
22275 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22276 when optimizing backedge uses.
22278 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22280 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22282 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22283 to dl_section_ref. On AIX, append an expression to subtract
22284 the size of the section length to dl_section_ref.
22286 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22288 * configure.ac: If any of the config.* scripts fail, exit 1.
22289 * configure: Regenerate.
22291 2017-07-25 Richard Biener <rguenther@suse.de>
22293 PR middle-end/81546
22294 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22295 of immediate uses, be more verbose on errors.
22297 2017-07-25 Richard Biener <rguenther@suse.de>
22299 PR tree-optimization/81510
22300 * tree-vect-loop.c (vect_is_simple_reduction): When the
22301 reduction stmt is not inside the loop bail out.
22303 2017-07-25 Richard Biener <rguenther@suse.de>
22305 PR tree-optimization/81303
22306 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22307 profitability check against LOOP_VINFO_NITERSM1.
22309 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22311 * domwalk.c (cmp_bb_postorder): Simplify.
22312 (sort_bbs_postorder): New function. Use it...
22313 (dom_walker::walk): ...here to optimize common cases.
22315 2017-07-25 Martin Liska <mliska@suse.cz>
22318 * ipa-visibility.c (function_and_variable_visibility): Make the
22319 redirection just on target that supports aliasing.
22320 Fix GNU coding style.
22322 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22325 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22327 * config/arm/rtems.h: Likewise.
22328 * config/bfin/rtems.h: Likewise.
22329 * config/i386/rtemself.h: Likewise.
22330 * config/lm32/rtems.h: Likewise.
22331 * config/m32c/rtems.h: Likewise.
22332 * config/m68k/rtemself.h: Likewise.
22333 * config/microblaze/rtems.h: Likewise.
22334 * config/mips/rtems.h: Likewise.
22335 * config/moxie/rtems.h: Likewise.
22336 * config/nios2/rtems.h: Likewise.
22337 * config/powerpcspe/rtems.h: Likewise.
22338 * config/rs6000/rtems.h: Likewise.
22339 * config/rtems.h: Likewise.
22340 * config/sh/rtems.h: Likewise.
22341 * config/sh/rtemself.h: Likewise.
22342 * config/sparc/rtemself.h: Likewise.
22344 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22347 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22348 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22349 * tree-ssa-structalias.c (alias_get_name): Same.
22351 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22354 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22355 instructions if no du chain is found.
22357 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22359 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22361 2017-07-25 Richard Biener <rguenther@suse.de>
22363 PR middle-end/81505
22364 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22367 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22369 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22370 upper-regs options.
22371 (ISA_2_7_MASKS_SERVER): Likewise.
22372 (ISA_3_0_MASKS_IEEE): Likewise.
22373 (OTHER_P8_VECTOR_MASKS): Likewise.
22374 (OTHER_VSX_VECTOR_MASKS): Likewise.
22375 (POWERPC_MASKS): Likewise.
22376 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22377 duplicate list of options.
22378 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22379 explicit -mupper-regs options.
22380 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22381 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22382 alias for -mupper-regs-df.
22383 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22384 (rs6000_init_hard_regno_mode_ok): Likewise.
22385 (rs6000_option_override_internal): Likewise.
22386 (rs6000_opt_masks): Likewise.
22387 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22388 options in terms of whether -mvsx or -mpower8-vector was used.
22389 (TARGET_UPPER_REGS_DI): Likewise.
22390 (TARGET_UPPER_REGS_SF): Likewise.
22391 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22392 -mupper-regs-* options.
22394 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22396 * passes.c (emergency_dump_function): Print some empty lines and a
22397 header before the RTL dump.
22399 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22401 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22403 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22406 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22407 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22409 2017-07-24 Carl Love <cel@us.ibm.com>
22411 * config/rs6000/rs6000-c.c: Add support for built-in functions
22412 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22413 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22414 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22415 vec_extract_fp_from_shortl): Add defines for the two builtins.
22416 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22417 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22419 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22420 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22421 * doc/extend.texi: Update the built-in documentation file for the
22422 new built-in function.
22424 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22427 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22429 * doc/generic.texi: Likewise.
22430 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22431 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22433 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22435 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22436 (aarch64_mls_elt_merge<mode>): Likewise.
22438 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22440 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22441 having __cxa_atexit.
22443 2017-07-23 Michael Collison <michael.collison@arm.com>
22445 * config/arm/arm.c (arm_option_override): Deprecate
22446 use of -mstructure-size-boundary.
22447 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22448 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22450 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22453 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22454 Reduce cost estimate for direct moves.
22456 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22459 * config/i386/i386.c (ix86_option_override_internal): Disable
22460 BMI, BMI2 and TBM instructions for -m16.
22462 2017-07-21 Carl Love <cel@us.ibm.com>
22464 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22465 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22466 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22467 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22468 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22469 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22470 VMULOSW): New enum "unspec" values.
22471 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22472 altivec_vmulosw): New patterns.
22473 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22474 VMULOSW): Add definitions.
22476 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22478 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22479 (qdf24xx): Likewise.
22480 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22481 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22482 (AARCH64_FL_V8_1): Renumber.
22483 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22484 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22485 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22486 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22487 rdma to feature modifiers list.
22489 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22491 PR middle-end/56727
22492 * ipa-visibility (function_and_variable_visibility): Convert
22493 recursive PLT call to direct call if appropriate.
22495 2017-07-21 Andrew Pinski <apinski@cavium.com>
22497 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22498 operand 1 to see if the types precision matches.
22499 * fold-const.c (operand_equal_p): Likewise.
22501 2017-07-21 Richard Biener <rguenther@suse.de>
22503 PR tree-optimization/81303
22504 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22505 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22506 (vect_peeling_hash_get_lowest_cost): Adjust.
22507 (vect_enhance_data_refs_alignment): Likewise. Use
22508 vect_get_peeling_costs_all_drs to compute the penalty for no
22509 peeling to match up costs.
22511 2017-07-21 Richard Biener <rguenther@suse.de>
22513 PR tree-optimization/81500
22514 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22515 we didn't identify a reduction path.
22517 2017-07-21 Tom de Vries <tom@codesourcery.com>
22518 Cesar Philippidis <cesar@codesourcery.com>
22520 PR gcov-profile/81442
22521 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22524 2017-07-21 Tom de Vries <tom@codesourcery.com>
22527 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22529 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22530 nvptx_override_options_after_change.
22532 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22534 * dwarf2out.c (output_file_names): Avoid double testing for
22535 dwarf_version >= 5.
22537 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22539 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22541 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22543 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22545 (try_crossjump_to_edge): Do not punt on partitioned functions.
22547 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22549 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22550 Put all BBs reachable only via paths crossing cold region to cold
22552 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22554 2016-07-21 Richard Biener <rguenther@suse.de>
22556 PR tree-optimization/81303
22557 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22558 into account prologue and epilogue iterations when raising
22559 min_profitable_iters to sth at least covering one vector iteration.
22561 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22563 * config/arm/arm.c (arm_test_cpu_arch_dat):
22566 2017-07-20 Nathan Sidwell <nathan@acm.org>
22568 Remove TYPE_METHODS.
22569 * tree.h (TYPE_METHODS): Delete.
22570 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22571 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22572 (dbxout_type_methods): Scan TYPE_FIELDS.
22573 (dbxout_type): Don't check TYPE_METHODS here.
22574 * function.c (use_register_for_decl): Always ignore register for
22575 class types when not optimizing.
22576 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22577 * tree.c (free_lang_data_in_type): Stitch out member functions and
22578 templates from TYPE_FIELDS.
22579 (build_distinct_type_copy, verify_type_variant,
22580 verify_type): Member fns are on TYPE_FIELDS.
22581 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22582 * tree-pretty-print.c (dump_generic_node): Likewise.
22584 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22587 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22588 V2TImode and V4TImode.
22589 (ix86_expand_vector_extract): Likewise.
22590 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22591 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22592 (ssescalarmode): Handle V4TImode and V2TImode.
22593 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22594 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22595 (VEXTRACTI128_MODE): New mode iterator.
22596 (splitter for *vec_extractv?ti first element): New.
22597 (VEC_INIT_MODE): New mode iterator.
22598 (vec_init<mode>): Consolidate 3 expanders into one using
22599 VEC_INIT_MODE mode iterator.
22601 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22603 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22604 non_spilled_static_chain_regno_p.
22606 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22608 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22610 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22612 * bb-reorder.c (connect_traces): Allow copying of blocks within
22615 2017-07-20 Richard Biener <rguenther@suse.de>
22617 * gimple.h (gimple_phi_result): Add gphi * overload.
22618 (gimple_phi_result_ptr): Likewise.
22619 (gimple_phi_arg): Likewise. Adjust index assert to only
22620 allow actual argument accesses rather than all slots available
22622 (gimple_phi_arg_def): Add gphi * overload.
22623 * tree-phinodes.c (make_phi_node): Initialize only actual
22625 (resize_phi_node): Clear memory not covered by old node,
22626 do not initialize excess argument slots.
22627 (reserve_phi_args_for_new_edge): Initialize new argument slot
22630 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22632 PR tree-optimization/81388
22634 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22636 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22637 by removing computation of may_be_zero.
22639 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22640 Tom de Vries <tom@codesourcery.com>
22642 PR middle-end/81030
22643 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22644 when gimple level profile disagrees with what RTL expander did.
22646 2017-07-20 Richard Biener <rguenther@suse.de>
22648 PR tree-optimization/61171
22649 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22650 (vect_analyze_stmt): Add slp instance parameter.
22651 (vectorizable_reduction): Likewise.
22652 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22653 (vect_is_simple_reduction): Deal with chains not detected
22654 as SLP reduction chain, specifically not properly associated
22655 chains containing a mix of plus/minus.
22656 (get_reduction_op): Remove.
22657 (get_initial_defs_for_reduction): Simplify, pass in whether
22658 this is a reduction chain, pass in the SLP node for the PHIs.
22659 (vect_create_epilog_for_reduction): Get the SLP instance as
22661 (vectorizable_reduction): Get the SLP instance as arg.
22662 During analysis remember the SLP node with the PHIs in the
22663 instance. Simplify getting at the vectorized reduction PHIs.
22664 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22665 through SLP instance.
22666 (vect_slp_analyze_operations): Likewise.
22667 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22668 (vect_transform_stmt): Likewise.
22670 2017-07-20 Tom de Vries <tom@codesourcery.com>
22672 PR tree-optimization/81489
22673 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22674 read of phi arg location to before loop that modifies phi.
22676 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22678 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22681 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22683 PR middle-end/81331
22684 * except.c (execute): Fix ordering issue.
22686 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22688 PR rtl-optimization/81423
22689 * combine.c (make_compound_operation_int): Don't try to optimize
22690 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22692 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22694 PR rtl-optimization/81423
22695 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22696 with a constant that is -1 in the truncated to mode.
22698 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22700 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22701 (determine_unlikely_bbs): ... here.
22702 * predict.h (propagate_unlikely_bbs_forward): Declare.
22703 * cfgexpand.c (pass_expand::execute): Use it.
22704 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22706 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22707 propagate_unlikely_bbs_forward.
22709 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22711 PR middle-end/81331
22712 * except.c (maybe_add_nop_after_section_switch): New function.
22715 2017-07-19 Tom de Vries <tom@codesourcery.com>
22717 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22719 2017-07-19 Tom de Vries <tom@codesourcery.com>
22721 * gimple.h (gimple_phi_arg): Make assert more strict.
22723 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22725 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22726 * config/rs6000/mmintrin.h: New file.
22727 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22729 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22731 PR tree-optimization/81346
22732 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22734 2017-07-19 Tom de Vries <tom@codesourcery.com>
22736 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22738 2017-07-19 Tom de Vries <tom@codesourcery.com>
22740 * config/nvptx/nvptx-modes.def: Add V2DImode.
22741 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22742 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22743 (nvptx_output_mov_insn): Handle lack of mov.b128.
22744 (nvptx_print_operand): Handle 'H' and 'L' codes.
22745 (nvptx_vector_mode_supported): Allow V2DImode.
22746 (nvptx_preferred_simd_mode): New function.
22747 (nvptx_data_alignment): New function.
22748 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22749 nvptx_preferred_simd_mode.
22750 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22752 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22754 2017-07-19 Tom de Vries <tom@codesourcery.com>
22756 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22757 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22758 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22759 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22760 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22761 (mov<VECIM>_insn): New define_insn.
22762 (define_expand "mov<VECIM>): New define_expand.
22764 2017-07-19 Tom de Vries <tom@codesourcery.com>
22766 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22768 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22770 PR tree-optimization/81346
22771 * fold-const.h (fold_div_compare, range_check_type): Declare.
22772 * fold-const.c (range_check_type): New function.
22773 (build_range_check): Use range_check_type.
22774 (fold_div_compare): No longer static, rewritten into
22775 a match.pd helper function.
22776 (fold_comparison): Don't call fold_div_compare here.
22777 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22778 as helper function.
22780 2017-07-19 Nathan Sidwell <nathan@acm.org>
22782 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22783 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22784 * tree.c (find_decls_types_r, verify_type): Use
22785 TYPE_{MIN,MAX}_VALUE_RAW.
22786 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22787 (hash_tree): Likewise.
22788 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22790 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22793 2017-07-18 Tom de Vries <tom@codesourcery.com>
22795 PR middle-end/81464
22796 * omp-expand.c (expand_omp_for_static_chunk): Handle
22797 equal-argument loop exit phi.
22799 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22802 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22803 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22804 operand 2 predicate.
22805 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22806 operand 2 predicate.
22807 (ror,rol -> rorx splitters): Use const_int_operand as
22808 operand 2 predicate.
22810 2017-06-18 Richard Biener <rguenther@suse.de>
22812 PR tree-optimization/81410
22813 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22814 the gap in the ! slp_perm SLP case after each group.
22816 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22818 PR middle-end/81463
22819 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22822 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22824 PR middle-end/81462
22825 * predict.c (set_even_probabilities): Cleanup; do not affect
22826 probabilities that are already known.
22827 (combine_predictions_for_bb): Call even when count is set.
22829 2017-07-18 Nathan Sidwell <nathan@acm.org>
22831 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22834 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22837 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22838 optimization for loop niter analysis.
22840 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22843 * config/avr/avr.c (avr_optimize_casesi): Don't use
22844 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22846 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22848 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22849 body_cost_vec from _vect_peel_extended_info.
22850 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
22851 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
22854 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22856 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
22858 2017-07-18 Richard Biener <rguenther@suse.de>
22860 PR tree-optimization/80620
22861 PR tree-optimization/81403
22862 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
22863 info when re-using a VN table entry.
22865 2017-07-18 Richard Biener <rguenther@suse.de>
22867 PR tree-optimization/81418
22868 * tree-vect-loop.c (vectorizable_reduction): Properly compute
22869 vectype_in. Verify that with lane-reducing reduction operations
22870 we have a single def-use cycle.
22872 2017-07-17 Carl Love <cel@us.ibm.com>
22874 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
22876 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22877 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22878 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22879 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22880 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22881 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22882 VMULOSW): New enum "unspec" values.
22883 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22884 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22885 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22886 altivec_vmulosw): New patterns.
22887 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22888 VMULOSW): Add definitions.
22890 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
22892 * config/alpha/alpha.c: Include predict.h.
22894 2017-07-17 Yury Gribov <tetra2005@gmail.com>
22896 * tree-vrp.c (compare_assert_loc): Fix comparison function
22897 to return predictable results.
22899 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22901 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
22903 (subdi3): Likewise.
22904 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
22905 * doc/invoke.texi (mexpand-adddi): Update text.
22907 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22909 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
22910 that also clobbers the CC register. The old expand code is moved
22912 (*arc_clzsi2): ... here.
22913 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
22914 the CC register. The old expand code is moved to ...
22915 (arc_ctzsi2): ... here.
22917 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22919 * config/arc/arc.opt (mindexed-loads): Use initial value
22920 TARGET_INDEXED_LOADS_DEFAULT.
22921 (mauto-modify-reg): Use initial value
22922 TARGET_AUTO_MODIFY_REG_DEFAULT.
22923 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22924 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22925 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22926 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22928 2017-07-17 Martin Liska <mliska@suse.cz>
22931 * opts.c (finish_options): Do not allow -fgnu-tm
22932 w/ -fsanitize={kernel-,}address. Say sorry.
22934 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22937 * tree-loop-distribution.c (classify_partition): Only assert on
22938 numer of iterations.
22939 (merge_dep_scc_partitions): Delete prameter. Update function call.
22940 (distribute_loop): Remove code handling loop with unknown niters.
22941 (pass_loop_distribution::execute): Skip loop with unknown niters.
22943 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22946 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
22947 function sort_partitions_by_post_order.
22949 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22951 PR tree-optimization/81374
22952 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
22953 the max index of basic blocks, rather than number of basic blocks.
22955 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22957 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
22959 (arc_legitimate_pic_operand_p): Likewise.
22960 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
22962 (arc_needs_pcl_p): Likewise.
22963 (arc_legitimate_pc_offset_p): Likewise.
22964 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
22965 function is also used in constrains.md.
22966 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
22967 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
22968 PLUS. Only return true/false in known cases, otherwise assert.
22969 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
22970 is already called in arc_legitimate_constant_p.
22971 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
22973 (LEGITIMATE_PIC_OPERAND_P): Use
22974 arc_raw_symbolic_reference_mentioned_p function.
22975 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
22980 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22981 Andrew Burgess <andrew.burgess@embecosm.com>
22983 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
22984 (arc_return_address_register): New function.
22985 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
22986 (arc_handle_fndecl_attribute): Add naked attribute.
22987 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22988 (TARGET_WARN_FUNC_RETURN): Likewise.
22989 (arc_allocate_stack_slots_for_args): New function.
22990 (arc_warn_func_return): Likewise.
22991 (machine_function): Change type fn_type.
22992 (arc_compute_function_type): Consider new naked function type,
22993 change function return type.
22994 (arc_must_save_register): Adapt to handle new
22995 arc_compute_function_type's return type.
22996 (arc_expand_prologue): Likewise.
22997 (arc_expand_epilogue): Likewise.
22998 (arc_return_address_regs): Delete.
22999 (arc_return_address_register): New function.
23000 (arc_epilogue_uses): Use above function.
23001 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23002 (arc_function_type): Change encoding, add naked type.
23003 (ARC_INTERRUPT_P): Change to handle the new encoding.
23004 (ARC_FAST_INTERRUPT_P): Likewise.
23005 (ARC_NORMAL_P): Define.
23006 (ARC_NAKED_P): Likewise.
23007 (arc_compute_function_type): Delete prototype.
23008 * config/arc/arc.md (in_ret_delay_slot): Use
23009 arc_return_address_register function.
23010 (simple_return): Likewise.
23011 (p_return_i): Likewise.
23013 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23015 PR tree-optimization/81428
23016 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23017 can't be built for those types.
23019 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23021 Remove stuff dead since r239246.
23023 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23024 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23025 (avr_inform_devices): Remove dead stuff.
23027 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23029 * config/arm/arm_neon.h: Fix softp typo.
23031 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23033 PR tree-optimization/81365
23034 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23035 aggregate moves onto bb predecessor edges, make sure there are no
23036 loads that could alias the lhs in between the start of bb and the
23039 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23042 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23043 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23044 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23046 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23048 PR tree-optimization/81396
23049 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23050 (init_symbolic_number): Initialize it to 1.
23051 (perform_symbolic_merge): Add n_ops from both operands into the new
23053 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23054 without base_addr as useless if they need more than one operation.
23055 (bswap_replace): Handle !bswap case for NULL base_addr.
23057 2017-07-17 Tom de Vries <tom@codesourcery.com>
23060 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23063 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23065 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23066 conditional builtin define __FIX_LEON3FT_B2BST.
23068 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23070 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23071 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23074 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23076 PR rtl-optimization/81424
23077 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23078 to remove potential trapping from operands if -fnon-call-exceptions.
23080 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23082 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23083 profile_proability for scalling.
23084 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23086 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23088 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23090 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23092 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23093 fixpoint arithmetics.
23095 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23097 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23098 fixpoint arithmetics.
23100 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23102 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23103 fixpoint arithmetics.
23105 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23107 * profile-count.h (profile_probability::from_reg_br_prob_note,
23108 profile_probability::to_reg_br_prob_note): New functions.
23109 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23110 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23111 * predict.c (probability_reliable_p): Update.
23112 (edge_probability_reliable_p): Update.
23113 (br_prob_note_reliable_p): Update.
23114 (invert_br_probabilities): Update.
23115 (add_reg_br_prob_note): New function.
23116 (combine_predictions_for_insn): Update.
23117 * asan.c (asan_clear_shadow): Update.
23118 * cfgbuild.c (compute_outgoing_frequencies): Update.
23119 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23120 (update_br_prob_note): Update.
23121 (rtl_verify_edges): Update.
23122 (purge_dead_edges): Update.
23123 (fixup_reorder_chain): Update.
23124 * emit-rtl.c (try_split): Update.
23125 * ifcvt.c (cond_exec_process_insns): Update.
23126 (cond_exec_process_if_block): Update.
23127 (dead_or_predicable): Update.
23128 * internal-fn.c (expand_addsub_overflow): Update.
23129 (expand_neg_overflow): Update.
23130 (expand_mul_overflow): Update.
23131 * loop-doloop.c (doloop_modify): Update.
23132 * loop-unroll.c (compare_and_jump_seq): Update.
23133 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23134 * predict.h: Update.
23135 * reorg.c (mostly_true_jump): Update.
23137 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23138 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23139 * config/arc/arc.c: (emit_unlikely_jump): Update.
23140 * config/arm/arm.c: (emit_unlikely_jump): Update.
23141 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23142 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23143 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23144 (ix86_print_operand): Update.
23145 (ix86_split_fp_branch): Update.
23146 (predict_jump): Update.
23147 * config/ia64/ia64.c (ia64_print_operand): Update.
23148 * config/mmix/mmix.c (mmix_print_operand): Update.
23149 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23150 (rs6000_expand_split_stack_prologue): Update.
23151 * config/rs6000/rs6000.c: Update.
23152 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23153 (s390_expand_vec_movstr): Update.
23154 (s390_expand_cs_tdsi): Update.
23155 (s390_expand_split_stack_prologue): Update.
23156 * config/sh/sh.c (sh_print_operand): Update.
23157 (expand_cbranchsi4): Update.
23158 (expand_cbranchdi4): Update.
23159 * config/sparc/sparc.c (output_v9branch): Update.
23160 * config/spu/spu.c (get_branch_target): Update.
23161 (ea_load_store_inline): Update.
23162 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23163 * config/tilepro/tilepro.c: Update.
23165 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23167 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23168 (gimplify_save_expr): Likewise.
23170 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23172 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23174 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23176 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23178 * timevar.def (TV_IPA_FNSUMMARY): Define.
23180 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23182 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23183 to back store errata sensitive sequence from being generated.
23184 (sqrtdf2_fix): Likewise.
23186 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23188 * tree-ssa-threadupdate.c (compute_path_counts,
23189 update_joiner_offpath_counts): Use profile_probability.
23191 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23194 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23196 * config/arm/arm-c.c (arm_cpu_builtins): Define
23197 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23199 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23201 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23202 array entries to represent __ieee128 versions of the
23203 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23204 scalar_extract_sig, and scalar_insert_exp built-in functions.
23205 (altivec_resolve_overloaded_builtin): Add special case handling
23206 for the __builtin_scalar_insert_exp function, as represented by
23207 the P9V_BUILTIN_VEC_VSIEDP constant.
23208 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23209 exponent support for __ieee128 argument.
23210 (VSESQP): Add scalar extract signature support for __ieee128
23212 (VSTDCNQP): Add scalar test negative support for __ieee128
23214 (VSIEQP): Add scalar insert exponent support for __int128 argument
23215 with __ieee128 result.
23216 (VSIEQPF): Add scalar insert exponent support for __ieee128
23217 argument with __ieee128 result.
23218 (VSTDCQP): Add scalar test data class support for __ieee128
23220 (VSTDCNQP): Add overload support for scalar test negative with
23221 __ieee128 argument.
23222 (VSTDCQP): Add overload support for scalar test data class
23223 __ieee128 argument.
23224 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23225 UNSPEC_VSX_SXSIGDP.
23226 (UNSPEC_VSX_SIEXPQP): New constant.
23227 (xsxexpqp): New insn for VSX scalar extract exponent quad
23229 (xsxsigqp): New insn for VSX scalar extract significand quad
23231 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23232 precision with floating point argument.
23233 (xststdcqp): New expand for VSX scalar test data class quad
23235 (xststdcnegqp): New expand for VSX scalar test negative quad
23237 (xststdcqp): New insn to match expansions for VSX scalar test data
23238 class quad precision and VSX scalar test negative quad precision.
23239 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23240 special case operand checking to enforce that second operand of
23241 VSX scalar test data class with quad precision argument is a 7-bit
23243 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23244 prototypes and descriptions of __ieee128 versions of
23245 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23246 scalar_test_data_class, and scalar_test_neg built-in functions.
23248 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23250 PR tree-optimization/81162
23251 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23252 replace a negate with an add.
23254 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23256 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23258 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23260 * config/arm/arm-c.c (arm_cpu_builtins): Define
23261 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23263 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23265 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23266 (armv8-r): Set ARM Cortex-R52 as default CPU.
23267 * config/arm/arm-tables.opt: Regenerate.
23268 * config/arm/arm-tune.md: Regenerate.
23269 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23271 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23272 extension for -mcpu=cortex-r52.
23274 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23276 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23277 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23278 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23279 (fp-armv8): Define it as FP_ARMv8 only.
23280 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23281 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23283 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23284 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23285 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23286 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23287 than TARGET_FPU_ARMV8.
23288 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23289 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23290 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23292 * config/arm/neon.md (neon_vrint): Likewise.
23293 (neon_vcvt): Likewise.
23294 (neon_<fmaxmin_op><mode>): Likewise.
23295 (<fmaxmin><mode>3): Likewise.
23296 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23297 * config/arm/predicates.md (arm_cond_move_operator): Check against
23298 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23300 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23302 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23303 to top of function.
23305 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23307 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23308 loop in comment with memset.
23310 2017-07-14 Martin Liska <mliska@suse.cz>
23312 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23313 * dwarf2out.c (is_java): Remove the function.
23314 (output_pubname): Remove usage of the function.
23315 (lower_bound_default): Remove usage of DW_LANG_Java.
23316 (gen_compile_unit_die): Likewise.
23317 * gcc.c: Remove compiler defaults for .java and .zip files.
23318 * gimple-expr.c (remove_suffix): Change as there's no longer
23319 extension than 4-letter one.
23320 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23321 (gimplify_save_expr): Likewise.
23322 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23323 as it's possible even for other languages than Java.
23324 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23325 * lto-opts.c (lto_write_options): Remove reference to Java.
23326 * opts.c (strip_off_ending): Update file extension handling.
23327 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23328 * tree-eh.c (lower_resx): Likewise.
23329 * tree.c (free_lang_data_in_type): Remove dead code.
23330 (find_decls_types_r): Likewise.
23331 (build_common_builtin_nodes): Remove Java from a comment.
23332 (verify_type): Remove dead code.
23333 * varasm.c (assemble_external): Remove Java from a comment.
23335 2017-07-14 Martin Liska <mliska@suse.cz>
23337 * opts.c (finish_options): Add quotes.
23338 (common_handle_option): Likewise.
23340 2017-07-14 Martin Liska <mliska@suse.cz>
23342 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23343 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23344 Remove N_SO_PASCAL.
23345 * dwarf2out.c (lower_bound_default): Do not handle
23347 (gen_compile_unit_die): Likewise.
23348 * gcc.c: Remove default extension binding for GNU Pascal.
23349 * stmt.c: Remove Pascal language from a comment.
23350 * xcoffout.c: Likewise.
23352 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23355 * diagnostic-show-locus.c (fixit_cmp): New function.
23356 (layout::layout): Sort m_fixit_hints.
23357 (column_range::column_range): Assert that the values are valid.
23358 (struct char_span): New struct.
23359 (correction::overwrite): New method.
23360 (struct source_line): New struct.
23361 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23362 calls in terms of classes source_line and char_span, and
23363 correction::overwrite.
23364 (selftest::test_overlapped_fixit_printing_2): New function.
23365 (selftest::diagnostic_show_locus_c_tests): Call it.
23367 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23369 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23370 early if there is no lhs.
23372 2017-07-13 Martin Liska <mliska@suse.cz>
23374 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23375 (gen_reference_type_die): Likewise.
23376 * stor-layout.c: Remove Pascal-related comment.
23378 2017-07-13 Martin Liska <mliska@suse.cz>
23380 * opts.c (finish_options): Add quotes to error messages.
23381 (parse_sanitizer_options): Likewise.
23383 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23385 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23387 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23389 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23391 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23393 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23395 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23397 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23400 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23401 provides the hardware capability bits, define the macro
23402 __BUILTIN_CPU_SUPPORTS__.
23403 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23404 if GLIBC does not provide the hardware capability bits. Add a
23405 gcc_unreachable call if the built-in cpu function is neither
23406 __builtin_cpu_is nor __builtin_cpu_supports.
23407 (rs6000_get_function_versions_dispatcher): Change the warning
23408 that an old GLIBC is used which does not export the capability
23409 bits to be an error.
23410 * doc/extend.texi (target_clones attribute): Document the
23411 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23412 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23413 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23414 the macros defined by GCC if the newer GLIBC is available.
23416 2017-07-12 Jeff Law <law@redhat.com>
23418 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23419 remaining includes slightly.
23420 * config/riscv/riscv-builtins.c: Include profile-count.h.
23422 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23425 * config/avr/avr.c (avr_set_current_function): In diagnostic
23426 messages: Quote keywords and (parts of) identifiers.
23427 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23430 2017-07-12 Carl Love <cel@us.ibm.com>
23432 * config/rs6000/rs6000-c.c: Add support for built-in functions
23433 vector bool char vec_revb (vector bool char);
23434 vector bool short vec_revb (vector short char);
23435 vector bool int vec_revb (vector bool int);
23436 vector bool long long vec_revb (vector bool long long);
23437 * doc/extend.texi: Update the built-in documentation file for the
23438 new built-in functions.
23440 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23442 * config/s390/s390.md: Remove movcc splitter.
23444 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23446 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23447 load/store on condition.
23449 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23452 * config/avr/avr.c (avr_encode_section_info)
23453 [progmem && !TREE_READONLY]: Error if progmem object needs
23456 2017-07-11 Michael Collison <michael.collison@arm.com>
23458 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23461 2017-07-11 Carl Love <cel@us.ibm.com>
23463 * config/rs6000/rs6000-c.c: Add support for builtins
23464 vector unsigned int vec_parity_lsbb (vector signed int);
23465 vector unsigned int vec_parity_lsbb (vector unsigned int);
23466 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23467 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23468 vector unsigned long long vec_parity_lsbb (vector signed long long);
23469 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23470 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23471 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23472 * doc/extend.texi: Update the built-in documentation file for the
23473 new built-in functions.
23475 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23477 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23478 (layout::m_primary_loc): New field.
23479 (layout::layout): Initialize new field. Move location filtering
23480 logic from here to...
23481 (layout::maybe_add_location_range): ...this new method. Add
23482 support for filtering to just the lines already specified by other
23484 (layout::will_show_line_p): New method.
23485 (gcc_rich_location::add_location_if_nearby): New method.
23486 (selftest::test_add_location_if_nearby): New test function.
23487 (selftest::diagnostic_show_locus_c_tests): Call it.
23488 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23491 2017-07-11 Tom de Vries <tom@codesourcery.com>
23493 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23494 (bb_first_real_insn): New function.
23495 (nvptx_single): Add extra initialization of broadcasted condition
23498 2017-07-11 Nathan Sidwell <nathan@acm.org>
23500 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23502 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23504 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23505 Binutils doc as TEXI will mess them up.
23506 * doc/invoke.texi (AVR Options): Same here.
23508 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23510 * config/sparc/sparc.opt (mfix-ut700): New option.
23511 (mfix-gr712rc): Likewise.
23512 (sparc_fix_b2bst): New variable.
23513 * doc/invoke.texi (SPARC options): Document them.
23514 (ARM options): Fix warnings.
23515 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23516 instructions to prevent sequences that can trigger the store-store
23517 errata for certain LEON3FT processors.
23518 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23519 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23520 * config/sparc/sparc.md (fix_b2bst): New attribute.
23521 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23523 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23526 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23528 (*rsqrtsf2_sse): Ditto.
23530 (div<mode>3): Macroize insn from divdf3 and divsf3
23531 using MODEF mode iterator.
23533 2017-07-10 Martin Sebor <msebor@redhat.com>
23535 PR tree-optimization/80397
23536 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23537 instead of testing for equality to INTEGER_TYPE.
23539 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23541 * config.gcc: Remove uclibc from arc target spec.
23543 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23545 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23547 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23550 * lto-wrapper.c (remove_option): New function.
23551 (merge_and_complain): Merge PIC/PIE options more realistically.
23553 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23555 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23559 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23560 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23561 * config.in: Regenerate.
23562 * configure: Regenerate.
23563 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23564 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23565 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23566 (TARGET_GASISR_PROLOGUES): ...target mask.
23567 * common/config/avr/avr-common.c
23568 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23569 Set -mgas-isr-prologues.
23570 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23571 INSERT_PASS_BEFORE for it.
23572 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23573 * config/avr/avr.c (avr_option_override)
23574 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23575 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23576 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23577 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23578 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23580 (make_avr_pass_pre_proep): New function.
23581 (emit_push_sfr) <treg>: Add argument to function and use it
23582 instead of TMP_REG.
23583 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23584 and set machine->gasisr.yes.
23585 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23586 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23587 __gcc_isr.n_pushed to .L__stack_usage.
23588 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23589 (avr_asm_final_postscan_insn): ...this new static function.
23590 * config/avr/avr.h (machine_function)
23591 <is_no_gccisr, use_L__stack_usage>: New fields.
23592 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23593 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23594 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23595 (gasisr, *gasisr): New expander and insn.
23596 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23597 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23598 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23600 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23602 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23605 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23607 Move jump-tables out of .text again.
23610 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23611 (ASM_OUTPUT_ADDR_VEC): New function.
23612 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23613 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23614 INSN_ADDRESSes as asm comment.
23615 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23616 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23617 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23618 * config/avr/avr.md (*tablejump): Adjust comment.
23619 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23620 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23622 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23623 (avr_output_addr_vec): New proto.
23624 (avr_log_t) <insn_addresses>: New field.
23626 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23629 * config/i386/i386.c (ix86_function_arg_advance): Set
23630 outgoing_args_on_stack to true if there are outgoing arguments
23632 (ix86_function_arg): Likewise.
23633 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23634 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23635 * config/i386/i386.h (machine_function): Add
23636 outgoing_args_on_stack.
23638 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23640 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23641 supporting pthreds.
23642 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23644 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23646 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23647 (REAL_H): Remove $(MACHMODE_H).
23648 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23650 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23651 $(MACHMODE_H) and double-int.h.
23652 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23654 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23657 2017-07-07 Andrew Pinski <apinski@cavium.com>
23659 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23660 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23662 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23664 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23665 Add warning if GCC was not configured to link against a GLIBC that
23666 exports the hardware capability bits.
23667 (make_resolver_func): Make resolver function private and not a
23668 COMDAT function. Create the name with clone_function_name instead
23669 of make_unique_name.
23672 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23673 correct operand in doing the split.
23675 2017-07-07 Carl Love <cel@us.ibm.com>
23677 * config/rs6000/rs6000-c: Add support for built-in function
23678 vector unsigned short vec_pack_to_short_fp32 (vector float,
23680 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23681 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23682 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23683 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23684 (convert_4f32_8i16): Add define_expand.
23685 * doc/extend.texi: Update the built-in documentation file for the
23686 new built-in function.
23688 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23690 * config/sparc/m8.md: New file.
23691 * config/sparc/sparc.md: Include m8.md.
23693 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23695 * config/sparc/sparc.opt: New option -mvis4b.
23696 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23697 (sparc_option_override): Handle VIS4B.
23698 (enum sparc_builtins): Define
23699 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23700 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23701 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23702 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23703 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23704 (check_constant_argument): New function.
23705 (sparc_vis_init_builtins): Define builtins
23706 __builtin_vis_dictunpack{8,16,32},
23707 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23708 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23709 __builtin_vis_fpcmpde{8,16,32}shl and
23710 __builtin_vis_fpcmpur{8,16,32}shl.
23711 (sparc_expand_builtin): Check that the constant operands to
23712 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23713 constant and in range.
23714 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23716 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23717 (SPARC_IMM5_P): Likewise.
23718 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23719 (enabled): Handle vis4b.
23720 (UNSPEC_DICTUNPACK): New unspec.
23721 (UNSPEC_FPCMPSHL): Likewise.
23722 (UNSPEC_FPUCMPSHL): Likewise.
23723 (UNSPEC_FPCMPDESHL): Likewise.
23724 (UNSPEC_FPCMPURSHL): Likewise.
23725 (cpu_feature): New CPU feature `vis4b'.
23726 (dictunpack{8,16,32}): New insns.
23727 (FPCSMODE): New mode iterator.
23728 (fpcscond): New code iterator.
23729 (fpcsucond): Likewise.
23730 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23731 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23732 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23733 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23734 * config/sparc/constraints.md: Define constraints `q' for unsigned
23735 2-bit integer constants and `t' for unsigned 5-bit integer
23737 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23739 (imm5_operand_dictunpack16): Likewise.
23740 (imm5_operand_dictunpack32): Likewise.
23741 (imm2_operand): Likewise.
23742 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23743 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23744 ditunpack* and fpcmp*shl builtins.
23746 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23748 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23749 * config.in: Add HAVE_AS_SPARC6 define.
23750 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23752 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23754 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23755 (CPP_CPU_SPEC): Handle m8.
23756 (ASM_CPU_SPEC): Likewise.
23757 * config/sparc/sparc-opts.h (enum processor_type): Add
23759 * config/sparc/sparc.c (m8_costs): New struct.
23760 (sparc_option_override): Handle TARGET_CPU_m8.
23761 (sparc32_initialize_trampoline): Likewise.
23762 (sparc64_initialize_trampoline): Likewise.
23763 (sparc_issue_rate): Likewise.
23764 (sparc_register_move_cost): Likewise.
23765 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23766 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23767 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23768 (CPP_CPU_SPEC): Handle M8.
23769 (ASM_CPU_SPEC): Likewise.
23770 (AS_M8_FLAG): Define.
23771 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23772 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23773 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23775 * configure: Regenerate.
23776 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23779 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23781 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23783 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23784 ("*movdi_insn_sp32"): Do not set v3pipe.
23785 ("*movsi_insn"): Likewise.
23786 ("*movdi_insn_sp64"): Likewise.
23787 ("*movsf_insn"): Likewise.
23788 ("*movdf_insn_sp32"): Likewise.
23789 ("*movdf_insn_sp64"): Likewise.
23790 ("*zero_extendsidi2_insn_sp64"): Likewise.
23791 ("*sign_extendsidi2_insn"): Likewise.
23792 ("*mov<VM32:mode>_insn"): Likewise.
23793 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23794 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23795 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23796 ("<vlop:code><VL:mode>3"): Likewise.
23797 ("*not_<vlop:code><VL:mode>3"): Likewise.
23798 ("*nand<VL:mode>_vis"): Likewise.
23799 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23800 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23801 ("one_cmpl<VL:mode>2"): Likewise.
23802 ("faligndata<VM64:mode>_vis"): Likewise.
23803 ("alignaddrsi_vis"): Likewise.
23804 ("alignaddrdi_vis"): Likweise.
23805 ("alignaddrlsi_vis"): Likewise.
23806 ("alignaddrldi_vis"): Likewise.
23807 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23808 ("bmaskdi_vis"): Likewise.
23809 ("bmasksi_vis"): Likewise.
23810 ("bshuffle<VM64:mode>_vis"): Likewise.
23811 ("cmask8<P:mode>_vis"): Likewise.
23812 ("cmask16<P:mode>_vis"): Likewise.
23813 ("cmask32<P:mode>_vis"): Likewise.
23814 ("pdistn<P:mode>_vis"): Likewise.
23815 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23817 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23819 * config/sparc/sparc.md ("subtype"): New insn attribute.
23820 ("*wrgsr_sp64"): Set insn subtype.
23821 ("*rdgsr_sp64"): Likewise.
23822 ("alignaddrsi_vis"): Likewise.
23823 ("alignaddrdi_vis"): Likewise.
23824 ("alignaddrlsi_vis"): Likewise.
23825 ("alignaddrldi_vis"): Likewise.
23826 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23827 ("fexpand_vis"): Likewise.
23828 ("fpmerge_vis"): Likewise.
23829 ("faligndata<VM64:mode>_vis"): Likewise.
23830 ("bshuffle<VM64:mode>_vis"): Likewise.
23831 ("cmask8<P:mode>_vis"): Likewise.
23832 ("cmask16<P:mode>_vis"): Likewise.
23833 ("cmask32<P:mode>_vis"): Likewise.
23834 ("fchksm16_vis"): Likewise.
23835 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23836 ("fmean16_vis"): Likewise.
23837 ("fp<plusminus_insn>64_vis"): Likewise.
23838 ("<plusminus_insn>v8qi3"): Likewise.
23839 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23840 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23841 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23842 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23843 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23844 ("*movqi_insn"): Likewise.
23845 ("*movhi_insn"): Likewise.
23846 ("*movsi_insn"): Likewise.
23847 ("movsi_pic_gotdata_op"): Likewise.
23848 ("*movdi_insn_sp32"): Likewise.
23849 ("*movdi_insn_sp64"): Likewise.
23850 ("movdi_pic_gotdata_op"): Likewise.
23851 ("*movsf_insn"): Likewise.
23852 ("*movdf_insn_sp32"): Likewise.
23853 ("*movdf_insn_sp64"): Likewise.
23854 ("*zero_extendhisi2_insn"): Likewise.
23855 ("*zero_extendqihi2_insn"): Likewise.
23856 ("*zero_extendqisi2_insn"): Likewise.
23857 ("*zero_extendqidi2_insn"): Likewise.
23858 ("*zero_extendhidi2_insn"): Likewise.
23859 ("*zero_extendsidi2_insn_sp64"): Likewise.
23860 ("ldfsr"): Likewise.
23861 ("prefetch_64"): Likewise.
23862 ("prefetch_32"): Likewise.
23863 ("tie_ld32"): Likewise.
23864 ("tie_ld64"): Likewise.
23865 ("*tldo_ldub_sp32"): Likewise.
23866 ("*tldo_ldub1_sp32"): Likewise.
23867 ("*tldo_ldub2_sp32"): Likewise.
23868 ("*tldo_ldub_sp64"): Likewise.
23869 ("*tldo_ldub1_sp64"): Likewise.
23870 ("*tldo_ldub2_sp64"): Likewise.
23871 ("*tldo_ldub3_sp64"): Likewise.
23872 ("*tldo_lduh_sp32"): Likewise.
23873 ("*tldo_lduh1_sp32"): Likewise.
23874 ("*tldo_lduh_sp64"): Likewise.
23875 ("*tldo_lduh1_sp64"): Likewise.
23876 ("*tldo_lduh2_sp64"): Likewise.
23877 ("*tldo_lduw_sp32"): Likewise.
23878 ("*tldo_lduw_sp64"): Likewise.
23879 ("*tldo_lduw1_sp64"): Likewise.
23880 ("*tldo_ldx_sp64"): Likewise.
23881 ("*mov<VM32:mode>_insn"): Likewise.
23882 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23883 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23885 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23887 * config/sparc/sparc.md ("type"): New insn type viscmp.
23888 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
23890 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
23891 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
23892 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23893 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
23895 ("n7_vis_logical_11cycle"): Likewise.
23896 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
23897 * config/sparc/niagara2.md ("niag3_vis": Likewise.
23898 * config/sparc/niagara.md ("niag_vis"): Likewise.
23899 * config/sparc/ultra3.md ("us3_fga"): Likewise.
23900 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
23902 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23904 * config/sparc/sparc.md: New instruction type `bmask'.
23905 (bmaskdi_vis): Use the `bmask' type.
23906 (bmasksi_vis): Likewise.
23907 * config/sparc/ultra3.md (us3_array): Likewise.
23908 * config/sparc/niagara7.md (n7_array): Likewise.
23909 * config/sparc/niagara4.md (n4_array): Likewise.
23910 * config/sparc/niagara2.md (niag2_vis): Likewise.
23911 (niag3_vis): Likewise.
23912 * config/sparc/niagara.md (niag_vis): Likewise.
23914 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23916 * ipa-comdats.c: Remove optimize check from gate.
23917 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
23918 for functions not optimized.
23919 (ipa_fn_summary_read): Skip optimize check.
23920 (ipa_fn_summary_write): Likewise.
23921 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
23923 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
23925 (can_inline_edge_p): Check flag_pcc_struct_return for match.
23926 (check_callers): Give up on caller which is not optimized.
23927 (inline_small_functions): Likewise.
23928 (ipa_inline): Do not give up when not optimizing.
23929 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
23930 away unoptimizes cdtors.
23931 (whole_program_function_and_variable_visibility): Do
23932 ipa_discover_readonly_nonaddressable_vars in LTO mode.
23933 * ipa.c (process_references): Do not check optimize.
23934 (symbol_table::remove_unreachable_nodes): Update optimize check.
23935 (set_writeonly_bit): Update optimize check.
23936 (pass_ipa_cdtor_merge::gate): Do not check optimize.
23937 (pass_ipa_single_use::gate): Remove.
23939 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23941 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
23942 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
23943 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
23944 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
23945 permute_load, permute_store, adjust_extract, adjust_splat,
23946 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
23947 replace_swap_with_copy, dump_swap_insn_table,
23948 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
23949 recombine_lvx_pattern, recombine_stvx_pattern,
23950 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
23951 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
23952 to file rs6000-p8swap.c.
23953 * config/rs6000/rs6000-p8swap.c: New file.
23954 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
23955 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
23956 and rs6000*-*-* targets.
23958 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23960 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
23962 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23964 * lto-wrapper.c (merge_and_complain): Do not merge
23965 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
23966 fsigned_zeros, ftrapping_math, fwrapv.
23967 (append_compiler_options): Do not track these options.
23968 (append_linker_options): Likewie
23970 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23972 * cgraphunit.c (cgraph_node::finalize_function): When
23973 !flag_toplevel_reorde set no_reorder flag.
23974 (varpool_node::finalize_decl): Likewise.
23975 (symbol_table::compile): Drop no toplevel reorder path.
23977 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23979 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
23980 edges; zero probability is not better than uninitialized.
23982 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23984 * asan.h (asan_sanitize_allocas_p): Declare.
23985 * asan.c (asan_sanitize_allocas_p): New function.
23986 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
23987 (handle_builtin_alloca): Likewise.
23988 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
23989 if !asan_sanitize_allocas_p.
23990 * params.def (asan-instrument-allocas): Add new option.
23991 * params.h (ASAN_PROTECT_ALLOCAS): Define.
23992 * opts.c (common_handle_option): Disable allocas sanitization for
23995 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23997 * asan.c: Include gimple-fold.h.
23998 (get_last_alloca_addr): New function.
23999 (handle_builtin_stackrestore): Likewise.
24000 (handle_builtin_alloca): Likewise.
24001 (asan_emit_allocas_unpoison): Likewise.
24002 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24003 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24004 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24005 (instrument_builtin_call): Pass gimple iterator to
24006 get_mem_refs_of_builtin_call.
24007 (last_alloca_addr): New global.
24008 * asan.h (asan_emit_allocas_unpoison): Declare.
24009 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24010 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24011 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24012 if function calls alloca.
24013 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24014 * gimple-fold.h (replace_call_with_value): Declare.
24015 * internal-fn.c: Include asan.h.
24016 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24017 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24019 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24021 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24022 (C_SELFTEST_FLAGS): New.
24023 (CPP_SELFTEST_FLAGS): New.
24024 (SELFTEST_DEPS): New, from deps of s-selftest.
24025 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24026 (CPP_SELFTEST_DEPS): New.
24027 (selftest): Add dependency on s-selftest-c++.
24028 (s-selftest): Rename to...
24029 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24030 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24031 than SELFTEST_FLAGS.
24032 (selftest-gdb): Rename to...
24033 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24035 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24036 (selftest-valgrind): Rename to...
24037 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24039 (selftest-valgrind): Reintroduce as an alias for
24040 selftest-c-valgrind.
24041 (s-selftest-c++): New.
24042 (selftest-c++-gdb): New.
24043 (selftest-c++-valgrind): New.
24045 2017-07-06 Olivier Hainque <hainque@adacore.com>
24047 * gcc.c (process_command): When deciding if undefined variables
24048 should be ignored when processing specs, accept "gcc -v" as well.
24050 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24052 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24053 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24055 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24057 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24058 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24059 * config/arm/arm-tables.opt: Regenerate.
24060 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24062 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24064 2017-07-06 Carl Love <cel@us.ibm.com>
24066 * ChangeLog: Clean up from mid air collision
24068 2017-07-06 Carl Love <cel@us.ibm.com>
24070 * config/rs6000/rs6000-c.c: Add support for built-in functions
24071 vector signed int vec_subc (vector signed int, vector signed int);
24072 vector signed __int128 vec_subc (vector signed __int128,
24073 vector signed __int128);
24074 vector unsigned __int128 vec_subc (vector unsigned __int128,
24075 vector unsigned __int128);
24076 vector signed int vec_sube (vector signed int, vector signed int,
24077 vector signed int);
24078 vector unsigned int vec_sube (vector unsigned int,
24079 vector unsigned int,
24080 vector unsigned int);
24081 vector signed __int128 vec_sube (vector signed __int128,
24082 vector signed __int128,
24083 vector signed__int128);
24084 vector unsigned __int128 vec_sube (vector unsigned __int128,
24085 vector unsigned __int128,
24086 vector unsigned __int128);
24087 vector signed int vec_subec (vector signed int, vector signed int,
24088 vector signed int);
24089 vector unsigned int vec_subec (vector unsigned int,
24090 vector unsigned int,
24091 vector unsigned int);
24092 vector signed __int128 vec_subec (vector signed __int128,
24093 vector signed __int128,
24094 vector signed__int128);
24095 vector unsigned __int128 vec_subec (vector unsigned __int128,
24096 vector unsigned __int128,
24097 vector unsigned __int128);
24098 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24099 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24100 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24101 BU_ALTIVEC_OVERLOAD_X definitions.
24102 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24103 * doc/extend.texi: Update the built-in documentation file for the new
24104 built-in functions.
24106 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24109 * diagnostic-show-locus.c (layout::layout): Use start and finish
24110 spelling location for the start and finish of each range.
24111 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24112 Add unused aspect param.
24113 * input.c (expand_location_1): Add "aspect" param, and use it
24114 to access the correct part of the location.
24115 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24117 (expand_location_to_spelling_point): Likewise.
24118 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24119 param, and pass it to expand_location_1.
24121 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24123 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24124 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24125 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24126 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24127 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24128 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24129 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24130 _mm_maskz_getmant_ss): New intrinsics.
24131 (__builtin_ia32_getexpss128_mask): Changed to ...
24132 __builtin_ia32_getexpss128_round ... this.
24133 (__builtin_ia32_getexpsd128_mask): Changed to ...
24134 __builtin_ia32_getexpsd128_round ... this.
24135 * config/i386/i386-builtin-types.def
24136 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24137 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24138 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24139 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24140 __builtin_ia32_getmantss_mask_round): New builtins.
24141 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24142 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24143 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24144 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24145 * config/i386/sse.md
24146 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24147 avx512f_sgetexp<mode><mask_scalar_name>
24148 <round_saeonly_scalar_name> ... this.
24149 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24150 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24151 vgetexp<ssescalarmodesuffix>
24152 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24153 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24154 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24155 avx512f_vgetmant<mode><mask_scalar_name>
24156 <round_saeonly_scalar_name> ... this.
24157 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24158 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24159 vgetmant<ssescalarmodesuffix>
24160 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24161 %0<mask_scalar_operand4>, %1, %2
24162 <round_saeonly_scalar_mask_op4>, %3} ... this.
24163 * config/i386/subst.md (mask_scalar_operand4,
24164 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24165 round_saeonly_scalar_nimm_predicate): New subst attributes.
24167 2017-07-06 Julia Koval <julia.koval@intel.com>
24169 * config/i386/i386.c (ix86_erase_embedded_rounding):
24170 Remove code for old rounding pattern.
24172 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24174 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24176 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24178 * doc/sourcebuild.texi (Test Directives, Variants of
24179 dg-require-support): Add documentation for dg-require-stack-check.
24181 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24183 * config/i386/subst.md (mask_scalar, round_scalar,
24184 round_saeonly_scalar): New meta-templates.
24185 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24186 round_scalar_mask_operand3, round_scalar_mask_op3,
24187 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24188 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24189 round_saeonly_scalar_constraint,
24190 round_saeonly_scalar_prefix): New subst attribute.
24191 * config/i386/sse.md
24192 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24193 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24194 <round_scalar_name> ... this.
24195 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24196 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24197 <round_scalar_name> ... this.
24198 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24199 <sse>_vm<code><mode>3<mask_scalar_name>
24200 <round_saeonly_scalar_name> ... this.
24201 (v<plusminus_mnemonic><ssescalarmodesuffix>
24202 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24203 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24204 v<plusminus_mnemonic><ssescalarmodesuffix>
24205 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24206 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24207 (v<multdiv_mnemonic><ssescalarmodesuffix>
24208 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24209 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24210 v<multdiv_mnemonic><ssescalarmodesuffix>
24211 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24212 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24213 (v<maxmin_float><ssescalarmodesuffix>
24214 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24215 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24216 v<maxmin_float><ssescalarmodesuffix>
24217 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24218 %0<mask_scalar_operand3>, %1, %<iptr>2
24219 <round_saeonly_scalar_mask_op3>} ... this.
24221 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24223 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24224 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24226 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24227 Alan Hayward <alan.hayward@arm.com>
24228 David Sherwood <david.sherwood@arm.com>
24230 * combine.c (simplify_if_then_else): Remove "enum" before
24232 * compare-elim.c (can_eliminate_compare): Likewise.
24233 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24235 (aarch64_lookup_simd_builtin_type): Likewise.
24236 (aarch64_simd_builtin_type): Likewise.
24237 (aarch64_init_simd_builtin_types): Likewise.
24238 (aarch64_simd_expand_args): Likewise.
24239 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24241 (aarch64_reverse_mask): Likewise.
24242 (aarch64_simd_emit_reg_reg_move): Likewise.
24243 (aarch64_gen_adjusted_ldpstp): Likewise.
24244 (aarch64_ccmp_mode_to_code): Likewise.
24245 (aarch64_operands_ok_for_ldpstp): Likewise.
24246 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24247 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24249 (aarch64_min_divisions_for_recip_mul): Likewise.
24250 (aarch64_reassociation_width): Likewise.
24251 (aarch64_get_condition_code_1): Likewise.
24252 (aarch64_simd_emit_reg_reg_move): Likewise.
24253 (aarch64_simd_attr_length_rglist): Likewise.
24254 (aarch64_reverse_mask): Likewise.
24255 (aarch64_operands_ok_for_ldpstp): Likewise.
24256 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24257 (aarch64_gen_adjusted_ldpstp): Likewise.
24258 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24260 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24261 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24262 (arm_lookup_simd_builtin_type): Likewise.
24263 (arm_simd_builtin_type): Likewise.
24264 (arm_init_simd_builtin_types): Likewise.
24265 (arm_expand_builtin_args): Likewise.
24266 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24267 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24268 (ft32_setup_incoming_varargs): Likewise.
24269 (ft32_function_arg): Likewise.
24270 (ft32_function_arg_advance): Likewise.
24271 (ft32_pass_by_reference): Likewise.
24272 (ft32_arg_partial_bytes): Likewise.
24273 (ft32_valid_pointer_mode): Likewise.
24274 (ft32_addr_space_pointer_mode): Likewise.
24275 (ft32_addr_space_legitimate_address_p): Likewise.
24276 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24278 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24279 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24280 (iamcu_alignment): Likewise.
24281 (canonicalize_vector_int_perm): Likewise.
24282 (ix86_noce_conversion_profitable_p): Likewise.
24283 (ix86_mpx_bound_mode): Likewise.
24284 (ix86_operands_ok_for_move_multiple): Likewise.
24285 * config/microblaze/microblaze-protos.h
24286 (microblaze_expand_conditional_branch_reg): Likewise.
24287 * config/microblaze/microblaze.c
24288 (microblaze_expand_conditional_branch_reg): Likewise.
24289 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24291 (rs6000_reassociation_width): Likewise.
24292 (rs6000_invalid_binary_op): Likewise.
24293 (fusion_p9_p): Likewise.
24294 (emit_fusion_p9_load): Likewise.
24295 (emit_fusion_p9_store): Likewise.
24296 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24298 (riscv_hard_regno_mode_ok_p): Likewise.
24299 (riscv_address_insns): Likewise.
24300 (riscv_split_symbol): Likewise.
24301 (riscv_legitimize_move): Likewise.
24302 (riscv_function_value): Likewise.
24303 (riscv_hard_regno_nregs): Likewise.
24304 (riscv_expand_builtin): Likewise.
24305 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24306 (riscv_build_integer): Likewise.
24307 (riscv_split_integer): Likewise.
24308 (riscv_legitimate_constant_p): Likewise.
24309 (riscv_cannot_force_const_mem): Likewise.
24310 (riscv_regno_mode_ok_for_base_p): Likewise.
24311 (riscv_valid_base_register_p): Likewise.
24312 (riscv_valid_offset_p): Likewise.
24313 (riscv_valid_lo_sum_p): Likewise.
24314 (riscv_classify_address): Likewise.
24315 (riscv_legitimate_address_p): Likewise.
24316 (riscv_address_insns): Likewise.
24317 (riscv_load_store_insns): Likewise.
24318 (riscv_force_binary): Likewise.
24319 (riscv_split_symbol): Likewise.
24320 (riscv_force_address): Likewise.
24321 (riscv_legitimize_address): Likewise.
24322 (riscv_move_integer): Likewise.
24323 (riscv_legitimize_const_move): Likewise.
24324 (riscv_legitimize_move): Likewise.
24325 (riscv_address_cost): Likewise.
24326 (riscv_subword): Likewise.
24327 (riscv_output_move): Likewise.
24328 (riscv_canonicalize_int_order_test): Likewise.
24329 (riscv_emit_int_order_test): Likewise.
24330 (riscv_function_arg_boundary): Likewise.
24331 (riscv_pass_mode_in_fpr_p): Likewise.
24332 (riscv_pass_fpr_single): Likewise.
24333 (riscv_pass_fpr_pair): Likewise.
24334 (riscv_get_arg_info): Likewise.
24335 (riscv_function_arg): Likewise.
24336 (riscv_function_arg_advance): Likewise.
24337 (riscv_arg_partial_bytes): Likewise.
24338 (riscv_function_value): Likewise.
24339 (riscv_pass_by_reference): Likewise.
24340 (riscv_setup_incoming_varargs): Likewise.
24341 (riscv_print_operand): Likewise.
24342 (riscv_elf_select_rtx_section): Likewise.
24343 (riscv_save_restore_reg): Likewise.
24344 (riscv_for_each_saved_reg): Likewise.
24345 (riscv_register_move_cost): Likewise.
24346 (riscv_hard_regno_mode_ok_p): Likewise.
24347 (riscv_hard_regno_nregs): Likewise.
24348 (riscv_class_max_nregs): Likewise.
24349 (riscv_memory_move_cost): Likewise.
24350 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24351 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24352 (rl78_addr_space_address_mode): Likewise.
24353 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24355 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24356 (rs6000_reassociation_width): Likewise.
24357 (rs6000_invalid_binary_op): Likewise.
24358 (fusion_p9_p): Likewise.
24359 (emit_fusion_p9_load): Likewise.
24360 (emit_fusion_p9_store): Likewise.
24361 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24362 (ok_for_simple_move_operands): Likewise.
24363 (ok_for_simple_move_strict_operands): Likewise.
24364 (ok_for_simple_arith_logic_operands): Likewise.
24365 (visium_legitimize_reload_address): Likewise.
24366 (visium_select_cc_mode): Likewise.
24367 (output_cbranch): Likewise.
24368 (visium_split_double_move): Likewise.
24369 (visium_expand_copysign): Likewise.
24370 (visium_expand_int_cstore): Likewise.
24371 (visium_expand_fp_cstore): Likewise.
24372 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24373 (visium_function_arg): Likewise.
24374 (visium_function_arg_advance): Likewise.
24375 (visium_libcall_value): Likewise.
24376 (visium_setup_incoming_varargs): Likewise.
24377 (visium_legitimate_constant_p): Likewise.
24378 (visium_legitimate_address_p): Likewise.
24379 (visium_legitimize_address): Likewise.
24380 (visium_secondary_reload): Likewise.
24381 (visium_register_move_cost): Likewise.
24382 (visium_memory_move_cost): Likewise.
24383 (prepare_move_operands): Likewise.
24384 (ok_for_simple_move_operands): Likewise.
24385 (ok_for_simple_move_strict_operands): Likewise.
24386 (ok_for_simple_arith_logic_operands): Likewise.
24387 (visium_function_value_1): Likewise.
24388 (rtx_ok_for_offset_p): Likewise.
24389 (visium_legitimize_reload_address): Likewise.
24390 (visium_split_double_move): Likewise.
24391 (visium_expand_copysign): Likewise.
24392 (visium_expand_int_cstore): Likewise.
24393 (visium_expand_fp_cstore): Likewise.
24394 (visium_split_cstore): Likewise.
24395 (visium_select_cc_mode): Likewise.
24396 (visium_split_cbranch): Likewise.
24397 (output_cbranch): Likewise.
24398 (visium_print_operand_address): Likewise.
24399 * expmed.c (flip_storage_order): Likewise.
24400 * expmed.h (emit_cstore): Likewise.
24401 (flip_storage_order): Likewise.
24402 * genrecog.c (validate_pattern): Likewise.
24403 * hsa-gen.c (gen_hsa_addr): Likewise.
24404 * internal-fn.c (expand_arith_overflow): Likewise.
24405 * ira-color.c (allocno_copy_cost_saving): Likewise.
24406 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24407 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24408 (process_invariant_for_inheritance): Likewise.
24409 * lra-eliminations.c (move_plus_up): Likewise.
24410 * omp-low.c (lower_oacc_reductions): Likewise.
24411 * simplify-rtx.c (simplify_subreg): Likewise.
24412 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24413 (TARGET_CHKP_BOUND_MODE): Likewise..
24414 * targhooks.c (default_chkp_bound_mode): Likewise.
24415 (default_setup_incoming_vararg_bounds): Likewise.
24416 * targhooks.h (default_chkp_bound_mode): Likewise.
24417 (default_setup_incoming_vararg_bounds): Likewise.
24418 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24419 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24420 (have_whole_vector_shift): Likewise.
24421 * tree-vect-stmts.c (vectorizable_load): Likewise.
24422 * doc/tm.texi: Regenerate.
24424 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24426 Graceful degrade if Binutils PR21472 is not available.
24429 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24430 .rodata in flash test fails.
24431 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24432 * confgure: Regenerate.
24433 * config.in: Regenerate.
24434 * config/avr/avr.c (avr_asm_named_section)
24435 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24436 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24437 (avr_asm_init_sections): Same.
24439 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24441 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24442 (fma<VH:mode>4_intrinsic): Likewise.
24443 (*fmsub<VCVTF:mode>4): Likewise.
24444 (*fmsub<VH:mode>4_intrinsic): Likewise.
24446 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24449 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24450 Don't depend on "optimize > 0".
24451 (out_movhi_r_mr, out_movqi_mr_r): Same.
24452 (out_movhi_mr_r, out_movqi_r_mr): Same.
24453 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24454 io_address_operand on "optimize > 0".
24456 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24458 * tree-loop-distribution.c: Add general explanantion on the pass.
24459 (generate_loops_for_partition): Mark distributed loop.
24460 (pg_add_dependence_edges): New parameter. Handle alias data
24461 dependence specially and record it in the parameter if asked.
24462 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24463 (init_partition_graph_vertices, add_partition_graph_edge): New.
24464 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24465 (free_partition_graph_vdata, build_partition_graph): New.
24466 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24467 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24468 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24469 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24470 (version_for_distribution_p, finalize_partitions): New.
24471 (distribute_loop): Handle alias data dependence specially. Factor
24472 out loop fusion code as functions and call these functions.
24474 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24476 * tree-loop-distribution.c (classify_partition): New parameter and
24477 better handle reduction statement.
24478 (rdg_build_partitions): Revise comment.
24479 (distribute_loop): Compute statements in all partitions and pass it
24480 to classify_partition.
24482 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24484 * tree-loop-distribution.c (enum partition_type): New.
24485 (struct partition): New field type.
24486 (partition_merge_into): Add parameter. Update partition type.
24487 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24488 (build_rdg_partition_for_vertex): Compute partition type.
24489 (rdg_build_partitions): Dump partition type.
24490 (distribute_loop): Update calls to partition_merge_into.
24492 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24494 * tree-loop-distribution.c (struct ddr_hasher): New.
24495 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24497 (classify_partition): Call get_data_dependence.
24498 (pg_add_dependence_edges): Ditto.
24499 (distribute_loop): Release data dependence hash table.
24501 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24503 * tree-loop-distribution.c (ref_base_address): Delete.
24504 (similar_memory_accesses): Rename ...
24505 (share_memory_accesses): ... to this. Check if partitions access
24506 the same memory reference.
24507 (distribute_loop): Call share_memory_accesses.
24509 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24511 * tree-loop-distribution.c (struct partition): New field recording
24512 its data reference.
24513 (partition_alloc, partition_free): Init and release data refs.
24514 (partition_merge_into): Merge data refs.
24515 (build_rdg_partition_for_vertex): Collect data refs for partition.
24516 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24518 (distribute_loop): Remve data refs from vertice data of partition
24521 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24523 * tree-loop-distribution.c (params.h): Include header file.
24524 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24525 (datarefs_vec): New global var.
24526 (create_rdg_vertices): Use datarefs_vec directly.
24527 (free_rdg): Don't free data references.
24528 (build_rdg): Update use. Don't free data references.
24529 (distribute_loop): Compute global variable for data references.
24530 Bail out if there are too many data references.
24532 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24534 * tree-loop-distribution.c (loop_nest): New global var.
24535 (build_rdg): Use loop directly, rather than loop nest.
24536 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24538 (distribute_loop): Compute global variable loop nest. Update use.
24540 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24542 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24543 (partition_merge_into): New parameter. Dump reason for fusion.
24544 (distribute_loop): Update use of partition_merge_into.
24546 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24548 * tree-loop-distribution.c (bb_top_order_index): New.
24549 (bb_top_order_index_size, bb_top_order_cmp): New.
24550 (stmts_from_loop): Use topological order.
24551 (pass_loop_distribution::execute): Compute and release topological
24552 order for basic blocks.
24554 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24556 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24559 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24561 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24562 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24563 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24564 * internal-fn.def (LOOP_DIST_ALIAS): New.
24565 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24566 (fold_loop_internal_call): ... this.
24567 (vect_loop_dist_alias_call): New function.
24568 (set_uid_loop_bbs): Call fold_loop_internal_call.
24569 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24572 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24575 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24576 Require dead FLAGS_REG at the beginning of a peephole.
24578 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24581 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24582 arguments in the call to __builtin_ia32_sbb_u32.
24583 (_subborrow_u64): Swap _X and _Y arguments in the call to
24584 __builtin_ia32_sbb_u64.
24586 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24589 * tree-vrp.c (compare_assert_loc): Turn into a function template
24590 with stable template parameter. Only test if a->e is NULL,
24591 !a->e == !b->e has been verified already. Use e == NULL or
24592 e != NULL instead of e or ! e tests. If stable is true, don't use
24593 iterative_hash_expr, on the other side allow a or b or both NULL
24594 and sort the NULLs last.
24595 (process_assert_insertions): Sort using compare_assert_loc<false>
24596 instead of compare_assert_loc, later sort using
24597 compare_assert_loc<true> before calling process_assert_insertions_for
24598 in a loop. Use break instead of continue once seen NULL pointer.
24600 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24602 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24603 Cortex-R7 and Cortex-R8 processors.
24605 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24607 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24608 uninitialized while src is not.
24610 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24612 * common/config/arm/arm-common.c: Adjust include path for
24614 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24615 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24616 (arm-cpu-data.h): Likewise.
24617 (arm-cpu-cdata.h): Likewise.
24618 * config/arm/arm-cpu.h: Delete.
24619 * config/arm/arm-cpu-cdata.h: Delete.
24620 * config/arm/arm-cpu-data.h: Delete.
24622 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24624 * config/arm/arm-cpus.in (cortex-a55): New.
24625 (cortex-a75): Likewise.
24626 (cortex-a75.cortex-a55): Likewise.
24627 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24629 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24630 * config/arm/arm-cpu-cdata.h: Regenerate.
24631 * config/arm/arm-cpu-data.h: Regenerate.
24632 * config/arm/arm-cpu.h: Regenerate.
24633 * config/arm/arm-tables.opt: Regenerate.
24634 * config/arm/arm-tune.md: Regenerate.
24636 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24638 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24640 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24642 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24645 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24647 PR tree-optimization/81292
24648 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24649 full_string_p, also call adjust_related_strinfos if the adjustment
24650 is simple, otherwise invalidate related strinfos.
24652 2017-07-04 Martin Liska <mliska@suse.cz>
24655 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24656 newly created variable as DECL_IGNORED_P.
24658 2017-07-04 Martin Liska <mliska@suse.cz>
24661 * ipa-inline.c (inline_small_functions):
24662 Use xstrdup_for_dump.
24664 2017-07-04 Tom de Vries <tom@codesourcery.com>
24666 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24668 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24671 * config/darwin.c (darwin_function_switched_text_sections):
24674 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24676 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24678 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24680 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24682 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24684 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24685 min_profitable_iters, and th as inclusive lower bounds.
24686 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24687 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24688 for min_profitable_iters and min_profitable_estimate.
24689 (vect_transform_loop): Treat th as an inclusive lower bound.
24690 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24692 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24695 * config/darwin.c (darwin_function_switched_text_sections):
24696 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24697 in two pieces, and suppress the use of buf.
24699 2017-07-03 Nathan Sidwell <nathan@acm.org>
24701 * hash-table.h (hash_table_mod1): Fix indentation.
24703 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24705 PR middle-end/81290
24706 * predict.c (force_edge_cold): Be more careful about propagation
24708 * profile-count.h (profile_probability::guessed,
24709 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24711 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24713 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24715 * doc/invoke.texi (rcpc architecture extension): Document it.
24717 2017-07-03 Richard Biener <rguenther@suse.de>
24719 PR tree-optimization/60510
24720 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24721 the scalar reduction PHI and use it.
24722 (vectorizable_reduction): Properly guard the single_defuse_cycle
24723 path for non-SLP reduction chains where we cannot use it.
24724 Rework reduc_def/index and vector type deduction. Rework
24725 vector operand gathering during reduction op code-gen.
24726 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24727 chains dissolve the chain and leave it to non-SLP reduction
24730 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24732 * tree-data-ref.h (dr_alignment): Declare.
24733 * tree-data-ref.c (dr_alignment): New function.
24734 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24735 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24737 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24739 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24741 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24742 and base_misalignment fields.
24743 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24744 * tree-data-ref.c: Include builtins.h.
24745 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24746 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24747 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24748 * tree-vect-data-refs.c: Include tree-cfg.h.
24749 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24750 fields instead of calculating an alignment here.
24751 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24752 innermost_loop_behavior fields.
24754 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24756 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24758 (DR_STEP_ALIGNMENT): New macro.
24759 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24760 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24761 (create_data_ref): Print it.
24762 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24763 to tell whether the step preserves vector (mis)alignment.
24764 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24765 Move the check for an integer step and generalise to all INTEGER_CST.
24766 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24767 Print the outer step alignment.
24769 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24771 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24772 with offset_alignment.
24773 (DR_ALIGNED_TO): Delete.
24774 (DR_OFFSET_ALIGNMENT): New macro.
24775 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24776 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24777 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24778 (create_data_ref): Likewise.
24779 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24780 (vect_analyze_data_refs): Likewise.
24781 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24782 creating dummy innermost behavior.
24784 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24786 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24787 with a "innermost_loop_behavior *" and refeence tree.
24788 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24789 (create_data_ref): Update call accordingly.
24790 * tree-predcom.c (find_looparound_phi): Likewise.
24792 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24794 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24795 fields with dr_wrt_vec_loop.
24796 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24797 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24798 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24799 (vect_dr_behavior): New function.
24800 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24801 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24802 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24803 track whether the step preserves the misalignment.
24804 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24805 Use vect_dr_behavior.
24806 (vect_setup_realignment): Update call accordingly.
24807 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24808 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24809 call to vect_create_addr_base_for_vector_ref.
24810 (vect_create_cond_for_align_checks): Likewise.
24811 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24812 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24813 (vect_recog_mask_conversion_pattern): Likewise.
24814 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24815 (new_stmt_vec_info): Remove redundant zeroing.
24817 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24819 * common/config/arm/arm-common.c (arm_be8_option): New function.
24820 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24821 (ISA_ARMv6): Add isa_bit_be8.
24822 * config/arm/arm.h (arm_be8_option): Add prototype.
24823 (BE8_SPEC_FUNCTION): New define.
24824 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24825 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24826 (mlittle-endian): Similarly.
24827 (mbe8, mbe32): New options.
24828 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24829 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24831 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24833 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24835 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24837 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24838 (cleanup_tree_cfg_bb): Use it.
24839 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24841 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24843 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24846 * loop-doloop.c (add_test): Update profile.
24848 2017-07-03 Martin Liska <mliska@suse.cz>
24851 * sanopt.c (rewrite_usage_of_param): New function.
24852 (sanitize_rewrite_addressable_params): Likewise.
24853 (pass_sanopt::execute): Call rewrite_usage_of_param.
24855 2017-07-03 Richard Biener <rguenther@suse.de>
24857 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
24858 back to using VIEW_CONVERT_EXPR.
24860 2017-07-03 Martin Liska <mliska@suse.cz>
24863 * doc/extend.texi: Document when a resolver function is
24864 generated for target_clones.
24866 2017-07-03 Martin Liska <mliska@suse.cz>
24868 * asan.c (asan_emit_stack_protection): Unpoison just red zones
24869 and shadow memory of auto variables which are subject of
24870 use-after-scope sanitization.
24871 (asan_expand_mark_ifn): Add do set only when is_poison.
24873 2016-07-03 Richard Biener <rguenther@suse.de>
24875 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
24877 (vect_force_simple_reduction): Record reduction def -> phi mapping.
24878 (vectorizable_reduction): Perform reduction PHI creation when
24879 visiting a reduction PHI and adjust and simplify code generation
24880 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
24881 (vect_transform_loop): Visit reduction PHIs.
24882 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
24883 defs into the SLP tree.
24884 (vect_build_slp_tree): Reduction defs terminate the recursion.
24885 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
24887 (vect_get_vec_defs_for_stmt_copy): Export.
24888 (vect_get_vec_defs): Likewise.
24889 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
24891 (vect_get_vec_defs_for_stmt_copy): Declare.
24892 (vect_get_vec_defs): Likewise.
24894 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24896 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
24897 parameter with a "loop" parameter and use it instead of the
24898 loop containing DR_STMT. Don't check simple_iv when doing
24899 BB analysis. Describe the two analysis modes in the comment.
24901 2017-07-03 Tom de Vries <tom@codesourcery.com>
24903 PR tree-optimization/69468
24904 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
24905 (find_same_succ_bb): Handle ignore_edge_flags.
24907 2017-07-03 Tom de Vries <tom@codesourcery.com>
24909 PR tree-optimization/81192
24910 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
24912 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
24914 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
24916 2017-07-03 Tom de Vries <tom@codesourcery.com>
24918 PR tree-optimization/81192
24919 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
24920 BB_SAME_SUCC (bb) == NULL.
24922 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24924 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
24927 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24929 * dumpfile.c: Include profile-count.h
24930 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
24932 (insert_cond_bb): Update profile.
24933 * tree-cfg.h (insert_cond_bb): Update prototype.
24934 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
24935 * tree-dump.c: Do not include tree-cfg.
24937 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24939 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
24941 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24943 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
24946 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24948 * tree-complex.c (expand_complex_div_wide): update profile.
24950 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24951 Alan Hayward <alan.hayward@arm.com>
24952 David Sherwood <david.sherwood@arm.com>
24954 * Makefile.in (MACHMODE_H): Remove insn-modes.h
24955 (CORETYPES_H): New define.
24956 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
24957 (insn-modes-inline.h, s-modes-inline-h): New rules.
24958 (generated_files): Add insn-modes-inline.h.
24959 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
24960 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
24961 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
24962 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
24963 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
24964 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
24965 (build/gencodes.o, build/genconditions.o): Likewise.
24966 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
24967 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
24968 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
24969 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
24970 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
24971 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
24972 * coretypes.h: Include everything up to real.h for generators.
24973 Include insn-modes.h first. Include wide-int-print.h after
24974 wide-int.h. Include insn-modes-inline.h and then machmode.h.
24975 * machmode.h: Don't include insn-modes.h here.
24976 * function-tests.c: Remove includes of signop.h, machmode.h,
24977 double-int.h and wide-int.h.
24979 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
24981 * optc-save-gen.awk: Likewise.
24982 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
24983 * godump.c: Remove include of wide-int-print.h.
24984 * pretty-print.h: Likewise.
24985 * wide-int-print.cc: Likewise.
24986 * wide-int.cc: Likewise.
24987 * hash-map-tests.c: Remove include of signop.h.
24988 * hash-set-tests.c: Likewise.
24989 * rtl-tests.c: Likewise.
24990 * mkconfig.sh: Remove include of machmode.h.
24991 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
24993 (emit_insn_modes_inline_h): ...this new function. Emit the code
24994 into an insn-modes-inline.h header file, adding appropriate
24995 include guards and end comments.
24996 (emit_insn_modes_c_header): Remove include of machmode.h.
24997 (emit_min_insn_modes_c_header): Include coretypes.h rather than
24999 (main): Handle -i flag and call emit_insn_modes_inline_h when
25002 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25004 * tree-ssa-strlen.c (strinfo): Rename the length field to
25005 nonzero_chars. Add a full_string_p field.
25006 (compare_nonzero_chars, zero_length_string_p): New functions.
25007 (get_addr_stridx): Add an offset_out parameter.
25008 Use compare_nonzero_chars.
25009 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25010 (new_strinfo): Update after above changes to strinfo.
25011 (set_endptr_and_length): Set full_string_p.
25012 (get_string_length): Update after above changes to strinfo.
25013 (unshare_strinfo): Update call to new_strinfo.
25014 (maybe_invalidate): Likewise.
25015 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25016 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25017 as a uhwi instead of an shwi. Update after above changes to
25018 strinfo and new_strinfo.
25019 (zero_length_string): Assert that chainsi contains full strings.
25020 Use zero_length_string_p. Update call to new_strinfo.
25021 (adjust_related_strinfos): Update after above changes to strinfo.
25022 Copy full_string_p from origsi.
25023 (adjust_last_stmt): Use zero_length_string_p.
25024 (handle_builtin_strlen): Update after above changes to strinfo and
25025 new_strinfo. Install the lhs as the string length if the previous
25026 entry didn't describe a full string.
25027 (handle_builtin_strchr): Update after above changes to strinfo
25029 (handle_builtin_strcpy): Likewise.
25030 (handle_builtin_strcat): Likewise.
25031 (handle_builtin_malloc): Likewise.
25032 (handle_pointer_plus): Likewise.
25033 (handle_builtin_memcpy): Likewise. Track nonzero characters
25034 that aren't necessarily followed by a nul terminator.
25035 (handle_char_store): Likewise.
25037 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25039 PR tree-optimization/80769
25040 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25041 for malloc and calloc. Document the new invariant that all related
25042 strinfos have delayed lengths or none do.
25043 (verify_related_strinfos): Move earlier in file.
25044 (set_endptr_and_length): New function, split out from...
25045 (get_string_length): ...here. Also set the lengths of related
25047 (zero_length_string): Assert that chainsi has known (rather than
25049 (adjust_related_strinfos): Likewise.
25051 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25053 PR tree-optimization/81136
25054 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25055 assert that two references with the same misalignment have the same
25056 compile-time misalignment if those compile-time misalignments
25059 2017-07-01 Andi Kleen <ak@linux.intel.com>
25061 * print-tree.c (print_node): Print all attributes.
25063 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25065 * cfg.c (scale_bbs_frequencies): New function.
25066 * cfg.h (scale_bbs_frequencies): Declare it.
25067 * cfgloopanal.c (single_likely_exit): Cleanup.
25068 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25070 (scale_loop_profile): Likewise.
25071 (loop_version): Likewise.
25072 (create_empty_loop_on_edge): Update.
25073 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25074 scale_loop_frequencies, scale_loop_profile, loopify,
25075 loop_version): Update prototypes.
25076 * modulo-sched.c (sms_schedule): Update.
25077 * predict.c (unlikely_executed_edge_p): Also check probability.
25078 (probably_never_executed_edge_p): Fix typo.
25079 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25080 * tree-parloops.c (gen_parallel_loop): Update.
25081 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25082 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25083 * tree-ssa-loop-split.c (split_loop): Update.
25084 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25085 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25086 (vect_loop_versioning): Update.
25087 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25089 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25091 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25093 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25095 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25096 to keep profile consistent.
25098 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25100 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25101 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25102 * profile-count.h (max_safe_multiplier): Make unsigned.
25103 (profile_count::guessed_zero): New.
25105 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25107 * bb-reorder.c (fix_up_crossing_landing_pad,
25108 fix_crossing_conditional_branches): Use make_single_succ_edge
25109 to keep profile consistent.
25111 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25113 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25116 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25119 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25120 the right scopes, make sure cond_jump isn't preserved between multiple
25121 iterations. Search for fallthru edge whenever there are 3+ edges and
25122 use find_fallthru_edge for it.
25124 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25126 Patch by Alexander Monakov <amonakov@ispras.ru>
25127 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25128 probabilities consistently.
25130 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25132 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25133 emit_cmp_and_jump_insns.
25135 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25138 * tree-inline.c (expand_call_inline): Combine profile statuses.
25140 2017-06-30 Andrew Pinski <apinski@cavium.com>
25142 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25143 fold_stmt returned true.
25145 2017-06-30 Nathan Sidwell <nathan@acm.org>
25147 * ggc.h (empty_string): Delete.
25148 * cfgexpand.c (expand_asm_stmt): Use plain "".
25149 * optabs.c (expand_asm_memory_barrier): Likewise.
25150 * stringpool.c (empty_string): Delete.
25151 (digit_vector, digit_string): Delete.
25152 (ggc_alloc_string): Use plain "", don't optimize single digit
25153 strings. Use ggc_alloc_atomic.
25155 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25157 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25158 comparison set and one other set, use the cost of the non-comparison
25161 2017-06-30 Nathan Sidwell <nathan@acm.org>
25163 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25166 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25168 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25169 loops. Remove now unneeded calls to gimple_switch_set_label() that
25170 just set removed labels to NULL_TREE.
25172 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25174 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25175 (set_range_info): ...here. Only call set_range_info_raw if domain
25177 (set_nonzero_bits): Call set_range_info_raw.
25178 * tree-ssanames.h (set_range_info_raw): New.
25180 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25183 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25184 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25185 of nonimmediate_operand and <store_mask_constraint> instead of m
25186 for the input operand. For V8FI iterator, always split if input
25187 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25188 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25189 <store_mask_predicate> instead of register_operand and
25190 <store_mask_constraint> instead of v for the input operand. Make
25191 sure both operands aren't MEMs for if not <mask_applied>.
25193 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25195 * lto-wrapper.c (copy_file) Close both file descriptors before
25198 2017-06-30 Martin Liska <mliska@suse.cz>
25201 * multiple_target.c (create_dispatcher_calls): Make ifunc
25202 also for function that don't have calls or are not referenced.
25204 2017-06-30 Richard Biener <rguenther@suse.de>
25206 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25207 analyze the first scalar stmt. Move vector type computation
25208 for the BB case here from ...
25209 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25210 live operation processing in the SLP case properly.
25212 2017-06-30 Richard Biener <rguenther@suse.de>
25214 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25216 2017-06-30 Martin Liska <mliska@suse.cz>
25219 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25220 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25222 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25224 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25227 2017-06-30 Richard Biener <rguenther@suse.de>
25229 PR tree-optimization/81249
25230 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25231 condition reduction result to original scalar type.
25233 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25235 * profile-count.h (enum profile_quality): Fix typos and whitespace
25238 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25240 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25241 type for branch probabilities.
25243 2017-06-29 Julian Brown <julian@codesourcery.com>
25244 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25246 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25247 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25248 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25249 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25251 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25253 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25254 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25255 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25256 CC usage from generic code to here.
25257 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25258 CC usage into the target macros.
25260 2017-06-29 Maya Rashish <coypu@sdf.org>
25262 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25265 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25267 * arm/arm-builtins.c: Include profile-count.h
25268 * except.c (sjlj_emit_function_enter): Use
25269 profile_probability::unlikely.
25271 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25273 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25274 and tocrel_offset be pointer args rather than implicitly using
25276 (legitimate_constant_pool_address_p, rs6000_emit_move,
25277 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25278 tocrel_offset and use in toc_relative_expr_p call.
25279 (print_operand, print_operand_address): Use static tocrel_base_oac
25280 and tocrel_offset_oac.
25281 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25284 2017-06-29 Maya Rashish <coypu@sdf.org>
25286 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25288 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25290 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25291 objects, take into account only the alignment of 'op0' and 'mode1' if
25294 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25296 * ccmp.c (ccmp_tree_comparison_p): New function.
25297 (ccmp_candidate_p): Update to use above function.
25298 (get_compare_parts): New function.
25299 (expand_ccmp_next): Update to use new functions.
25300 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25302 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25303 take mode as argument.
25304 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25305 * expr.c (expand_expr_real_1): Pass mode as argument.
25307 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25309 * combine.c (combine_instructions): Print insns to dump_file, together
25312 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25314 * asan.c (asan_emit_stack_protection): Update.
25315 (create_cond_insert_point): Update.
25316 * auto-profile.c (afdo_propagate_circuit): Update.
25317 * basic-block.h (struct edge_def): Turn probability to
25318 profile_probability.
25319 (EDGE_FREQUENCY): Update.
25320 * bb-reorder.c (find_traces_1_round): Update.
25321 (better_edge_p): Update.
25322 (sanitize_hot_paths): Update.
25323 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25324 (make_single_succ_edge): Update.
25325 (check_bb_profile): Update.
25326 (dump_edge_info): Update.
25327 (update_bb_profile_for_threading): Update.
25328 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25329 probabilitycount to 0.
25330 * cfgbuild.c (compute_outgoing_frequencies): Update.
25331 * cfgcleanup.c (try_forward_edges): Update.
25332 (outgoing_edges_match): Update.
25333 (try_crossjump_to_edge): Update.
25334 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25335 (expand_gimple_tailcall): Update.
25336 (construct_init_block): Use make_single_succ_edge.
25337 (construct_exit_block): Use make_single_succ_edge.
25338 * cfghooks.c (verify_flow_info): Update.
25339 (redirect_edge_succ_nodup): Update.
25340 (split_edge): Update.
25341 (account_profile_record): Update.
25342 * cfgloopanal.c (single_likely_exit): Update.
25343 * cfgloopmanip.c (scale_loop_profile): Update.
25344 (set_zero_probability): Remove.
25345 (duplicate_loop_to_header_edge): Update.
25346 * cfgloopmanip.h (loop_version): Update prototype.
25347 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25348 (force_nonfallthru_and_redirect): Update.
25349 (update_br_prob_note): Update.
25350 (rtl_verify_edges): Update.
25351 (purge_dead_edges): Update.
25352 (rtl_lv_add_condition_to_bb): Update.
25353 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25354 * cgraphunit.c (init_lowered_empty_function): Update.
25355 (cgraph_node::expand_thunk): Update.
25356 * cilk-common.c: Include profile-count.h
25357 * dojump.c (inv): Remove.
25358 (jumpifnot): Update.
25359 (jumpifnot_1): Update.
25360 (do_jump_1): Update.
25362 (do_jump_by_parts_greater_rtx): Update.
25363 (do_compare_rtx_and_jump): Update.
25364 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25365 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25366 * dwarf2cfi.c: Include profile-count.h
25367 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25368 (sjlj_emit_dispatch_table): Likewise.
25369 * explow.c: Include profile-count.h
25370 * expmed.c (emit_store_flag_force): Update.
25371 (do_cmp_and_jump): Update.
25372 * expr.c (compare_by_pieces_d::generate): Update.
25373 (compare_by_pieces_d::finish_mode): Update.
25374 (emit_block_move_via_loop): Update.
25375 (store_expr_with_bounds): Update.
25376 (store_constructor): Update.
25377 (expand_expr_real_2): Update.
25378 (expand_expr_real_1): Update.
25379 * expr.h (try_casesi, try_tablejump): Update prototypes.
25380 * gimple-pretty-print.c (dump_probability): Update.
25381 (dump_profile): New.
25382 (dump_gimple_label): Update.
25383 (dump_gimple_bb_header): Update.
25384 * graph.c (draw_cfg_node_succ_edges): Update.
25385 * hsa-gen.c (convert_switch_statements): Update.
25386 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25387 (find_if_case_1): Update.
25388 (find_if_case_2): Update.
25389 * internal-fn.c (expand_arith_overflow_result_store): Update.
25390 (expand_addsub_overflow): Update.
25391 (expand_neg_overflow): Update.
25392 (expand_mul_overflow): Update.
25393 (expand_vector_ubsan_overflow): Update.
25394 * ipa-cp.c (good_cloning_opportunity_p): Update.
25395 * ipa-split.c (split_function): Use make_single_succ_edge.
25396 * ipa-utils.c (ipa_merge_profiles): Update.
25397 * loop-doloop.c (add_test): Update.
25398 (doloop_modify): Update.
25399 * loop-unroll.c (compare_and_jump_seq): Update.
25400 (unroll_loop_runtime_iterations): Update.
25401 * lra-constraints.c (lra_inheritance): Update.
25402 * lto-streamer-in.c (input_cfg): Update.
25403 * lto-streamer-out.c (output_cfg): Update.
25404 * mcf.c (adjust_cfg_counts): Update.
25405 * modulo-sched.c (sms_schedule): Update.
25406 * omp-expand.c (expand_omp_for_init_counts): Update.
25407 (extract_omp_for_update_vars): Update.
25408 (expand_omp_ordered_sink): Update.
25409 (expand_omp_for_ordered_loops): Update.
25410 (expand_omp_for_generic): Update.
25411 (expand_omp_for_static_nochunk): Update.
25412 (expand_omp_for_static_chunk): Update.
25413 (expand_cilk_for): Update.
25414 (expand_omp_simd): Update.
25415 (expand_omp_taskloop_for_outer): Update.
25416 (expand_omp_taskloop_for_inner): Update.
25417 * omp-simd-clone.c (simd_clone_adjust): Update.
25418 * optabs.c (expand_doubleword_shift): Update.
25419 (expand_abs): Update.
25420 (emit_cmp_and_jump_insn_1): Update.
25421 (expand_compare_and_swap_loop): Update.
25422 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25423 * predict.c (predictable_edge_p): Update.
25424 (edge_probability_reliable_p): Update.
25425 (set_even_probabilities): Update.
25426 (combine_predictions_for_insn): Update.
25427 (combine_predictions_for_bb): Update.
25428 (propagate_freq): Update.
25429 (estimate_bb_frequencies): Update.
25430 (force_edge_cold): Update.
25431 * profile-count.c (profile_count::dump): Add missing space into dump.
25432 (profile_count::debug): Add newline.
25433 (profile_count::differs_from_p): Explicitly convert to unsigned.
25434 (profile_count::stream_in): Update.
25435 (profile_probability::dump): New member function.
25436 (profile_probability::debug): New member function.
25437 (profile_probability::differs_from_p): New member function.
25438 (profile_probability::differs_lot_from_p): New member function.
25439 (profile_probability::stream_in): New member function.
25440 (profile_probability::stream_out): New member function.
25441 * profile-count.h (profile_count_quality): Rename to ...
25442 (profile_quality): ... this one.
25443 (profile_probability): New.
25444 (profile_count): Update.
25445 * profile.c (compute_branch_probabilities): Update.
25446 * recog.c (peep2_attempt): Update.
25447 * sched-ebb.c (schedule_ebbs): Update.
25448 * sched-rgn.c (find_single_block_region): Update.
25449 (compute_dom_prob_ps): Update.
25450 (schedule_region): Update.
25451 * sel-sched-ir.c (compute_succs_info): Update.
25452 * stmt.c (struct case_node): Update.
25453 (do_jump_if_equal): Update.
25454 (get_outgoing_edge_probs): Update.
25455 (conditional_probability): Update.
25456 (emit_case_dispatch_table): Update.
25457 (expand_case): Update.
25458 (expand_sjlj_dispatch_table): Update.
25459 (emit_case_nodes): Update.
25460 * targhooks.c: Update.
25461 * tracer.c (better_p): Update.
25462 (find_best_successor): Update.
25463 * trans-mem.c (expand_transaction): Update.
25464 * tree-call-cdce.c: Update.
25465 * tree-cfg.c (gimple_split_edge): Upate.
25466 (move_sese_region_to_fn): Upate.
25467 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25468 * tree-eh.c (lower_resx): Upate.
25469 (cleanup_empty_eh_move_lp): Upate.
25470 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25471 * tree-inline.c (copy_edges_for_bb): Update.
25472 (copy_cfg_body): Update.
25473 * tree-parloops.c (gen_parallel_loop): Update.
25474 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25475 (gimple_gen_time_profiler): Update.
25476 * tree-ssa-dce.c (remove_dead_stmt): Update.
25477 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25478 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25479 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25480 (unloop_loops): Update.
25481 (try_peel_loop): Update.
25482 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25483 * tree-ssa-loop-split.c (connect_loops): Update.
25484 (split_loop): Update.
25485 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25486 (hoist_guard): Update.
25487 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25488 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25489 (value_replacement): Update.
25490 * tree-ssa-reassoc.c (branch_fixup): Update.
25491 * tree-ssa-tail-merge.c (replace_block_by): Update.
25492 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25493 (create_edge_and_update_destination_phis): Update.
25494 (compute_path_counts): Update.
25495 (recompute_probabilities): Update.
25496 (update_joiner_offpath_counts): Update.
25497 (freqs_to_counts_path): Update.
25498 (duplicate_thread_path): Update.
25499 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25500 (struct switch_conv_info): Update.
25501 (gen_inbound_check): Update.
25502 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25503 (vect_do_peeling): Update.
25504 (vect_loop_versioning): Update.
25505 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25506 (optimize_mask_stores): Update.
25507 * ubsan.c (ubsan_expand_null_ifn): Update.
25508 * value-prof.c (gimple_divmod_fixed_value): Update.
25509 (gimple_divmod_fixed_value_transform): Update.
25510 (gimple_mod_pow2): Update.
25511 (gimple_mod_pow2_value_transform): Update.
25512 (gimple_mod_subtract): Update.
25513 (gimple_mod_subtract_transform): Update.
25514 (gimple_ic): Update.
25515 (gimple_stringop_fixed_value): Update.
25516 (gimple_stringops_transform): Update.
25517 * value-prof.h: Update.
25519 2017-06-29 Carl Love <cel@us.ibm.com>
25521 * config/rs6000/rs6000-c.c: Add support for built-in functions
25522 vector signed int vec_signed (vector float);
25523 vector signed long long vec_signed (vector double);
25524 vector signed int vec_signed2 (vector double, vector double);
25525 vector signed int vec_signede (vector double);
25526 vector signed int vec_signedo (vector double);
25527 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25528 instruction generator.
25529 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25530 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25531 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25533 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25534 vunsignede_v2df): Add define_expands.
25535 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25536 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25537 VEC_UNSIGNEDO): Add definitions.
25538 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25539 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25540 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25541 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25542 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25543 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25544 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25545 vec_unsignede, vec_unsignedo): Add builtin defines.
25546 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25548 * doc/extend.texi: Update the built-in documentation file for the
25549 new built-in functions.
25551 2017-06-29 Richard Biener <rguenther@suse.de>
25553 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25554 reduction chains to LOOP_VINFO_REDUCTIONS.
25555 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25556 SLP reductions after processing reduction chains.
25558 2017-06-29 Nathan Sidwell <nathan@acm.org>
25560 * builtins.c (fold_builtin_FUNCTION): Use
25561 lang_hooks.decl_printable_name.
25563 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25565 PR middle-end/81194
25566 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25567 with only one label.
25568 * stmt.c (expand_case): Assert NCASES is greater than one.
25570 2017-06-29 Richard Biener <rguenther@suse.de>
25572 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25574 (group_case_labels): Likewise.
25575 (find_taken_edge): Push sanity checking on val to workers...
25576 (find_taken_edge_cond_expr): ... here
25577 (find_taken_edge_switch_expr): ... and here, handle cases
25578 with just a default label.
25579 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25580 (group_case_labels): Likewise.
25581 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25582 group_case_labels does anything cleanup the CFG again.
25584 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25586 PR tree-optimization/81196
25587 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25588 exit condition comparing two IVs.
25590 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25592 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25593 profile to the dummy entry at the end of the list of architectures.
25594 * config/arm/arm-cpu-cdata.h: Regenerated.
25596 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25597 Michael Collison <michael.collison@arm.com>
25600 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25602 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25603 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25604 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25605 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25606 of shift when the shift amount is masked with constant equal to
25607 the size of the mode.
25608 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25611 2017-06-29 Martin Liska <mliska@suse.cz>
25613 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25615 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25618 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25621 (edge_growth_cache_entry::edge_growth_cache_entry): New
25623 (reset_edge_growth_cache): Update to use constructor.
25625 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25627 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25628 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25629 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25631 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25633 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25634 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25636 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25638 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25639 (*-linux-uclibc*): Add t-uclibc tmake_file.
25640 * config/t-musl: New.
25641 * config/t-uclibc: New.
25643 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25645 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25647 (gen_comm_data): Emit architectural setting of arch_prof.
25648 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25650 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25651 (armv8-m.base, armv8-m.main): Likewise.
25652 * arm-protos.h (arm_build_target): Add profile field.
25653 (arch_option): Likewise.
25654 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25656 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25657 arm_active_target.profile.
25659 2017-06-28 Richard Biener <rguenther@suse.de>
25661 PR middle-end/81227
25662 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25663 TYPE_OVERFLOW_WRAPS.
25664 * match.pd (negate_expr_p): Likewise.
25665 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25666 fold_build2, not fold_binary.
25668 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25670 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25671 Convert memory address to Pmode.
25672 (aarch64_print_operand): Assert MEM operands are always Pmode.
25674 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25677 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25678 Remove redundant if.
25679 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25680 * config/arm/aarch-common-protos.h
25681 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25682 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25684 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25687 * multiple_target.c (create_dispatcher_calls): Set the default
25688 clone to be static, not public.
25690 2017-06-28 Richard Biener <rguenther@suse.de>
25692 * tree-vect-loop.c (vectorizable_reduction): Move special
25693 cond reduction IV var creation ...
25694 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25695 parameter. Use STMT_VINFO_VECTYPE.
25696 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25699 2017-06-28 Martin Liska <mliska@suse.cz>
25702 * ipa-visibility.c (non_local_p): Handle visibility.
25704 2017-06-28 Martin Liska <mliska@suse.cz>
25707 * common.opt: Add IntegerRange to various options.
25708 * opt-functions.awk (integer_range_info): New function.
25709 * optc-gen.awk: Add integer_range_info to cl_options struct.
25710 * opts-common.c (decode_cmdline_option): Handle
25711 CL_ERR_INT_RANGE_ARG.
25712 (cmdline_handle_error): Likewise.
25713 * opts.c (print_filtered_help): Show valid interval in
25714 when --help is provided.
25715 * opts.h (struct cl_option): Add range_min and range_max fields.
25716 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25718 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25720 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25721 (x * C EQ/NE y * C): New transformation.
25723 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25725 * genmultilib (combination_space): Accept '+' in option names.
25727 2017-06-28 Martin Liska <mliska@suse.cz>
25730 * asan.c (instrument_derefs): Bail out inner references
25731 that are hard register variables.
25733 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25736 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25737 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25739 2017-06-28 Richard Biener <rguenther@suse.de>
25741 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25742 (vect_get_slp_defs): Adjust.
25743 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25745 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25747 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25748 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25749 (vectorizable_reduction): Adjust.
25750 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25752 (vect_get_slp_defs): Likewise.
25753 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25754 (vectorizable_bswap): Adjust.
25755 (vectorizable_call): Likewise.
25756 (vectorizable_conversion): Likewise.
25757 (vectorizable_assignment): Likewise.
25758 (vectorizable_shift): Likewise.
25759 (vectorizable_operation): Likewise.
25760 (vectorizable_store): Likewise.
25761 (vectorizable_condition): Likewise.
25762 (vectorizable_comparison): Likewise.
25764 2017-06-28 Michael Collison <michael.collison@arm.com>
25767 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25769 (arm_gen_movmemqi): Removed unused variable 'i'.
25770 Convert 'for' loop into 'while' loop.
25771 (arm_expand_prologue): Remove last unnecessary set of insn.
25772 (thumb_pop): Remove unused variable 'pushed_words'.
25773 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25775 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25777 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25778 * config/s390/s390-protos.h: Add prototype of
25779 s390_rel_address_ok_p.
25780 * config/s390/s390.c (s390_got_symbol): New function.
25781 (s390_rel_address_ok_p): New function.
25782 (legitimize_pic_address): Use s390_rel_address_ok_p.
25783 (s390_load_got): Use s390_got_symbol.
25784 (s390_option_override): Issue error if
25785 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25786 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25788 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25790 2017-06-27 Andrew Pinski <apinski@cavium.com>
25792 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25793 (X * copysign (1.0, X)): New pattern.
25794 (X * copysign (1.0, -X)): New pattern.
25795 (copysign (-1.0, CST)): New pattern.
25797 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25799 * genmultilib (combination_space): Remove variable.
25800 Validate reuse rules against regular expression for any sequence
25801 of multilib options in any order.
25803 2017-06-27 Michael Collison <michael.collison@arm.com>
25805 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25806 call aarch64_split_simd_combine.
25807 * (aarch64_combine_internal<mode>): Delete pattern.
25808 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25809 Allow register and subreg operands.
25811 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25813 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25814 specific need, just fallback on defaults.
25815 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25817 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25818 Olivier Hainque <hainque@adacore.com>
25820 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25822 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25823 targets. Pick a default if no particular attempt applied.
25824 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25827 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25829 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25830 (x86_64-wrs-vxworks7): Likewise.
25832 2017-06-27 Marek Polacek <polacek@redhat.com>
25835 * ubsan.c (instrument_null): Check get_base_address's result for null.
25837 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25839 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25841 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25843 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25844 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25845 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25846 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25847 New function types.
25848 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25849 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25850 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
25851 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
25852 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
25853 BUILT_IN_FEUPDATEENV): New builtins.
25854 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
25855 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
25856 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
25857 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
25859 (builtin_structptr_types): Adjust size.
25860 * tree.c (builtin_structptr_types): Add four entries.
25862 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25863 Olivier Hainque <hainque@adacore.com>
25865 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
25866 (TLS_SYM): New local macro, forcing reference to __tls__ on
25867 link command lines for VxWorks 7 RTPs, triggering initialization
25869 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
25870 OS features TLS support, true for RTPs on VxWorks 7.
25871 * config/vxworks.c (vxworks_override_options): Setup emutls
25874 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25876 * predict.c (test_prediction_value_range): Use -1U instead of -1
25877 to avoid narrowing conversion warning.
25878 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
25879 to avoid narrowing conversion warning.
25880 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
25882 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
25884 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25886 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
25887 64bit configurations.
25888 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
25889 (SIZE_TYPE): Likewise.
25890 * config/vxworks.c (vxworks_emutls_var_fields): Use
25891 long_unsigned_type_node instead of unsigned_type_node as the offset
25892 field type, which is "pointer" mode in emutls.c.
25894 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25897 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
25899 PR middle-end/81207
25900 * gimple-fold.c (replace_call_with_call_and_fold): Handle
25901 gimple_vuse copying separately from gimple_vdef copying.
25903 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25905 * value-prof.c (free_hist): Remove call to memset and the enclosing if
25908 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25909 Olivier Hainque <hainque@adacore.com>
25911 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
25912 for all vxworks7 targets.
25913 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
25914 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
25915 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
25916 variations for VX6/VX7 and 32/64bits later on in ...
25917 (VXWORKS_LIB_SPEC): Leverage new macros.
25918 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
25919 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
25921 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25923 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
25924 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
25926 2017-06-26 Carl Love <cel@us.ibm.com>
25928 * config/rs6000/rs6000-c.c: Add support for built-in functions
25929 vector bool char vec_reve (vector bool char);
25930 vector signed char vec_reve (vector signed char);
25931 vector unsigned char vec_reve (vector unsigned char);
25932 vector bool int vec_reve (vector bool int);
25933 vector signed int vec_reve (vector signed int);
25934 vector unsigned int vec_reve (vector unsigned int);
25935 vector bool long long vec_reve (vector bool long long);
25936 vector signed long long vec_reve (vector signed long long);
25937 vector unsigned long long vec_reve (vector unsigned long long);
25938 vector bool short vec_reve (vector bool short);
25939 vector signed short vec_reve (vector signed short);
25940 vector double vec_reve (vector double);
25941 vector float vec_reve (vector float);
25942 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
25943 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
25944 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
25945 (altivec_vreve): New pattern.
25946 * config/rs6000/altivec.h (vec_reve): New define.
25947 * doc/extend.texi (vec_rev): Update the built-in documentation file
25948 for the new built-in functions.
25950 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25952 PR tree-optimization/71815
25953 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
25955 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
25957 (slsr_process_phi): Likewise.
25958 (replace_uncond_cands_and_profitable_phis): Don't replace a
25959 multiply candidate with a stride of 1 (copy or cast).
25960 (phi_incr_cost): Call uses_consumed_by_stmt rather than
25962 (lowest_cost_path): Likewise.
25963 (total_savings): Likewise.
25965 2017-06-26 Richard Biener <rguenther@suse.de>
25968 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
25969 Use def_builtin_pure for all gather builtins.
25971 2017-06-26 Richard Biener <rguenther@suse.de>
25973 PR tree-optimization/81203
25974 * tree-tailcall.c (find_tail_calls): Do not move stmts into
25975 non-dominating BBs.
25977 2017-06-26 Marek Polacek <polacek@redhat.com>
25980 * doc/invoke.texi: Document -Wmultistatement-macros.
25982 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
25984 * doc/sourcebuild.texi (ARM-specific attributes): Document new
25985 arm_neon_ok_no_float_abi effective target.
25987 2017-06-26 Richard Biener <rguenther@suse.de>
25989 PR tree-optimization/80928
25990 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
25991 (copy_bbs): Set BB_DUPLICATED flag early.
25992 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
25994 (execute_on_shrinking_pred): Likewise.
25995 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
25996 BB_DUPLICATED blocks.
25997 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
25998 iterate over all PHIs considering removal of *gsi.
26000 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26002 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26005 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26007 * config/rs6000/rs6000-string.c: (expand_block_clear,
26008 do_load_for_compare, select_block_compare_mode,
26009 compute_current_alignment, expand_block_compare,
26010 expand_strncmp_align_check, expand_strn_compare,
26011 expand_block_move, rs6000_output_load_multiple)
26012 Move functions related to string/block move/compare
26013 to a separate file.
26014 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26015 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26016 for this function which is now used in two files.
26017 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26018 * config.gcc: Add rs6000-string.o to extra_objs for
26019 targets powerpc*-*-* and rs6000*-*-*.
26021 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26024 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26025 32-bit, since indexed is not valid for DImode.
26026 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26027 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26028 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26029 (define_peephole2 for Altivec d-form store): Likewise.
26032 * multiple_target.c (create_dispatcher_calls): Only create the
26033 dispatcher call if the function is the default clone of a
26034 versioned function.
26036 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26038 PR middle-end/80902
26039 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26040 a call, force the call to not be a tail call.
26042 2017-06-23 Jeff Law <law@redhat.com>
26044 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26047 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26049 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26050 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26051 handling for early expansion of vector shifts (sl,sr,sra,rl).
26052 (builtin_function_type): Add vector shift right instructions
26053 to the unsigned argument list.
26055 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26057 rtl-optimizatoin/79286
26058 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26059 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26060 trap. PIC register plus a const unspec without offset can never trap.
26062 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26064 * tree.h (builtin_structptr_type): New type.
26065 (builtin_structptr_types): Declare new array.
26066 * tree.c (builtin_structptr_types): New array.
26067 (free_lang_data, build_common_tree_nodes): Use it.
26069 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26072 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26075 2017-06-22 Matt Turner <mattst88@gmail.com>
26077 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26078 Lake models to skylake case. Assume skylake for unknown
26079 models with clflushopt.
26081 2017-06-22 Jeff Law <law@redhat.com>
26083 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26084 frame sizes that do not satisfy aarch64_uimm12_shift.
26086 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26088 * profile-count.h (apply_probability,
26089 apply_scale, probability_in): Fix checks for zero.
26091 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26093 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26094 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26096 2016-06-22 Richard Biener <rguenther@suse.de>
26098 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26099 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26100 REDUC_MAX_EXPR support.
26101 (vectorizable_reduction): Likewise.
26102 (vect_create_epilog_for_reduction): Likewise.
26104 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26106 * match.pd (A / (1 << B) -> A >> B): New.
26107 * generic-match-head.c: Include optabs-tree.h.
26108 * gimple-match-head.c: Likewise.
26109 * optabs-tree.h (target_supports_op_p): New.
26110 * optabs-tree.c (target_supports_op_p): New.
26112 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26114 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26115 $gcc_cv_ld --help output.
26116 (gcc_cv_ld_demangle): Likewise.
26117 (gcc_cv_ld_eh_frame_hdr): Likewise.
26118 (gcc_cv_ld_pie): Likewise.
26119 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26120 (gcc_cv_ld_buildid): Likewise.
26121 (gcc_cv_ld_sysroot): Likewise.
26122 (ld_bndplt_support): Likewise.
26123 (ld_pushpopstate_support): Likewise.
26124 * configure: Regenerate.
26125 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26127 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26130 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26131 operands indexes to avoid gap between operands and match_dups.
26133 2017-06-21 Andrew Pinski <apinski@cavium.com>
26135 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26136 Increment Arith_shift and Arith_shift_reg by 1.
26137 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26139 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26140 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26141 (aarch64_strip_extend): Add new argument and test for it.
26142 (aarch64_cheap_mult_shift_p): New function.
26143 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26144 add a cost if it is true.
26145 Update calls to aarch64_strip_extend.
26146 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26148 2017-06-21 Andrew Pinski <apinski@cavium.com>
26150 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26152 (thunderxt88): Likewise.
26153 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26154 (thunderx_prefetch_tune): New variable.
26155 (thunderx2t99_prefetch_tune): Update for the correct values.
26156 (thunderxt88_tunings): New variable.
26157 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26158 generic_prefetch_tune.
26159 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26161 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26163 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26164 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26165 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26166 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26167 (aarch64_atomic_cas<mode>, GPI): Likewise.
26169 2017-06-21 Martin Liska <mliska@suse.cz>
26171 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26172 statements on cold and hot labels.
26173 * predict.c (tree_estimate_probability_bb): Remove the
26174 prediction from this place.
26176 2017-06-21 Martin Liska <mliska@suse.cz>
26178 PR tree-optimization/79489
26179 * gimplify.c (maybe_add_early_return_predict_stmt): New
26181 (gimplify_return_expr): Call the function.
26182 * predict.c (tree_estimate_probability_bb): Remove handling
26184 * predict.def: Update comment about early return predictor.
26185 * gimple-predict.h (is_gimple_predict): New function.
26186 * predict.def: Change default value of early return to 66.
26187 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26189 * passes.def: Put pass_strip_predict_hints to the beginning of
26192 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26194 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26195 FUNCTION_DECL declarations.
26196 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26198 (dwaf2out_decl): Likewise.
26199 * godump.c (go_early_global_decl): Skip call to the real debug hook
26200 for FUNCTION_DECL declarations.
26201 * passes.c (rest_of_decl_compilation): Skip call to the
26202 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26203 -fdump-go-spec is passed.
26205 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26207 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26208 Reorder for compactness.
26209 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26210 (def_builtin_pure, def_builtin_pure2): New functions.
26211 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26213 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26215 * match.pd (nop_convert): New predicate.
26216 ((A +- CST1) +- CST2): Allow some NOP conversions.
26218 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26221 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26222 with ctors/dtors if GOVD_SHARED is set.
26224 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26226 * config/aarch64/aarch64.md (movti_aarch64):
26227 Emit mov rather than orr.
26228 (movtf_aarch64): Likewise.
26229 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26230 Emit mov rather than orr.
26232 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26234 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26235 Swap alternatives, make integer dup more expensive.
26237 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26239 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26240 Return true for non-tls symbols.
26242 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26244 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26245 (cortex-a75): Likewise.
26246 (cortex-a75.cortex-a55): Likewise.
26247 * config/aarch64/aarch64-tune.md: Regenerate.
26248 * doc/invoke.texi (-mtune): Document new values for -mtune.
26250 2017-06-21 Tom de Vries <tom@codesourcery.com>
26252 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26253 stack_size feature.
26254 (Effective-Target Keywords, Other attributes): Suggest using
26255 dg-add-options stack_size feature to get stack limit in stack_size
26256 effective target documentation.
26258 2017-06-21 Julian Brown <julian@codesourcery.com>
26259 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26261 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26262 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26263 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26265 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26266 attribute type list for neon_multiply.
26267 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26268 attribute type list for neon_multiply.
26269 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26270 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26271 attribute type list for neon_multiply.
26272 * config/arm/types.md (crypto_pmull): Add.
26273 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26274 attribute type list.
26276 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26278 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26281 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26283 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26284 to make sure not to dereference a NULL cost_classes_ptr pointer.
26286 2017-06-20 Carl Love <cel@us.ibm.com>
26288 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26289 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26290 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26291 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26292 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26293 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26294 VMULOSW): New enum "unspec" values.
26295 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26296 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26297 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26298 altivec_vmulosw): New patterns.
26299 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26300 VMULOSW): Add definitions.
26302 2017-06-20 Julia Koval <julia.koval@intel.com>
26304 * config/i386/i386.c: Fix rounding expand for new pattern.
26305 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26307 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26309 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26310 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26312 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26314 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26317 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26319 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26320 architecture, then by implementer ID.
26321 * config/aarch64/aarch64-tune.md: Regenerate.
26323 2017-06-20 Richard Biener <rguenther@suse.de>
26325 PR middle-end/81097
26326 * fold-const.c (split_tree): Fold to type before negating.
26328 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26330 * diagnostic-show-locus.c
26331 (selftest::test_fixit_deletion_affecting_newline): New function.
26332 (selftest::diagnostic_show_locus_c_tests): Call it.
26334 2017-06-20 Andreas Schwab <schwab@suse.de>
26337 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26340 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26342 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26343 __ARM_FEATURE_COPROC according to support.
26345 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26347 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26348 Rewritten to avoid overflow for > 32-bit pointers.
26351 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26352 by removing enum keyword.
26353 (ubsan_type_descriptor): Likewise. Formatting fix.
26356 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26357 splitter): Require TARGET_SSE2 in the condition.
26359 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26362 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26363 for doing vector set of SFmode on ISA 3.0.
26364 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26365 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26367 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26368 SFmode value into a V4SF variable that was extracted from another
26369 V4SF variable without converting the element to double precision
26370 and back to single precision vector format.
26371 (vsx_insert_extract_v4sf_p9_2): Likewise.
26373 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26375 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26376 in UWHI to avoid undefined overflow.
26379 * ubsan.h (enum ubsan_encode_value_phase): New.
26380 (ubsan_encode_value): Change second argument to
26381 enum ubsan_encode_value_phase with default value of
26382 UBSAN_ENCODE_VALUE_GENERIC.
26383 * ubsan.c (ubsan_encode_value): Change second argument to
26384 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26385 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26386 create_tmp_var_raw instead of create_tmp_var and use a
26388 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26389 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26390 ubsan_encode_value callers.
26393 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26394 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26395 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26397 2017-06-19 Richard Biener <rguenther@suse.de>
26399 PR middle-end/81118
26400 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26401 estimates if we changed anything.
26403 2017-06-19 Richard Biener <rguenther@suse.de>
26405 PR tree-optimization/80887
26406 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26407 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26408 simplified lookups, then reset mprts_hook.
26409 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26411 (try_to_simplify): Likewise.
26413 2017-06-19 Martin Liska <mliska@suse.cz>
26416 * gimplify.c (gimplify_switch_expr):
26417 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26419 2017-06-19 Martin Liska <mliska@suse.cz>
26421 * doc/install.texi: Document that PGO runs in 4 stages.
26423 2017-06-19 Martin Liska <mliska@suse.cz>
26426 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26427 to dispatcher function name.
26428 * multiple_target.c (replace_function_decl): New function.
26429 (create_dispatcher_calls): Redirect both edges and references.
26431 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26433 * profile-count.c (profile_count::dump): Dump quality.
26434 (profile_count::differs_from_p): Update for unsigned val.
26435 * profile-count.h (profile_count_quality): New enum.
26436 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26438 2017-06-19 Richard Biener <rguenther@suse.de>
26440 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26441 struct function as arg.
26442 (estimate_numbers_of_iterations): Export overload with loop arg.
26443 (free_numbers_of_iterations_estimates_loop): Use an overload of
26444 free_numbers_of_iterations_estimates instead.
26445 * tree-cfg.c (remove_bb): Adjust.
26446 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26447 * tree-parloops.c (gen_parallel_loop): Likewise.
26448 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26450 (tree_unroll_loops_completely): Likewise.
26451 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26452 Use an overload instead and export.
26453 (estimated_loop_iterations): Adjust.
26454 (max_loop_iterations): Likewise.
26455 (likely_max_loop_iterations): Likewise.
26456 (estimate_numbers_of_iterations): Take struct function as arg
26458 (loop_exits_before_overflow): Adjust.
26459 (free_numbers_of_iterations_estimates_loop): Use an overload.
26460 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26461 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26463 2017-06-19 Richard Biener <rguenther@suse.de>
26466 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26467 RANGE_EXPR conservatively.
26469 2017-06-16 Carl Love <cel@us.ibm.com>
26471 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26472 definitions for vec_float, vec_float2, vec_floato,
26473 vec_floate built-ins.
26474 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26475 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26477 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26478 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26479 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26480 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26481 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26482 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26483 vec_floato): Add builtin defines.
26484 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26485 Update the built-in documentation file for the new built-in
26488 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26490 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26491 (mthumb): Mark as the negative of -marm.
26493 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26495 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26497 (ARM Options, -mtune): Document that this accepts the same
26498 extension options as -mcpu.
26499 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26501 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26503 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26504 permitted extensions.
26506 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26508 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26509 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26510 (armv8-m.main): Add option +nodsp.
26511 * config/arm/arm-cpu-cdata.h: Regenerated.
26513 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26515 * config/arm/t-fuchsia: New file.
26516 * config.gcc (arm*-*-fuchsia*): Use it.
26518 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26520 * config/arm/t-symbian: Rewrite for new option infrastructure.
26522 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26524 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26525 (MULTILIB_REQUIRED): Likewise.
26527 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26529 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26530 (MULTILIB_RESUE): Likewise.
26531 (MULTILIB_MATCHES): Likewise.
26532 (MULTLIB_REQUIRED): Likewise.
26534 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26536 * config/arm/t-rtems: Rewrite for new option framework.
26538 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26540 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26541 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26542 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26543 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26544 * config/arm/t-multilib: ... here.
26545 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26546 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26547 armv7-a and armv8*-a when A-profile libraries have not been built.
26548 * config/arm/t-rmprofile: Rewrite.
26550 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26552 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26553 with a backslash. Remove the backslash after substituting unescaped
26555 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26557 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26559 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26560 the driver to pass through the default setting of -mfloat-abi.
26561 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26563 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26564 (all_feat_combs): New rule.
26565 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26567 * config/arm/t-aprofile: Rewrite.
26569 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26571 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26572 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26573 fpu is not specified by the user/command-line.
26574 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26575 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26576 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26577 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26578 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26579 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26581 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26583 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26584 * config/arm/t-arm-elf: Rewritten.
26586 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26588 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26589 have some floating-point instructions.
26590 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26591 (TARGET_MAYBE_HARD_FLOAT): New macro.
26592 * config/arm/arm-builtins.c (arm_init_builtins): Use
26593 TARGET_MAYBE_HARD_FLOAT.
26594 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26596 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26598 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26599 (configargs.h): Include it.
26600 (arm_print_hint_for_fpu_option): New function.
26601 (arm_parse_fpu_option): New function.
26602 (candidate_extension): New class.
26603 (arm_canon_for_multilib): New function.
26604 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26605 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26606 (ARCH_CANONICAL_SPECS): New macro.
26607 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26609 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26611 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26612 are set after handling multilib fragments. Set target_cpu_default2
26615 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26617 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26619 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26620 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26622 (check_arch): Likewise.
26623 * config/arm/arm.c (arm_configure_build_target): Handle
26624 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26625 options in the default.
26627 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26629 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26630 when an option is an alias of another.
26631 * config/arm/parsecpu.awk (optalias): New parser token.
26632 (gen_comm_data): Mark non-alias options as such. Emit entries
26633 for extension aliases.
26634 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26635 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26636 (armv6kz, armv6zk, armv6t2): Likewise.
26637 (armv7): Make vfpv3-d16 an alias.
26638 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26640 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26641 Sort in canonical order.
26642 (armv8-a): Sort in canonical order.
26643 (armv8.1-a, armv8.2-a): Likewise.
26644 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26646 (cortex-a9): Sort in canonical order.
26647 * config/arm/arm.c (selftests.h): Include it.
26648 (arm_test_cpu_arch_data): New function.
26649 (arm_run_self_tests): New function.
26650 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26651 (targetm): Move declaration to the end of the file.
26652 * arm-cpu-cdata.h: Regenerated.
26654 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26656 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26657 call to target_mode_check describing the type of option passed.
26658 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26659 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26660 arm_parse_cpu_option_name to match parameters against list of
26662 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26663 arm_arch_core_flags data structure.
26664 * config/arm/arm-cpu_cdata.h: Regenerated.
26666 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26668 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26670 (arm_print_hint_for_cpu_option): Likewise.
26671 (arm_print_hint_for_arch_option): Likewise.
26672 (arm_parse_cpu_option_name): Likewise.
26673 (arm_parse_arch_option_name): Likewise.
26674 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26675 of entries in the all_fpus list.
26676 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26677 (arm_parse_cpu_option_name): Declare.
26678 (arm_parse_arch_option_name): Declare.
26679 (arm_parse_option_features): Declare.
26680 (arm_intialize_isa): Declare.
26681 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26683 (gen_comm_data): ... here. Make definitions non-static.
26684 * config/arm/arm-cpu-data.h: Regenerated.
26685 * config/arm/arm-cpu-cdata.h: Regenerated.
26687 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26689 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26690 (cpu_arch_extension): New structure.
26691 (cpu_arch_option, arch_option, cpu_option): New structures.
26692 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26693 architecture types.
26694 (gen_data): Generate new format data tables.
26695 * config/arm/arm.c (cpu_tune): New structure.
26696 (cpu_option, processors): Delete.
26697 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26698 (arm_print_hint_for_cpu_option): ... this and ...
26699 (arm_print_hint_for_arch_option): ... this.
26700 (arm_parse_arch_cpu_name): Delete. Replace with ...
26701 (arm_parse_cpu_option_name): ... this and ...
26702 (arm_parse_arch_option_name): ... this.
26703 (arm_unrecognized_feature): Change type of target parameter to
26705 (arm_parse_arch_cpu_features): Delete. Replace with ...
26706 (arm_parse_option_features): ... this.
26707 (arm_configure_build_target): Rework to use new configuration data
26709 (arm_print_tune_info): Rework for new configuration data tables.
26710 * config/arm/arm-cpu-data.h: Regenerated.
26711 * config/arm/arm-cpu.h: Regenerated.
26713 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26715 * Makefile.in (OBJS): Move sbitmap.o from here ...
26716 (OBJS-libcommon): ... to here.
26718 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26720 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26721 (ISA_ALL_CRYPTO): New macro.
26722 (ISA_ALL_SIMD): New macro
26723 (ISA_ALL_FP): New macro.
26724 * config/arm/arm.c (fpu_bitlist): Update initializer.
26725 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26727 (arm9e): Add fpu. Add option for nofp
26728 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26729 (arm926ej-s, arm1026ej-s): Likewise.
26730 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26731 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26732 neon-fp16, neon-vfpv4, nofp and nosimd.
26733 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26734 (cortex-a8): Add fpu. Add option for nofp.
26735 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26736 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26737 (cortex-r4f): Add fpu.
26738 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26739 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26741 (cortex-r8): Likewise.
26742 (cortex-m4): Add fpu. Add option for nofp.
26743 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26744 (cortex-a17.cortex-a7): Likewise.
26745 (cortex-a32): Add fpu. Add options for crypto and nofp.
26746 (cortex-a35, cortex-a53): Likewise.
26747 (cortex-a57): Add fpu. Add option for crypto.
26748 (cortex-a72, cortex-a73): Likewise.
26749 (exynos-m1): Likewise.
26750 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26751 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26752 (cortex-m33): Add fpu. Add option for nofp.
26753 * config/arm/arm-cpu-cdata.h: Regenerated
26754 * config/arm/arm-cpu-data.h: Regenerated.
26756 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26758 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26759 (armv5te, armv5tej): Likewise.
26760 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26761 (armv7): Add options fp and vfpv3-d16.
26762 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26763 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26765 (armv7ve): Likewise.
26766 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26767 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26768 (armv8-a): Add nocrypto option.
26769 (armv8.1-a, armv8.2-a): Likewise.
26770 (armv8-m.main): add options fp, fp.dp and nofp.
26772 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26774 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26776 (armv8-a+crc): Delete.
26777 (armv8.1-a): Add options simd, crypto and nofp.
26778 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26779 (armv8.2-a+fp16): Delete.
26780 (armv8-m.main): Add option dsp.
26781 (armv8-m.main+dsp): Delete.
26782 (cortex-a8): Add fpu. Add nofp option.
26783 (cortex-a9): Add fpu. Add nofp and nosimd options.
26784 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26785 link to main cpu and architecture data structures.
26786 (gen_comm_data): Only put isa attributes from the main architecture
26788 (option): New statement for architecture and CPU entries.
26789 * arm.c (struct cpu_option): New structure.
26790 (struct processors): Add entry for options.
26791 (arm_unrecognized_feature): New function.
26792 (arm_parse_arch_cpu_name): Ignore any characters after the first
26794 (arm_parse_arch_cpu_feature): New function.
26795 (arm_configure_build_target): Separate out any CPU and architecture
26796 features and parse separately. Don't error out if -mfpu=auto is
26797 used with only an architecture string.
26798 (arm_print_asm_arch_directives): New function.
26799 (arm_file_start): Call it.
26800 * config/arm/arm-cpu-cdata.h: Regenerated.
26801 * config/arm/arm-cpu-data.h: Likewise.
26802 * config/arm/arm-tables.opt: Likewise.
26804 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26806 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26807 assembler when it is not -mfpu=auto.
26809 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26811 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26812 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26813 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26814 (ASM_CPU_SPEC): Rewrite.
26815 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26816 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26817 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26818 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26819 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26820 copied string is NUL-terminated. Also strip any characters prefixed
26822 (arm_rewrite_selected_arch): New function.
26823 (arm_rewrite_march): New function.
26825 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26827 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26828 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26829 (march, mcpu, mtune): Convert to string-based options.
26830 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26831 (arm_parse_arch_cpu_name): New function.
26832 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26833 identify selected architecture or CPU.
26834 (arm_option_save): New function.
26835 (TARGET_OPTION_SAVE): Redefine.
26836 (arm_option_restore): Restore string options.
26837 (arm_option_print): Print string options.
26839 2017-06-16 Martin Sebor <msebor@redhat.com>
26841 PR tree-optimization/80933
26842 PR tree-optimization/80934
26843 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26844 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26845 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26846 (gimple_fold_builtin): Call them.
26848 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26850 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
26851 as unlikely; update profile.
26853 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26855 * predict.c (force_edge_cold): Handle declaring edges impossible
26858 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26860 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
26862 (try_unroll_loop_completely): Fix reporting.
26864 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26866 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
26868 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
26871 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
26872 if given a non-constant argument for an intrinsic which requires a
26875 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26877 * profile.c (compare_freqs): New function.
26878 (branch_prob): Sort edge list.
26879 (find_spanning_tree): Assume that the list is priority sorted.
26881 2017-06-16 Richard Biener <rguenther@suse.de>
26883 PR tree-optimization/81090
26884 * passes.def (pass_record_bounds): Remove.
26885 * tree-pass.h (make_pass_record_bounds): Likewise.
26886 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
26887 make_pass_record_bounds): Likewise.
26888 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
26889 not free niter estimates at the beginning but at the end.
26890 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
26892 2017-06-16 Richard Biener <rguenther@suse.de>
26894 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
26895 initializer to workaround ICE in host GCC 4.8.
26897 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26899 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
26901 (clone_inlined_nodes): Update.
26903 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26905 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
26906 prefetch settings, and enable prefetching by default at -O3.
26908 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26910 * config/aarch64/aarch64.c (aarch64_override_options_internal):
26911 Set flag_prefetch_loop_arrays according to tuning data.
26913 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26915 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
26916 New tune structure.
26917 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
26918 [Unrelated to main purpose of the patch] Place the pointer field last
26919 to enable type checking errors when tune structure are wrongly merged.
26920 * config/aarch64/aarch64.c (generic_prefetch_tune,)
26921 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
26922 (thunderx2t99_prefetch_tune): New tune constants.
26923 (tune_params *_tunings): Update all tunings (no functional change).
26924 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
26925 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
26926 from tunings structures.
26928 2017-06-16 Jakub Jelinek <jakub@redhat.com>
26931 * ubsan.c (instrument_null): Add T argument, use it instead
26932 of computing it based on IS_LHS.
26933 (instrument_object_size): Likewise.
26934 (pass_ubsan::execute): Adjust instrument_null and
26935 instrument_object_size callers to pass gimple_get_lhs or
26936 gimple_assign_rhs1 result to it. Use instrument_null instead of
26937 calling get_base_address and instrument_mem_ref. Handle
26938 aggregate call arguments for object-size sanitization.
26940 2017-06-16 Yury Gribov <tetra2005@gmail.com>
26942 PR tree-optimization/81089
26943 * tree-vrp.c (is_masked_range_test): Validate operands of
26946 2017-06-15 Martin Sebor <msebor@redhat.com>
26949 * dumpfile.c (dump_register): Avoid calling memset to initialize
26950 a class with a default ctor.
26951 * gcc.c (struct compiler): Remove const qualification.
26952 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
26953 * hash-table.h: Ditto.
26954 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
26956 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
26957 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
26959 * params.h (struct param_info): Make struct members non-const.
26960 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
26961 with default initialization.
26962 * vec.h (vec_copy_construct, vec_default_construct): New helper
26964 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
26965 with vec_copy_construct.
26966 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
26967 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
26968 * doc/invoke.texi (-Wclass-memaccess): Document.
26970 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26972 * emit-rtl.h (is_leaf): Update comment about local
26973 register allocator.
26975 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26978 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
26979 for a variable to have a section before checking if the section has a
26981 Set section to.persistent if persistent attribute is set.
26982 Warn if .persistent attribute is used on an automatic variable.
26984 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
26986 PR rtl-optimization/80474
26987 * reorg.c (update_block): Do not ignore instructions in a delay slot.
26989 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
26991 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
26994 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
26996 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
26997 (casesi): Emit bounds checking as RTL.
26998 (casesi_internal_mips16_<mode>): Remove bounds checking.
27000 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27002 * config/xtensa/xtensa.c (xtensa_option_override): Append
27003 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27004 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27005 xtensa_doloop_hooks): Define unconditionally.
27006 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27008 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27009 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27010 for it in xtensa_option_override.
27011 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27012 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27014 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27016 * doc/cppopts.texi: Document '-' special value to -MF.
27018 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27020 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27021 (cortex_a53_fconst): Likewise.
27022 (cortex_a53_fpmul): Likewise.
27023 (cortex_a53_f_load_64): Likewise.
27024 (cortex_a53_f_load_many): Likewise.
27025 (cortex_a53_advsimd_alu): Likewise.
27026 (cortex_a53_advsimd_alu_q): Likewise.
27027 (cortex_a53_advsimd_mul): Likewise.
27028 (cortex_a53_advsimd_mul_q): Likewise.
27029 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27030 Add missing fmul, r2f_cvt and fconst cases.
27032 2017-06-14 Richard Biener <rguenther@suse.de>
27034 PR middle-end/81088
27035 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27037 (fold_binary_loc): When associating do not treat pre-existing
27038 TREE_OVERFLOW on literal constants as a reason to allow
27039 TREE_OVERFLOW on associated literal constants.
27041 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27043 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27044 (MASK_FEATURES): New macro.
27045 * config/sparc/sparc.c (sparc_option_override): Remove the special
27046 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27048 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27050 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27051 a division of 0 if non-call exceptions are enabled.
27053 2017-06-14 Andrew Pinski <apinski@cavium.com>
27054 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27057 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27058 Improve vector initialization code gen for only variable case.
27060 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27062 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27064 2017-06-14 Richard Biener <rguenther@suse.de>
27066 PR tree-optimization/81083
27067 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27070 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27072 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27073 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27074 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27075 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27076 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27077 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27079 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27081 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27082 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27084 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27086 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27088 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27090 * config/rs6000/t-rtems: Don't handle SPE.
27092 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27094 * config/rs6000/t-linux: Don't handle SPE.
27096 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27098 * config/rs6000/eabispe.h: Delete file.
27100 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27102 * config/rs6000/t-spe: Delete file.
27104 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27106 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27107 (rs6000_legitimate_offset_address_p): Return false for anything in
27108 V2SImode or V2SFmode.
27110 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27112 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27113 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27114 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27116 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27117 (rs6000_legitimate_offset_address_p): Ditto.
27118 (rs6000_emit_move): Ditto.
27119 (rs6000_init_builtins): Remove V4HI_type_node.
27121 2017-06-13 Martin Liska <mliska@suse.cz>
27124 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27125 (gate_asan): Likewise.
27126 * asan.h (asan_no_sanitize_address_p): Remove the function.
27127 (sanitize_flags_p): New function.
27128 * builtins.def: Fix coding style.
27129 * common.opt: Use renamed enum value.
27130 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27131 * doc/extend.texi: Document no_sanitize attribute.
27132 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27133 to SANITIZE_UNDEFINED_NONDEFAULT.
27134 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27135 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27136 Use sanitize_flags_p.
27137 * gimplify.c (gimplify_function_tree): Likewise.
27138 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27139 * opts.c (parse_no_sanitize_attribute): New function.
27140 (common_handle_option): Use renamed enum value.
27141 * opts.h (parse_no_sanitize_attribute): Declare.
27142 * tree.c (sanitize_flags_p): New function.
27143 * tree.h: Declared here.
27144 * tsan.c: Use sanitize_flags_p.
27145 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27146 (instrument_mem_ref): Likewise.
27147 (instrument_bool_enum_load): Likewise.
27148 (do_ubsan_in_current_function): Remove the function.
27149 (pass_ubsan::execute): Use sanitize_flags_p.
27150 * ubsan.h: Remove do_ubsan_in_current_function
27151 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27152 (dump_function_to_file): Use it here.
27154 2017-06-13 Martin Jambor <mjambor@suse.cz>
27156 PR tree-optimization/80803
27157 PR tree-optimization/81063
27158 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27159 (propagate_subaccesses_across_link): Enqueue subtree whenever
27160 necessary instead of relying on the caller.
27162 2017-06-13 Martin Jambor <mjambor@suse.cz>
27164 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27165 that have a first_link.
27166 (sort_and_splice_var_accesses): Do not check first_link before
27168 (subtree_mark_written_and_enqueue): Likewise.
27169 (propagate_all_subaccesses): Likewise and do not stop at first
27170 parent with a first_link.
27172 2017-06-13 Martin Jambor <mjambor@suse.cz>
27174 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27177 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27179 * match.pd: New pattern.
27181 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27183 * tree-vrp.c (is_masked_range_test): New function.
27184 (register_edge_assert_for): Determine ranges for
27187 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27189 PR tree-optimization/67328
27190 * fold-const.c (maskable_range_p): New function.
27191 (build_range_check): Generate bittests if possible.
27193 2017-06-13 Martin Liska <mliska@suse.cz>
27195 * gimple-pretty-print.c (dump_probability): Add new argument.
27196 (dump_edge_probability): Dump both probability and count.
27197 (dump_gimple_label): Likewise.
27198 (dump_gimple_bb_header): Likewise.
27200 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27203 * config/avr/avr-devices.c: Fix indentation.
27204 * config/avr/gen-avr-mmcu-specs.c: Dito.
27206 2017-06-13 Richard Biener <rguenther@suse.de>
27208 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27209 instead get vector type from stmt_info.
27210 (vectorizable_reduction): Adjust. Remove dead code.
27212 2017-06-13 Richard Biener <rguenther@suse.de>
27214 PR middle-end/81065
27215 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27216 case of C * (x * C2 + C3).
27217 (fold_addr_of_array_ref_difference): Properly fold index difference.
27219 2017-06-12 David S. Miller <davem@davemloft.net>
27222 * config/sparc/sparc.md (return expander): Emit frame blockage if
27223 function uses alloca.
27225 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27227 * combine.c (make_field_assignment): Check len rather than the mode
27228 precision when calling force_to_mode.
27230 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27232 Support multilibs and devices that see flash in RAM address range.
27235 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27236 (avr_mcu_t) <flash_pm_offset>: New field.
27237 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27238 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27239 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27240 (AVR_TINY_PM_OFFSET): Remove macro.
27241 * config/avr/avr.opt (-mshort-calls): New option.
27242 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27243 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27244 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27245 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27246 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27247 instead of avr_arch->have_jmp_call.
27248 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27249 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27250 avr_arch->flash_pm_offset to define.
27251 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27252 new field flash_pm_offset. Add entry for avrxmega3.
27253 (avr_texinfo): Add entry for avrxmega3.
27254 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27255 attiny212, attiny214,
27256 attiny412, attiny414, attiny416, attiny417,
27257 attiny814, attiny816, attiny817,
27258 attiny1614, attiny1616, attiny1617,
27259 attiny3214, attiny3216, attiny3217.
27260 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27261 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27262 (avr_print_operand_address) [AVR_TINY]: Same.
27263 (avr_asm_init_sections) <readonly_data_section>: Only patch
27264 callback if avr_arch->flash_pm_offset = 0.
27265 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27266 for rodata if avr_arch->flash_pm_offset != 0.
27267 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27268 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27269 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27270 (m_options): Append opt_rcall.
27271 (m_dirnames): Append dir_rcall.
27272 * config/avr/t-multilib: Regenerate.
27274 * configure.ac [target=avr]: Check whether avrxmega3 default
27275 linker description file works as needed.
27276 * configure: Regenerate.
27277 * doc/avr-mmcu.texi: Regenerate.
27278 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27279 <__AVR_ARCH__>: Document avrxmega3 and 103.
27280 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27281 <__AVR_SHORT_CALLS__>: Document it.
27282 <__AVR_PM_BASE_ADDRESS__>: Document it.
27283 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27284 (AVR Variable Attributes) <progmem>: Document this is
27285 not needed for avrxmega3.
27286 (AVR Named Address Spaces) <__flash>: Dito.
27288 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27290 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27292 2017-06-12 Doug Rupp <rupp@adacore.com>
27294 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27295 Append vxworks-stdint.h to the tm_file list.
27296 * config/vxworks-stdint.h: New file.
27298 2017-06-12 Martin Liska <mliska@suse.cz>
27300 PR tree-optimization/81041
27301 * tree-profile.c (gimple_gen_ic_func_profiler):
27302 Create an extra BB in profile-generate
27303 (gimple_gen_time_profiler): Likewise.
27305 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27307 PR tree-optimization/81003
27308 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27309 (update_range_test): Use it instead of force_gimple_operand_gsi.
27311 2017-06-12 Richard Biener <rguenther@suse.de>
27313 PR tree-optimization/81053
27314 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27315 with backedge value not defined in loop. Simplify def stmt
27318 2017-06-11 Tom de Vries <tom@codesourcery.com>
27321 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27323 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27324 nvptx_cannot_force_const_mem.
27326 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27328 * opts.c (finish_options): Move test for flag_split_stack after
27329 it has been initialized.
27331 2017-06-11 Jason Merrill <jason@redhat.com>
27333 * tree.h (id_equal): New.
27334 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27335 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27336 instead of strcmp of IDENTIFIER_POINTER.
27338 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27340 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27341 (mark_all_inlined_calls_cdtor): Fix formating.
27342 (inline_transform): Rescale profile before inlining.
27344 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27346 * cgraph.h (cgraph_edge::clone): Update prototype.
27347 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27348 (cgraph_node::create_clone): Update.
27349 (cgraph_node::create_version_clone): Update.
27350 * tree-inline.c (copy_bb): Update.
27351 (expand_call_inline): Update.
27353 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27355 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27356 factored out from ...
27357 (rs6000_emit_prologue): ... here.
27359 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27361 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27362 factored out from ...
27363 (rs6000_emit_prologue): ... here.
27365 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27367 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27369 (handle_missing_profiles): Fix computation of tp_first_run.
27370 (counts_to_freqs): Do not touch freqs when count is 0.
27372 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27374 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27377 2017-06-10 Tom de Vries <tom@codesourcery.com>
27379 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27380 attributes): Document signal effective target.
27382 2017-06-10 Tom de Vries <tom@codesourcery.com>
27384 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27385 Document effective target stack_size.
27387 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27389 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27390 to the edit_context if they can be auto-applied.
27392 2017-06-9 Ian Lance Taylor <iant@golang.org>
27394 * opts.c (finish_options): If -fsplit-stack, disable implicit
27395 -forder-blocks-and-partition.
27396 * doc/invoke.texi (Optimize Options): Document that when using
27397 -fsplit-stack -forder-blocks-and-partition is not implicitly
27400 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27402 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27403 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27404 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27405 * builtins.def (abort, trap, unreachable): Declare cold.
27406 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27407 * tree-core.h (ECF_COLD): New.
27408 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27409 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27411 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27413 * predict.c (unlikely_executed_stmt_p): Cleanup.
27415 2017-06-09 Richard Biener <rguenther@suse.de>
27417 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27418 model if the ref is always written to.
27420 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27422 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27424 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27426 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27429 2017-06-09 Tom de Vries <tom@codesourcery.com>
27432 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27433 "target cannot support label values" when encountering LABEL_REF.
27435 2017-06-09 Martin Liska <mliska@suse.cz>
27437 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27438 (gimple_gen_ic_func_profiler): Emit direct comparison
27439 of __gcov_indirect_call_callee with NULL.
27440 (gimple_gen_time_profiler): Change probability from
27441 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27443 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27445 * profile.c (edge_gcov_counts): Turn to pointer.
27446 (compute_branch_probabilities, compute_branch_probabilities): Update.
27447 (branch_prob): Do not clear edge_gcov_count.
27448 * profile.h (edge_gcov_counts): Turn to pointer.
27449 (edge_gcov_count): Update.
27451 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27453 * gimple.h (gimple_check_failed): Mark cold.
27455 2017-06-09 Richard Biener <rguenther@suse.de>
27457 PR tree-optimization/66623
27458 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27459 refactor check_reduction into two parts, properly computing
27460 whether we have to check reduction validity for outer loop
27463 2017-06-09 Richard Biener <rguenther@suse.de>
27465 PR tree-optimization/79483
27466 * graphite-scop-detection.c (order): New global.
27467 (get_order): Compute bb to order mapping that satisfies code
27468 generation constraints.
27469 (cmp_pbbs): New helper.
27470 (build_scops): Start domwalk at entry block, sort generated
27473 2017-06-09 Richard Biener <rguenther@suse.de>
27475 PR middle-end/81007
27476 * ipa-polymorphic-call.c
27477 (ipa_polymorphic_call_context::restrict_to_inner_class):
27478 Skip FIELD_DECLs with error_mark_node type.
27479 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27482 2017-06-09 Martin Liska <mliska@suse.cz>
27484 * predict.c (struct branch_predictor): New struct.
27485 (test_prediction_value_range): New test.
27486 (predict_c_tests): New function.
27487 * selftest-run-tests.c (selftest::run_tests): Run the function.
27488 * selftest.h: Declare new tests.
27490 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27493 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27494 gen_add3_insn did not fail.
27495 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27496 r0, construct that number in a temporary reg and add that reg to r0.
27497 If asked to put the result in r0 as well, fail.
27499 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27501 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27502 for early expansion of vec_eqv.
27504 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27506 PR middle-end/81005
27507 * ubsan.c (instrument_null): Avoid pointless code temporary.
27508 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27510 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27514 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27517 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27520 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27522 * predict.c (unlikely_executed_edge_p): Move ahead.
27523 (probably_never_executed_edge_p): Use it.
27525 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27527 PR middle-end/79988
27528 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27529 gimple_call_builtin_p call.
27531 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27533 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27534 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27535 rtl_check_failed_type2, rtl_check_failed_code1,
27536 rtl_check_failed_code2, rtl_check_failed_code_mode,
27537 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27538 rtvec_check_failed_bounds, rtl_check_failed_flag,
27539 _fatal_insn_not_found, _fatal_insn): Likewise.
27540 * tree.h (tree_contains_struct_check_failed,
27541 tree_check_failed, tree_not_check_failed,
27542 tree_class_check_failed, tree_range_check_failed,
27543 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27544 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27545 tree_operand_check_failed, omp_clause_check_failed,
27546 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27549 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27551 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27552 flag_branch_probabilities.
27553 * ipa-inline.c (edge_badness): Likewise.
27554 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27555 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27556 * predict.c (maybe_hot_frequency_p): Likewise.
27557 (probably_never_executed): Likewise.
27558 * sched-ebb.c (schedule_ebbs): Likewise.
27559 * sched-rgn.c (find_single_block_region): Likewise.
27560 * tracer.c (tail_duplicate): Likewise.
27562 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27564 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27565 longer requires x_flag_profile_use.
27567 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27569 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27570 instead of flag_reorder_blocks_and_partition.
27571 * dbxout.c (dbxout_function_end): Likewise.
27572 * dwarf2out.c (gen_subprogram_die): Likewise.
27573 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27574 * hw-doloop.c (reorg_loops): Likewise.
27575 * varasm.c (assemble_start_function,
27576 assemble_end_function): Likewise.
27577 (decide_function_section): Do not check for
27578 flag_reorder_blocks_and_partition.
27580 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27582 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27584 (chkp_get_hard_register_fake_addr_expr): Ditto.
27585 (chkp_build_addr_expr): Add check for hard reg case.
27586 (chkp_parse_array_and_component_ref): Ditto.
27587 (chkp_find_bounds_1): Ditto.
27588 (chkp_process_stmt): Don't generate bounds store for
27591 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27593 * predict.c (maybe_hot_bb_p): Do not check profile status.
27594 (maybe_hot_edge_p): Likewise.
27595 (probably_never_executed): Check for zero counts even if profile
27597 (unlikely_executed_edge_p): New function.
27598 (unlikely_executed_stmt_p): New function.
27599 (unlikely_executed_bb_p): New function.
27600 (set_even_probabilities): Use unlikely predicates.
27601 (combine_predictions_for_bb): Likewise.
27602 (predict_paths_for_bb): Likewise.
27603 (predict_paths_leading_to_edge): Likewise.
27604 (determine_unlikely_bbs): New function.
27605 (estimate_bb_frequencies): Use it.
27606 (compute_function_frequency): Use zero counts even if profile is
27608 * profile-count.h: Fix typo.
27610 2017-08-08 Julia Koval <julia.koval@intel.com>
27612 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27613 _mm512_mask_cvtsepi16_storeu_epi8,
27614 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27615 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27616 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27617 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27618 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27619 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27620 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27621 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27622 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27623 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27624 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27625 __builtin_ia32_pmovuswb256mem_mask,
27626 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27627 __builtin_ia32_pmovwb512mem_mask): New builtins.
27629 2017-08-08 Julia Koval <julia.koval@intel.com>
27631 PR target/73350,80862
27632 * config/i386/subst.md (round): Fix round pattern.
27633 * config/i386/i386.c (ix86_erase_embedded_rounding):
27634 Fix erasing rounding for the fixed pattern.
27636 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27638 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27640 2017-06-08 Martin Liska <mliska@suse.cz>
27642 PR gcov-profile/80911
27643 * gcov.c (block_info::block_info): New constructor.
27645 2017-06-07 Carl Love <cel@us.ibm.com>
27647 * config/rs6000/rs6000-c: The return type of the following
27648 built-in functions was implemented as int not long long. Fix sign
27649 of return value for the unsigned version of vec_mulo and vec_mule.
27650 vector unsigned long long vec_bperm (vector unsigned long long,
27651 vector unsigned char)
27652 vector signed long long vec_mule (vector signed int,
27654 vector unsigned long long vec_mule (vector unsigned int,
27655 vector unsigned int)
27656 vector signed long long vec_mulo (vector signed int,
27658 vector unsigned long long vec_mulo (vector unsigned int,
27659 vector unsigned int)
27660 * doc/extend.texi: Fix the documentation for the built-in
27663 2017-06-07 Carl Love <cel@us.ibm.com>
27666 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27669 2017-06-07 Carl Love <cel@us.ibm.com>
27671 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27672 support, Generate doublehv for signed int/float for BE case only.
27674 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27676 * doc/invoke.texi (mcx16): Rewrite.
27678 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27680 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27681 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27682 *mov<mode>_softfloat, and an anonymous splitter): Use
27683 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27685 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27687 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27689 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27690 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27691 (rs6000_debug_reg_global): Adjust.
27692 (rs6000_init_hard_regno_mode_ok): Adjust.
27693 (rs6000_dbx_register_number): Adjust.
27694 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27695 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27696 Remove SPE_ACC and SPEFSCR.
27697 (REG_ALLOC_ORDER): Ditto.
27698 (FRAME_POINTER_REGNUM): Change to 111.
27699 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27700 (REG_CLASS_NAMES): Ditto.
27701 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27702 (REGISTER_NAMES): Ditto.
27703 (ADDITIONAL_REG_NAMES): Ditto.
27704 (rs6000_reg_names): Ditto.
27705 * config/rs6000/rs6000.md: Renumber some register number
27708 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27710 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27712 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27713 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27715 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27716 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27717 Delete the SPE high registers.
27718 (REG_ALLOC_ORDER): Ditto.
27719 (enum reg_class): Remove SPE_HIGH_REGS.
27720 (REG_CLASS_NAMES): Ditto.
27721 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27722 (REGISTER_NAMES): Ditto.
27723 (rs6000_reg_names): Ditto.
27724 * doc/tm.texi.in: Remove SPE as example.
27725 * doc/tm.texi: Regenerate.
27727 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27729 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27730 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27731 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27732 * config/rs6000/rs6000.md (type): Remove "brinc".
27734 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27736 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27737 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27738 * config/rs6000/linuxspe.h: Delete file.
27739 * config/rs6000/rs6000.md: Don't include spe.md.
27740 * config/rs6000/spe.h: Delete file.
27741 * config/rs6000/spe.md: Delete file.
27742 * config/rs6000/t-rs6000: Remove spe.md.
27744 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27746 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27747 (reg_or_none500mem_operand): Delete.
27748 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27749 instead of reg_or_none500mem_operand.
27751 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27753 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27754 handling of SPE flags.
27755 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27757 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27759 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27761 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27762 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27763 paired_divv2sf3): Similar.
27764 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27765 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27766 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27768 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27769 Rename the paired_* instruction patterns.
27770 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27772 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27773 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27774 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27775 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27776 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27777 PAIRED_VECTOR_MODE.
27778 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27779 (spe_func_has_64bit_regs_p): Delete.
27780 (spe_expand_predicate_builtin): Delete.
27781 (spe_expand_evsel_builtin): Delete.
27782 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27783 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27784 (invalid_e500_subreg): Delete.
27785 (rs6000_legitimize_address): Always force_reg op2 as well, for
27786 paired single memory accesses.
27787 (rs6000_member_type_forces_blk): Delete.
27788 (rs6000_spe_function_arg): Delete.
27789 (rs6000_expand_unop_builtin): Delete SPE handling.
27790 (rs6000_expand_binop_builtin): Ditto.
27791 (spe_expand_stv_builtin): Delete.
27792 (bdesc_2arg_spe): Delete.
27793 (spe_expand_builtin): Delete.
27794 (spe_expand_predicate_builtin): Delete.
27795 (spe_expand_evsel_builtin): Delete.
27796 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27797 (spe_init_builtins): Delete.
27798 (spe_func_has_64bit_regs_p): Delete.
27799 (savres_routine_name): Delete "info" parameter. Adjust callers.
27800 (rs6000_emit_stack_reset): Ditto.
27801 (rs6000_dwarf_register_span): Delete.
27802 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27803 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27804 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27806 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27808 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27809 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27810 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27811 mulv2sf3, divv2sf3): Delete expanders.
27813 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27815 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27817 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27819 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27820 * config/rs6000/rs6000.c: Ditto.
27822 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27824 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27825 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27826 comparison_operator.
27828 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27830 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27831 * config/rs6000/rs6000.opt: Ditto.
27832 * config/rs6000/t-rtems: Ditto.
27834 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27836 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27837 TARGET_E500_SINGLE by 0, simplify.
27838 * config/rs6000/rs6000.c: Ditto.
27839 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27840 (spe_build_register_parallel): Delete.
27841 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27842 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27843 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27844 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27845 (E500_CONVERT): Delete.
27846 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27848 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27850 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
27851 * config/rs6000/dfp.md: Ditto.
27852 (negdd2, *negdd2_fpr): Merge.
27853 (absdd2, *absdd2_fpr): Merge.
27854 (negtd2, *negtd2_fpr): Merge.
27855 (abstd2, *abstd2_fpr): Merge.
27856 * config/rs6000/e500.h: Delete file.
27857 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
27858 TARGET_FPRS by 1 and simplify.
27859 * config/rs6000/rs6000-c.c: Ditto.
27860 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
27861 TARGET_DF_SPE by 0.
27862 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
27864 * config/rs6000/rs6000.md: Ditto.
27865 (floatdidf2, *floatdidf2_fpr): Merge.
27866 (move_from_CR_gt_bit): Delete.
27867 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
27868 (E500_CR_IOR_COMPARE): Delete.
27869 (All patterns that require !TARGET_FPRS): Delete.
27870 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
27872 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27874 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
27876 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27878 * graphds.c (add_edge): Intitialize edge's attached data.
27879 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
27880 pointer parameter. Call pointed function on each edge during
27881 graph traversing. Skip traversing the edge when the function
27883 (graphds_dfs, graphds_scc): Ditto.
27884 (for_each_edge): New parameter. Pass the new parameter to callback
27886 * graphds.h (skip_edge_callback): New function pointer type.
27887 (graphds_dfs, graphds_scc): New function pointer parameter.
27888 (graphds_edge_callback, for_each_edge): New parameter.
27890 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27892 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
27893 out code checking if runtime alias check is possible to below ...
27894 Call the new function.
27895 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
27896 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
27898 2017-06-07 Marek Polacek <polacek@redhat.com>
27901 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
27902 TYPE_OVERFLOW_WRAPS check.
27904 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27906 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
27907 if versioning is required.
27908 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
27909 peeling with the check for versioning.
27911 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27913 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
27914 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
27915 Set true to new parameter if new ssa variable is defined.
27916 (vect_gen_vector_loop_niters): Refactor. Set range information
27917 for the new vector loop bound variable.
27918 (vect_do_peeling): Ditto.
27920 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27922 * tree-affine.c (ssa.h): Include header file.
27923 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
27924 has wrapping overflow behavior.
27926 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27928 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
27930 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27932 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
27933 (tree_to_aff_combination): ... here.
27935 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27937 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
27938 reg_pressure model function.
27939 (ivopts_global_cost_for_size): Delete.
27940 (determine_set_costs, iv_ca_recount_cost): Call new model function
27941 ivopts_estimate_reg_pressure.
27943 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27945 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
27946 expensive than udiv. Remove floating point cases from mod.
27948 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27950 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
27951 Increase idiv cost.
27953 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27955 * config/aarch64/aarch64.md
27956 (copysignsf3): Fix mask generation.
27958 2017-06-07 Jakub Jelinek <jakub@redhat.com>
27960 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
27962 (class dump_manager): Add register_dumps method.
27963 * dumpfile.c: Include langhooks.h.
27964 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
27965 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
27966 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
27967 (dump_manager::dump_register): Start with 512 entries instead of 32.
27968 (dump_manager::register_dumps): New method.
27969 * toplev.c (general_init): Instead of invoking register_dumps
27970 langhook, invoke register_dumps method on the dump manager.
27971 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
27974 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
27976 * doc/md.texi: Clarify the restrictions on a define_insn condition.
27977 Say that # requires an associated define_split to exist, and that
27978 the define_split must be suitable for use after register allocation.
27980 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27982 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
27983 (compute_outgoing_frequencies): Also initialize zero counts.
27984 (find_many_sub_basic_blocks): Do not produce uninitialized profile
27985 around loops; preserve more of profile when nothing changes.
27987 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
27989 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
27991 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
27992 * config/arm/arm-cpu-cdata.h: Regenerate.
27993 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
27994 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
27995 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
27996 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
27997 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
27999 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28001 * config/arm/t-rmprofile: Likewise.
28002 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28004 2017-06-06 David S. Miller <davem@davemloft.net>
28007 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28008 blockage if function uses alloca.
28010 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28012 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28013 New "uid" fields to hold pretty-print IDs of group and ref.
28014 Memory references are now identified as <group_id>:<ref_id>
28015 instead of using [random] addresses.
28016 (dump_mem_details): Simplify, no functional change.
28017 (dump_mem_ref): Simplify and make output more concise.
28018 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28019 (find_or_create_group): Initialize group uid.
28020 (record_ref): Initialize ref uid. Improve debug output.
28021 (prune_group_by_reuse, should_issue_prefetch_p,)
28022 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28023 (mark_nontemporal_store, determine_loop_nest_reuse):
28024 Improve debug output.
28026 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28028 * dbgcnt.def (prefetch): New debug counter.
28029 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28030 (schedule_prefetches): Stop issueing prefetches if debug counter
28033 2017-06-06 Tom de Vries <tom@codesourcery.com>
28035 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28036 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28038 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28040 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28041 Use aarch64_reg_or_zero predicate for operand 4.
28042 (aarch64_compare_and_swap<mode> define_insn_and_split):
28043 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28044 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28046 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28048 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28049 (arm_compute_save_core_reg_mask): This.
28050 (thumb1_compute_save_reg_mask): Rename into ...
28051 (thumb1_compute_save_core_reg_mask): This.
28052 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28053 (arm_compute_frame_layout): Likewise.
28055 2017-06-06 Richard Biener <rguenther@suse.de>
28057 PR tree-optimization/80974
28058 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28059 keep or clear leaders SSA info.
28061 2017-06-06 Tom de Vries <tom@codesourcery.com>
28063 * config/nvptx/nvptx.c (split_mode_p): New function.
28064 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28066 2017-06-06 Tom de Vries <tom@codesourcery.com>
28068 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28070 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28073 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28076 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28078 * shrink-wrap.c (handle_simple_exit): Update profile.
28079 (try_shrink_wrapping): Upate profile.
28081 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28083 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28084 (tree_guess_outgoing_edge_probabilities): New.
28085 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28086 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28088 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28090 * ipa-split.c (split_function): Initialize return bb profile.
28092 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28094 * profile.c (compute_branch_probabilities): Also initialize
28095 EXIT_BLOCK profile.
28097 2017-06-06 Richard Biener <rguenther@suse.de>
28099 PR tree-optimization/80928
28100 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28101 (vect_analyze_loop_operations): Properly guard analysis for
28103 (vect_transform_loop): Likewise.
28104 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28105 (vect_model_induction_cost): Do not cost for pure SLP.
28106 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28107 of induction in inner loop vectorization.
28108 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28109 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28110 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28112 (vect_analyze_slp_cost_1): Cost induction.
28113 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28114 (vect_get_slp_vect_defs): Likewise.
28115 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28116 (vect_transform_stmt): Handle SLP reductions.
28117 * tree-vectorizer.h (vectorizable_induction): Adjust.
28119 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28121 * config/rs6000/rs6000.c (make_resolver_func): Update
28122 init_lowered_empty_function call.
28124 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28126 * doc/invoke.texi: Document the -fprofile-abs-path option.
28127 * common.opt (fprofile-abs-path): New option.
28128 * gcov-io.h (gcov_write_filename): Declare.
28129 * gcov-io.c (gcov_write_filename): New function.
28130 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28131 * profile.c (output_location): Likewise.
28133 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28135 * shring-wrap.c: Revert accidental commit.
28137 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28139 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28141 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28143 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28145 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28146 profile in callgraph edge.
28147 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28148 (apply_scale): Likewise.
28149 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28150 Also scale profile when inlining function with zero profile.
28151 (initialize_cfun): Update exit block profile even when it is zero.
28152 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28153 when profile is read.
28155 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28157 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28158 (CLONE_*): New constants to define the processors we can generate
28159 code for with the target_clone attribute.
28160 (rs6000_clone_map): New array to identify which clone processors
28161 the current program is running on.
28162 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28163 target_clone attribute.
28164 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28165 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28166 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28167 (cpu_expand_builtin): Add support for target_clone attribute.
28168 (rs6000_valid_attribute_p): Allow "default" attribute.
28169 (get_decl_name): New debug function to simplify printing the
28170 current function name in debugging statements.
28171 (rs6000_clone_priority): New functions to support the target_clone
28172 attribute, and be able to generate code to switch between ISA 2.05
28173 through ISA 3.0 (power6 through power9).
28174 (rs6000_compare_version_priority): Likewise.
28175 (rs6000_get_function_versions_dispatcher): Likewise.
28176 (make_resolver_func): Likewise.
28177 (add_condition_to_bb): Likewise.
28178 (dispatch_function_versions): Likewise.
28179 (rs6000_generate_version_dispatcher_body): Likewise.
28180 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28181 (fusion_gpr_load_p): Fix a spacing issue.
28182 * doc/extend.texi (Common Function Attributes): Document that the
28183 PowerPC supports the target_clone attribute.
28185 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28187 * config/arm/arm.h: explain F symbol found in description of ARM
28188 register allocation in its legend.
28190 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28192 * config/mips/frame-header-opt.c: Include profile-count.h.
28193 * config/riscv/riscv.c: Include profile-count.h
28195 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28197 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28199 (sm_set_flag_if_changed): Add bbs field.
28200 (execute_sm_if_changed_flag_set): Pass BBS.
28201 (execute_sm): Update.
28203 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28205 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28208 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28210 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28211 (peephole2): New peephole2 to emit the above.
28212 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28214 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28216 * config/aarch64/aarch64.c (define_peephole2 above
28217 *sub_<shift>_<mode>): New peephole.
28219 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28221 * config/i386/i386.c (make_resolver_func): Update.
28222 * Makefile.in: Add profile-count.h and profile-count.o
28223 * auto-profile.c (afdo_indirect_call): Update to new API.
28224 (afdo_set_bb_count): Update.
28225 (afdo_propagate_edge): Update.
28226 (afdo_propagate_circuit): Update.
28227 (afdo_calculate_branch_prob): Update.
28228 (afdo_annotate_cfg): Update.
28229 * basic-block.h: Include profile-count.h
28230 (struct edge_def): Turn count to profile_count.
28231 (struct basic_block_def): Likewie.
28232 (REG_BR_PROB_BASE): Move to profile-count.h
28233 (RDIV): Move to profile-count.h
28234 * bb-reorder.c (max_entry_count): Turn to profile_count.
28235 (find_traces): Update.
28236 (rotate_loop):Update.
28237 (connect_traces):Update.
28238 (sanitize_hot_paths):Update.
28239 * bt-load.c (migrate_btr_defs): Update.
28240 * cfg.c (RDIV): Remove.
28241 (init_flow): Use alloc_block.
28242 (alloc_block): Uninitialize count.
28243 (unchecked_make_edge): Uninitialize count.
28244 (check_bb_profile): Update.
28245 (dump_edge_info): Update.
28246 (dump_bb_info): Update.
28247 (update_bb_profile_for_threading): Update.
28248 (scale_bbs_frequencies_int): Update.
28249 (scale_bbs_frequencies_gcov_type): Update.
28250 (scale_bbs_frequencies_profile_count): New.
28251 * cfg.h (update_bb_profile_for_threading): Update.
28252 (scale_bbs_frequencies_profile_count): Declare.
28253 * cfgbuild.c (compute_outgoing_frequencies): Update.
28254 (find_many_sub_basic_blocks): Update.
28255 * cfgcleanup.c (try_forward_edges): Update.
28256 (try_crossjump_to_edge): Update.
28257 * cfgexpand.c (expand_gimple_tailcall): Update.
28258 (construct_exit_block): Update.
28259 * cfghooks.c (verify_flow_info): Update.
28260 (dump_bb_for_graph): Update.
28261 (split_edge): Update.
28262 (make_forwarder_block): Update.
28263 (duplicate_block): Update.
28264 (account_profile_record): Update.
28265 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28266 (get_estimated_loop_iterations): Update.
28267 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28268 (single_likely_exit): Update.
28269 * cfgloopmanip.c (scale_loop_profile): Update.
28271 (set_zero_probability): Update.
28272 (lv_adjust_loop_entry_edge): Update.
28273 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28274 (purge_dead_edges): Update.
28275 (rtl_account_profile_record): Update.
28276 * cgraph.c (cgraph_node::create): Uninitialize count.
28277 (symbol_table::create_edge): Uninitialize count.
28278 (cgraph_update_edges_for_call_stmt_node): Update.
28279 (cgraph_edge::dump_edge_flags): Update.
28280 (cgraph_node::dump): Update.
28281 (cgraph_edge::maybe_hot_p): Update.
28282 * cgraph.h: Include profile-count.h
28283 (create_clone), create_edge, create_indirect_edge): Update.
28284 (cgraph_node): Turn count to profile_count.
28285 (cgraph_edge0: Likewise.
28286 (make_speculative, clone): Update.
28287 (create_edge): Update.
28288 (init_lowered_empty_function): Update.
28289 * cgraphclones.c (cgraph_edge::clone): Update.
28290 (duplicate_thunk_for_node): Update.
28291 (cgraph_node::create_clone): Update.
28292 * cgraphunit.c (cgraph_node::analyze): Update.
28293 (cgraph_node::expand_thunk): Update.
28294 * final.c (dump_basic_block_info): Update.
28295 * gimple-streamer-in.c (input_bb): Update.
28296 * gimple-streamer-out.c (output_bb): Update.
28297 * graphite.c (print_global_statistics): Update.
28298 (print_graphite_scop_statistics): Update.
28299 * hsa-brig.c: Include basic-block.h.
28300 * hsa-dump.c: Include basic-block.h.
28301 * hsa-gen.c (T sum_slice): Update.
28302 (convert_switch_statements):Update.
28303 * hsa-regalloc.c: Include basic-block.h.
28304 * ipa-chkp.c (chkp_produce_thunks): Update.
28305 * ipa-cp.c (struct caller_statistics): Update.
28306 (init_caller_stats): Update.
28307 (gather_caller_stats): Update.
28308 (ipcp_cloning_candidate_p): Update.
28309 (good_cloning_opportunity_p): Update.
28310 (get_info_about_necessary_edges): Update.
28311 (dump_profile_updates): Update.
28312 (update_profiling_info): Update.
28313 (update_specialized_profile): Update.
28314 (perhaps_add_new_callers): Update.
28315 (decide_about_value): Update.
28316 (ipa_cp_c_finalize): Update.
28317 * ipa-devirt.c (struct odr_type_warn_count): Update.
28318 (struct decl_warn_count): Update.
28319 (struct final_warning_record): Update.
28320 (possible_polymorphic_call_targets): Update.
28321 (ipa_devirt): Update.
28322 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28323 * ipa-icf.c (sem_function::merge): Update.
28324 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28325 * ipa-inline.c (compute_uninlined_call_time): Update.
28326 (compute_inlined_call_time): Update.
28327 (want_inline_small_function_p): Update.
28328 (want_inline_self_recursive_call_p): Update.
28329 (edge_badness): Update.
28330 (lookup_recursive_calls): Update.
28331 (recursive_inlining): Update.
28332 (inline_small_functions): Update.
28333 (dump_overall_stats): Update.
28334 (dump_inline_stats): Update.
28335 * ipa-profile.c (ipa_profile_generate_summary): Update.
28336 (ipa_propagate_frequency): Update.
28337 (ipa_profile): Update.
28338 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28339 * ipa-utils.c (ipa_merge_profiles): Update.
28340 * loop-doloop.c (doloop_modify): Update.
28341 * loop-unroll.c (report_unroll): Update.
28342 (unroll_loop_runtime_iterations): Update.
28343 * lto-cgraph.c (lto_output_edge): Update.
28344 (lto_output_node): Update.
28345 (input_node): Update.
28346 (input_edge): Update.
28347 (merge_profile_summaries): Update.
28348 * lto-streamer-in.c (input_cfg): Update.
28349 * lto-streamer-out.c (output_cfg): Update.
28350 * mcf.c (create_fixup_graph): Update.
28351 (adjust_cfg_counts): Update.
28352 (sum_edge_counts): Update.
28353 * modulo-sched.c (sms_schedule): Update.
28354 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28355 * predict.c (maybe_hot_count_p): Update.
28356 (probably_never_executed): Update.
28357 (dump_prediction): Update.
28358 (combine_predictions_for_bb): Update.
28359 (propagate_freq): Update.
28360 (handle_missing_profiles): Update.
28361 (counts_to_freqs): Update.
28362 (rebuild_frequencies): Update.
28363 (force_edge_cold): Update.
28364 * predict.h: Include profile-count.h
28365 (maybe_hot_count_p, counts_to_freqs): UPdate.
28366 * print-rtl-function.c: Do not include cfg.h
28367 * print-rtl.c: Include basic-block.h
28368 * profile-count.c: New file.
28369 * profile-count.h: New file.
28370 * profile.c (is_edge_inconsistent): Update.
28371 (correct_negative_edge_counts): Update.
28372 (is_inconsistent): Update.
28373 (set_bb_counts): Update.
28374 (read_profile_edge_counts): Update.
28375 (compute_frequency_overlap): Update.
28376 (compute_branch_probabilities): Update; Initialize and deinitialize
28378 (branch_prob): Update.
28379 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28380 (edge_gcov_count): New.
28381 (bb_gcov_count): New.
28382 * shrink-wrap.c (try_shrink_wrapping): Update.
28383 * tracer.c (better_p): Update.
28384 * trans-mem.c (expand_transaction): Update.
28385 (ipa_tm_insert_irr_call): Update.
28386 (ipa_tm_insert_gettmclone_call): Update.
28387 * tree-call-cdce.c: Update.
28388 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28389 (gimple_duplicate_sese_tail): Update.
28390 (gimple_account_profile_record): Update.
28391 (execute_fixup_cfg): Update.
28392 * tree-inline.c (copy_bb): Update.
28393 (copy_edges_for_bb): Update.
28394 (initialize_cfun): Update.
28395 (freqs_to_counts): Update.
28396 (copy_cfg_body): Update.
28397 (expand_call_inline): Update.
28398 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28399 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28400 (try_unroll_loop_completely): Update.
28401 (try_peel_loop): Update.
28402 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28403 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28404 * tree-ssa-loop-split.c (connect_loops): Update.
28405 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28406 * tree-ssa-reassoc.c (branch_fixup): Update.
28407 * tree-ssa-tail-merge.c (replace_block_by): Update.
28408 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28409 (compute_path_counts): Update.
28410 (update_profile): Update.
28411 (recompute_probabilities): Update.
28412 (update_joiner_offpath_counts): Update.
28413 (estimated_freqs_path): Update.
28414 (freqs_to_counts_path): Update.
28415 (clear_counts_path): Update.
28416 (ssa_fix_duplicate_block_edges): Update.
28417 (duplicate_thread_path): Update.
28418 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28419 (struct switch_conv_info): Update.
28420 * tree-tailcall.c (decrease_profile): Update.
28421 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28422 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28423 * value-prof.c (check_counter): Update.
28424 (gimple_divmod_fixed_value): Update.
28425 (gimple_mod_pow2): Update.
28426 (gimple_mod_subtract): Update.
28427 (gimple_ic_transform): Update.
28428 (gimple_stringop_fixed_value): Update.
28429 * value-prof.h (gimple_ic): Update.
28431 2017-06-02 Carl Love <cel@us.ibm.com>
28433 * config/rs6000/rs6000-c: Add support for built-in functions
28434 vector double vec_doublee (vector signed int);
28435 vector double vec_doublee (vector unsigned int);
28436 vector double vec_doublee (vector float);
28437 vector double vec_doubleh (vector signed int);
28438 vector double vec_doubleh (vector unsigned int);
28439 vector double vec_doubleh (vector float);
28440 vector double vec_doublel (vector signed int);
28441 vector double vec_doublel (vector unsigned int);
28442 vector double vec_doublel (vector float);
28443 vector double vec_doubleo (vector signed int);
28444 vector double vec_doubleo (vector unsigned int);
28445 vector double vec_doubleo (vector float);.
28446 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28447 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28449 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28450 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28451 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28453 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28454 vec_doublel, vec_doubleh.
28455 * doc/extend.texi: Update the built-in documentation file for the
28456 new built-in functions.
28458 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28461 * ipa-inline-analysis.c (free_growth_caches): Set
28462 edge_removal_hook_holder to NULL after removing it.
28464 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28466 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28467 comparision with zero.
28469 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28470 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28471 for early expansion of vec_min and vec_max builtins.
28472 (builtin_function_type): Add min/max unsigned variants to those
28473 identified as having unsigned arguments.
28475 2017-06-02 Olivier Hainque <hainque@adacore.com>
28477 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28479 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28481 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28482 Use VALL_F16 iterator rather than VALL.
28484 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28486 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28487 Emit CBNZ inside loop when doing a strong exchange and comparing
28488 against zero. Generate the CC flags after the loop.
28490 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28492 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28493 (dl_section_ref): New.
28494 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28495 On AIX, append an expression to subtract the size of the
28496 section length to dl_section_ref.
28498 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28500 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28501 for early expansion of vector absolute builtins.
28503 2017-06-02 Richard Biener <rguenther@suse.de>
28505 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28506 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28508 2017-06-02 Richard Biener <rguenther@suse.de>
28510 PR tree-optimization/80948
28511 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28512 stmt order as well.
28514 2017-06-02 Richard Biener <rguenther@suse.de>
28516 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28518 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28519 for inductions as relevant.
28521 2017-06-02 Richard Biener <rguenther@suse.de>
28523 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28524 (vectorizable_induction): ... this. Remove dead code.
28526 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28528 * builtins. (expand_builtin_alloca): Remove second parameter and
28529 infer its value from the first parameter instead.
28530 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28532 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28534 PR rtl-optimization/80903
28535 * loop-doloop.c (add_test): Unshare sequence.
28537 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28539 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28541 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28543 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28545 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28546 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28547 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28548 (xlogue_layout::get_stub_rtx): Make static.
28549 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28550 (xlogue_layout::compute_stub_managed_regs): Rename to...
28551 (xlogue_layout::count_stub_managed_regs): ...this.
28552 (xlogue_layout::is_stub_managed_reg): New function.
28553 (xlogue_layout::m_stub_names): Rename to...
28554 (xlogue_layout::s_stub_names): ...this, make static.
28555 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28556 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28557 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28558 xlogue_layout::s_stub_names): Instantiate statics.
28559 (stub_managed_regs): Remove.
28560 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28561 (disable_call_ms2sysv_xlogues): Rename to...
28562 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28563 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28565 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28566 change after reload_completed.
28567 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28569 (ix86_expand_prologue): Likewise.
28570 (ix86_expand_epilogue): Likewise.
28571 (ix86_expand_split_stack_prologue): Likewise.
28572 (ix86_compute_frame_layout): Remove frame parameter ...
28573 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28574 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28576 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28577 (ix86_frame): Move from here ...
28578 * config/i386/i386.h (ix86_frame): ... to here.
28579 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28580 complete ix86_frame data structure instead. Remove some_ld_name.
28582 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28584 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28585 symbols that hold a DECL_VALUE_EXPR.
28587 2017-06-01 Martin Jambor <mjambor@suse.cz>
28589 PR tree-optimization/80898
28590 * tree-sra.c (process_subtree_disqualification): Removed.
28591 (disqualify_candidate): Do not acll
28592 process_subtree_disqualification.
28593 (subtree_mark_written_and_enqueue): New function.
28594 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28595 RHS has been disqualified and re-queue LHS if necessary. Apart
28596 from that, ignore disqualified RHS.
28598 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28600 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28601 address fetch for z10 or later.
28603 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28605 * config/arc/arc.md (tst_movb): Add guard when splitting.
28607 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28609 * config/arc/arc.c (arc_can_eliminate): Test against
28610 arc_frame_pointer_needed.
28612 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28614 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28615 to prevent store reordering.
28616 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28617 (type): Add block type.
28618 (stack_tie): Define special instruction to be used in
28621 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28623 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28624 constraint. It is not valid for the pattern.
28625 (noncommutative_binary_comparison): Likewise.
28627 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28629 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28632 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28634 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28635 be used by the reg-alloc.
28637 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28639 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28640 reg-alloc when having mul64 or mul32x16 instructions.
28641 (mulsidi3): Likewise.
28642 (umulsidi3): Likewise.
28643 (mulsi32x16): New pattern.
28644 (mulsi64): Likewise.
28645 (mulsidi64): Likewise.
28646 (umulsidi64): Likewise.
28647 (MUL32x16_REG): Define.
28648 (mul64_600): Use MUL32x16_REG.
28649 (mac64_600): Likewise.
28650 (umul64_600): Likewise.
28651 (umac64_600): Likewise.
28653 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28655 * config/arc/arc.md (mulsi3_700): Make it commutative.
28657 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28659 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28661 (*sign_extendsidi2_insn): Likewise for movstosw.
28663 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28665 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28666 the type of the input discriminant value. Convert the
28667 discriminant value of signedness vary.
28669 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28671 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28672 Add to -Wall section.
28674 2017-06-01 Richard Biener <rguenther@suse.de>
28676 PR middle-end/66313
28677 * fold-const.c (fold_plusminus_mult_expr): If the factored
28678 factor may be zero use a wrapping type for the inner operation.
28679 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28680 and handle moved defs.
28681 (process_assignment): Properly guard the unary op case. Return a
28682 tri-state indicating that moving the stmt before the call may allow
28683 to continue. Pass through to_move.
28684 (find_tail_calls): Handle moving unrelated defs before
28687 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28690 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28691 splitter result in the canonical way.
28693 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28695 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28696 also for 32bit target. Update insn attributes.
28697 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28699 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28701 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28702 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28703 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28704 (_mm_maskz_min_ss): New intrinsics.
28706 2017-05-31 Martin Liska <mliska@suse.cz>
28708 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28709 Change comment style to one we normally use.
28710 (vectorizable_reduction): Likewise.
28711 (vectorizable_induction): Likewise.
28712 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28713 (vectorizable_call): Likewise.
28714 (vectorizable_simd_clone_call): Likewise.
28715 (vectorizable_conversion): Likewise.
28716 (vectorizable_assignment): Likewise.
28717 (vectorizable_shift): Likewise.
28718 (vectorizable_operation): Likewise.
28719 (vectorizable_store): Likewise.
28720 (vectorizable_load): Likewise.
28721 * tree-vectorizer.h: Likewise.
28723 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28725 * passes.c (emergency_dump_function): New.
28726 * tree-pass.h (emergency_dump_function): Declare.
28727 * plugin.c (plugins_internal_error_function): Remove.
28728 * plugin.h (plugins_internal_error_function): Remove declaration.
28729 * toplev.c (internal_error_function): New static function. Use it...
28730 (general_init): ...here.
28732 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28734 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28735 (arc_rtx_costs): Add costs for new patterns.
28736 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28737 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28739 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28741 * tree-ssa-strlen.c (get_next_strinfo): New function.
28742 (get_stridx_plus_constant): Use it.
28743 (zero_length_string): Likewise.
28744 (adjust_related_strinfos): Likewise.
28745 (adjust_last_stmt): Likewise.
28747 2017-05-31 Richard Biener <rguenther@suse.de>
28750 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28751 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28753 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28755 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28756 loop_vinfo argument and use of dependence distance vectors.
28757 Check instead whether the two references differ only in their
28758 initial value and assume that they have the same alignment if the
28759 difference is a multiple of the vector alignment.
28760 (vect_analyze_data_refs_alignment): Update call accordingly.
28762 2017-05-31 Martin Liska <mliska@suse.cz>
28765 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28767 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28769 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28770 (create_intersect_range_checks): Move from ...
28771 * tree-data-ref.c (create_intersect_range_checks_index)
28772 (create_intersect_range_checks): ... to here.
28773 (create_runtime_alias_checks): New function factored from ...
28774 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28775 here. Call above function.
28776 * tree-data-ref.h (create_runtime_alias_checks): New function.
28778 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28780 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28781 segment length for dr_b and compute it in wide_int.
28783 2017-05-31 Richard Biener <rguenther@suse.de>
28785 PR tree-optimization/80906
28786 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28787 and pass through iv_map.
28788 (copy_bb_and_scalar_dependences): Adjust.
28789 (translate_pending_phi_nodes): Likewise.
28790 (copy_loop_close_phi_args): Handle code-generating IVs instead
28793 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28795 * diagnostic-color.c (color_dict): Add "type-diff".
28796 (parse_gcc_colors): Update comment.
28797 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28798 -fdiagnostics-show-template-tree and -fno-elide-type.
28799 (GCC_COLORS): Add type-diff to example.
28801 (-fdiagnostics-show-template-tree): New.
28802 (-fno-elide-type): New.
28803 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28804 the pp_format_decoder callback. Call any m_format_postprocessor's
28806 (pretty_printer::pretty_printer): Initialize
28807 m_format_postprocessor.
28808 (pretty_printer::~pretty_printer): Delete any
28809 m_format_postprocessor.
28810 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28811 (class format_postprocessor): New class.
28812 (struct pretty_printer::format_decoder): Document the new parameters.
28813 (struct pretty_printer::m_format_postprocessor): New field.
28814 * tree-diagnostic.c (default_tree_printer): Update for new
28815 bool and const char ** params.
28816 * tree-diagnostic.h (default_tree_printer): Likewise.
28818 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28820 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28821 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28822 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28823 rs6000_gen_cell_microcode code.
28824 (rs6000_final_prescan_insn): Delete.
28825 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28826 "warn-cell-microcode" entries.
28827 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28828 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28829 throughout. Change cc_reg_not_micro_cr0_operand to
28830 cc_reg_not_cr0_operand throughout.
28831 (*extendhi<mode>2_noload): Delete.
28832 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28833 (mwarn-cell-microcode): Delete.
28834 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28835 -mgen-cell-microcode and -mwarn-cell-microcode.
28837 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28840 * config/i386/constraints.md (Yd): New constraint.
28842 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28843 and (?Yd, r) alternatives. Update insn attributes.
28844 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28845 and (?*Yd, r) alternatives. Update insn attributes.
28846 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28848 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28850 * gimplify.c (gimplify_modify_expr): Don't create a
28851 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
28854 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
28856 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
28858 2017-05-30 Richard Biener <rguenther@suse.de>
28860 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
28861 and reduc_def fields.
28862 (STMT_VINFO_REDUC_TYPE): New define.
28863 (STMT_VINFO_REDUC_DEF): Likewise.
28864 (vect_force_simple_reduction): Adjust prototype.
28865 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
28866 (vect_is_simple_reduction): Remove check_reduction argument.
28867 (vect_force_simple_reduction): Adjust and set
28868 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28869 (vectorizable_reduction): Do not re-do reduction analysis
28870 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28871 * tree-parloops.c (gather_scalar_reductions): Adjust.
28873 2017-05-30 Richard Biener <rguenther@suse.de>
28875 PR middle-end/80901
28876 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
28879 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
28881 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
28882 Introduce unknown_misalignment parameter and remove vf.
28883 (vect_peeling_hash_get_lowest_cost):
28884 Pass unknown_misalignment parameter.
28885 (vect_enhance_data_refs_alignment):
28886 Fix unsupportable data ref treatment.
28888 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28890 * tree-vect-data-refs.c (vect_get_data_access_cost):
28891 Workaround for SLP handling.
28892 (vect_enhance_data_refs_alignment):
28893 Compute costs for doing no peeling at all, compare to the best
28894 peeling costs so far and avoid peeling if cheaper.
28896 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28898 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
28899 Return peeling info and set costs to zero for unlimited cost
28901 (vect_enhance_data_refs_alignment): Also inspect all datarefs
28902 with unknown misalignment. Compute and costs for unknown
28903 misalignment, compare them to the costs for known misalignment
28904 and choose the cheapest for peeling.
28906 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28908 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
28909 (vect_get_peeling_costs_all_drs): Create function.
28910 (vect_peeling_hash_get_lowest_cost):
28911 Use vect_get_peeling_costs_all_drs.
28912 (vect_peeling_supportable): Create function.
28913 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
28915 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28917 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
28918 DR_HAS_NEGATIVE_STEP.
28919 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
28920 (vect_enhance_data_refs_alignment): Use.
28921 (vect_duplicate_ssa_name_ptr_info): Use.
28922 * tree-vectorizer.h (dr_misalignment): Use.
28923 (known_alignment_for_access_p): Use.
28925 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28928 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
28930 (has_section_name): New function.
28932 2017-05-30 Martin Liska <mliska@suse.cz>
28935 * auto-profile.c (get_function_decl_from_block): Fix
28938 2017-05-30 Richard Biener <rguenther@suse.de>
28940 PR middle-end/80876
28941 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
28943 2017-05-30 Martin Liska <mliska@suse.cz>
28945 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
28946 * dumpfile.h (struct dump_file_info): Remove ctors.
28948 2017-05-30 Martin Liska <mliska@suse.cz>
28950 * predict.def: Fix GNU coding style.
28952 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28954 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28955 Mark 'to' argument with ATTRIBUTE_UNUSED.
28957 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28959 * config/xtensa/xtensa.c (xtensa_emit_call): Use
28960 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
28961 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
28964 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
28966 * doc/install.texi (Options specification): Restore entry of
28967 --enable-sjlj-exceptions.
28969 2017-05-27 Michael Eager <eager@eagercon.com>
28972 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
28974 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
28976 * config/microblaze/microblaze.h
28977 (FIXED_REGISTERS): Update in macro.
28978 (CALL_USED_REGISTERS): Update in macro.
28980 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
28982 * doc/install.texi: Add links to macOS binary distributions.
28984 2017-05-27 Jakub Jelinek <jakub@redhat.com>
28988 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28990 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28992 2017-05-26 Martin Liska <mliska@suse.cz>
28994 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
28996 2017-05-26 Martin Liska <mliska@suse.cz>
28998 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
28999 always leading ';; '.
29000 (dump_bb_info): Likewise.
29001 (brief_dump_cfg): Likewise.
29002 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29003 * dumpfile.c: Remove usage of TDF_VERBOSE.
29004 * dumpfile.h (enum dump_kind): Likewise.
29005 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29006 * print-tree.c (debug_verbose): Remove.
29007 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29008 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29009 * tree-diagnostic.c (default_tree_printer): Replace
29010 TDF_DIAGNOSTIC with TDF_SLIM.
29012 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29014 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29015 in parameter loop, rather than loop_vinfo.
29016 (create_intersect_range_checks): Ditto.
29017 (vect_create_cond_for_alias_checks): Update call to above functions.
29019 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29021 PR tree-optimization/80815
29022 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29023 for merging runtime alias checks. Handle negative DR_STEPs.
29025 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29027 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29029 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29030 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29031 out code pruning runtime alias checks.
29032 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29033 factored out from above.
29034 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29036 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29038 (prune_runtime_alias_test_list): New decalaration.
29040 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29042 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29043 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29044 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29045 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29046 (operator==, comp_dr_with_seg_len_pair): Ditto.
29047 (vect_prune_runtime_alias_test_list): Ditto.
29049 2017-05-26 Martin Liska <mliska@suse.cz>
29052 * params.def: Bound partial-inlining-entry-probability param.
29054 2017-05-26 Marek Polacek <polacek@redhat.com>
29057 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29060 2017-05-26 Richard Biener <rguenther@suse.de>
29062 PR tree-optimization/80842
29063 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29066 2017-05-26 Richard Biener <rguenther@suse.de>
29068 PR tree-optimization/80844
29069 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29071 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29073 * doc/md.texi (Machine Constraints): Update x86 family
29074 machine constraints section to match 'config/i386/constraints.md'.
29076 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29078 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29080 2017-05-25 Nathan Sidwell <nathan@acm.org>
29082 * doc/invoke.texi (--enable-languages): Update documentation.
29084 2017-05-25 Martin Liska <mliska@suse.cz>
29086 * dumpfile.c: Add TDF_FOLDING.
29087 * dumpfile.h (enum dump_kind): Likewise.
29088 * genmatch.c (dt_simplify::gen_1): Use it.
29090 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29092 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29094 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29096 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29097 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29099 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29101 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29102 * match.pd (X == C): Rewrite it here.
29103 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29104 with_certain_nonzero_bits2): New predicates.
29105 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29107 2017-05-24 Nathan Sidwell <nathan@acm.org>
29109 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29112 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29115 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29117 * config/powerpcspe: New port. Files are copied from the rs6000
29118 port, with "rs6000" in filenames replaced by "powerpcspe".
29120 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29122 PR rtl-optimization/80754
29123 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29125 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29127 * config/sparc/sparc.md (length): Return the correct value for -mflat
29128 sibcalls to match output_sibcall.
29130 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29134 * config/rs6000/rs6000.c (struct machine_function): Add new field
29136 (rs6000_get_separate_components): Init that field, use it.
29137 (rs6000_components_for_bb): Use the field.
29139 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29141 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29143 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29145 PR middle-end/80823
29146 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29148 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29151 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29152 address against address_operand predicate.
29153 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29155 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29157 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29158 parameters passed indirectly.
29160 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29162 * config/i386/i386.md (*movdi_internal): Remove SSE4
29163 alternative 18 (?r, *v). Update insn attributes.
29164 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29165 Update insn attributes.
29166 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29167 Update insn attributes.
29168 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29169 alternative 1 (r, v). Remove isa attribute.
29170 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29171 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29172 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29174 2017-05-23 Tom de Vries <tom@codesourcery.com>
29176 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29179 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29181 * cgraphunit.c (symbol_table::process_new_functions): Update.
29182 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29183 (inline_generate_summary): Rename to ...
29184 (ipa_fn_summary_generate): ... this one.
29185 (inline_read_summary): Rename to ...
29186 (ipa_fn_summary_read): ... this one.
29187 (inline_write_summary): Rename to ...
29188 (ipa_fn_summary_write): ... this one.
29189 (inline_free_summary): Rename to ...
29190 (ipa_free_fn_summary): ... this one.
29191 (pass_data_local_fn_summary, pass_local_fn_summary,
29192 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29193 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29194 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29195 make_pass_ipa_fn_summary): New.
29196 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29197 inline_write_summary, inline_free_summary): Remove.
29198 (ipa_free_fn_summary) : New.
29199 * ipa-inline.c (ipa_inline): Update.
29200 (pass_ipa_inline): Do not generate summaries.
29201 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29203 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29204 and add pass_ipa_fn_summary.
29205 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29207 (make_pass_inline_parameters): Remove.
29209 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29211 * omp-low.c (struct omp_context): Remove "default_kind" member.
29214 * omp-offload.c (execute_oacc_device_lower): Remove the
29215 parallelism dimensions function attributes for unparallelized
29216 OpenACC kernels constructs.
29218 2017-05-23 Martin Liska <mliska@suse.cz>
29220 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29222 (cgraph_edge::make_speculative): Likewise.
29223 (cgraph_edge::resolve_speculation): Likewise.
29224 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29225 (cgraph_node::dump): Likewise.
29226 * cgraph.h: Likewise.
29227 * cgraphunit.c (analyze_functions): Likewise.
29228 (symbol_table::compile): Likewise.
29229 * ipa-cp.c (print_all_lattices): Likewise.
29230 (determine_versionability): Likewise.
29231 (initialize_node_lattices): Likewise.
29232 (ipcp_verify_propagated_values): Likewise.
29233 (estimate_local_effects): Likewise.
29234 (update_profiling_info): Likewise.
29235 (create_specialized_node): Likewise.
29236 (perhaps_add_new_callers): Likewise.
29237 (decide_about_value): Likewise.
29238 (decide_whether_version_node): Likewise.
29239 (identify_dead_nodes): Likewise.
29240 (ipcp_store_bits_results): Likewise.
29241 * ipa-devirt.c (dump_targets): Likewise.
29242 (ipa_devirt): Likewise.
29243 * ipa-icf.c (sem_item::dump): Likewise.
29244 (sem_function::equals): Likewise.
29245 (sem_variable::equals): Likewise.
29246 (sem_item_optimizer::read_section): Likewise.
29247 (sem_item_optimizer::execute): Likewise.
29248 (congruence_class::dump): Likewise.
29249 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29250 (dump_inline_summary): Likewise.
29251 (estimate_node_size_and_time): Likewise.
29252 (inline_analyze_function): Likewise.
29253 * ipa-inline-transform.c (inline_call): Likewise.
29254 * ipa-inline.c (report_inline_failed_reason): Likewise.
29255 (want_early_inline_function_p): Likewise.
29256 (edge_badness): Likewise.
29257 (update_edge_key): Likewise.
29258 (inline_small_functions): Likewise.
29259 * ipa-profile.c (ipa_profile): Likewise.
29260 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29261 (ipa_make_edge_direct_to_target): Likewise.
29262 (remove_described_reference): Likewise.
29263 (ipa_impossible_devirt_target): Likewise.
29264 (propagate_controlled_uses): Likewise.
29265 (ipa_print_node_params): Likewise.
29266 (ipcp_transform_function): Likewise.
29267 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29268 (propagate_pure_const): Likewise.
29269 * ipa-reference.c (generate_summary): Likewise.
29270 (read_write_all_from_decl): Likewise.
29271 (propagate): Likewise.
29272 (ipa_reference_read_optimization_summary): Likewise.
29273 * ipa-utils.c (ipa_merge_profiles): Likewise.
29274 * ipa.c (walk_polymorphic_call_targets): Likewise.
29275 (symbol_table::remove_unreachable_nodes): Likewise.
29276 (ipa_single_use): Likewise.
29277 * passes.c (execute_todo): Likewise.
29278 * predict.c (drop_profile): Likewise.
29279 * symtab.c (symtab_node::get_dump_name): New function.
29280 (symtab_node::dump_name): Likewise.
29281 (symtab_node::dump_asm_name): Likewise.
29282 (symtab_node::dump_references): Likewise.
29283 (symtab_node::dump_referring): Likewise.
29284 (symtab_node::dump_base): Likewise.
29285 (symtab_node::debug_symtab): Likewise.
29286 * tree-sra.c (convert_callers_for_node): Likewise.
29287 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29288 * value-prof.c (init_node_map): Likewise.
29290 2017-05-23 Martin Liska <mliska@suse.cz>
29292 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29293 and symtab_node::debug_symtab to symbol_table::debug.
29294 * cgraphunit.c (analyze_functions): Use the renamed function.
29295 (symbol_table::compile): Likewise.
29296 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29297 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29298 * passes.c (execute_todo): Likewise.
29299 * symtab.c (symbol_table::dump): New function.
29300 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29302 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29304 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29305 that nonconst implies exec.
29307 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29309 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29310 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29311 (inline_edge_summary_vec): Turn into ...
29312 (ipa_call_summaries): ... this one.
29313 (redirect_to_unreachable, edge_set_predicate,
29314 evaluate_properties_for_edge, inline_summary_alloc,
29315 reset_ipa_call_summary, reset_inline_summary,
29316 inline_summary_t::duplicate): Update.
29317 (inline_edge_duplication_hook): Turn to ...
29318 (ipa_call_summary_t::duplicate): ... this one.
29319 (inline_edge_removal_hook): Turn to ...
29320 (ipa_call_summary_t::remove): ... this one.
29321 (dump_inline_edge_summary): Turn to ...
29322 (dump_ipa_call_summary): ... this one.
29323 (estimate_function_body_sizes): Update.
29324 (inline_update_callee_summaries): Update.
29325 (remap_edge_change_prob): Update.
29326 (remap_edge_summaries): Update.
29327 (inline_merge_summary): Update.
29328 (do_estimate_edge_time): Update.
29329 (inline_generate_summary): Update.
29330 (inline_read_section): Update.
29331 (inline_read_summary): Update.
29332 (inline_free_summary): Update.
29333 * ipa-inline.c (can_inline_edge_p): Update.
29334 (compute_inlined_call_time): Update.
29335 (want_inline_small_function_p): Update.
29336 (edge_badness): Update.
29337 (early_inliner): Update.
29338 * ipa-inline.h (inline_edge_summary): Turn to ...
29339 (ipa_call_summary): ... this one.
29340 (ipa_call_summary_t): New class.
29341 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29342 (ipa_call_summaries): New.
29343 (inline_edge_summary): Remove.
29344 (estimate_edge_growth): Update.
29345 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29346 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29347 * ipa-split.c (execute_split_functions): Update.
29348 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29350 2017-05-23 Tom de Vries <tom@codesourcery.com>
29352 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29353 attributes): Document rdrand effective target.
29355 2017-05-23 Tom de Vries <tom@codesourcery.com>
29357 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29358 attributes): Sort alphabetically.
29360 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29362 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29364 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29367 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29368 V2DF/V2DI splat into two separate patterns, one that handles
29369 registers, and the other that only handles memory. Drop support
29370 for splatting from a GPR on ISA 2.07 and then splitting the
29371 splat into direct move and splat.
29372 (vsx_splat_<mode>_reg): Likewise.
29373 (vsx_splat_<mode>_mem): Likewise.
29375 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29377 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29379 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29381 PR middle-end/80809
29382 * omp-low.c (finish_taskreg_remap): New function.
29383 (finish_taskreg_scan): If unit size of ctx->record_type
29384 is non-constant, unshare the size expression and replace
29385 decls in it with possible outer var refs.
29387 PR middle-end/80809
29388 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29389 GOVD_SHARED rather than GOVD_PRIVATE with it.
29390 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29391 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29393 PR middle-end/80853
29394 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29395 as last argument to build_outer_var_ref for pointer bases of array
29396 section reductions.
29398 2017-05-19 Martin Sebor <msebor@redhat.com>
29400 * print-tree.c (print_node): Print DECL_READ_P flag.
29402 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29404 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29405 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29406 * cgraph.c: Likewise.
29407 * cgraphunit.c: Likewise.
29408 * gengtype.c: Likewise.
29409 * ipa-cp.c: Likewise.
29410 * ipa-devirt.c: Likewise.
29411 * ipa-icf.c: Likewise.
29412 * ipa-predicate.c: Likewise.
29413 * ipa-profile.c: Likewise.
29414 * ipa-prop.c: Likewise.
29415 * ipa-split.c: Likewise.
29417 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29418 edge_predicate_pool, dump_inline_hints,
29419 inline_summary::account_size_time, redirect_to_unreachable,
29420 edge_set_predicate, set_hint_predicate,
29421 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29422 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29423 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29424 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29425 ipa_call_summary_t::remove, initialize_growth_caches,
29426 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29427 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29428 mark_modified, unmodified_parm_1, unmodified_parm,
29429 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29430 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29431 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29432 will_be_nonconstant_predicate, record_modified_bb_info,
29433 get_minimal_bb, record_modified, param_change_prob,
29434 phi_result_unknown_predicate, predicate_for_phi_result,
29435 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29436 estimate_function_body_sizes, compute_inline_parameters,
29437 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29438 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29439 inline_update_callee_summaries, remap_edge_change_prob,
29440 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29441 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29442 inline_analyze_function, inline_summary_t::insert,
29443 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29444 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29445 inline_free_summary): Move to ipa-fnsummary.h
29446 (predicate_t): Remove.
29447 * ipa-fnsummary.c: New file.
29448 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29449 (enum inline_hints_vals, inline_hints, agg_position_info,
29450 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29451 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29452 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29453 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29454 inline_read_summary, inline_write_summary, inline_free_summary,
29455 inline_analyze_function, initialize_inline_failed,
29456 inline_merge_summary, inline_update_overall_summary,
29457 compute_inline_parameters): Move to ipa-fnsummary.h
29458 * ipa-fnsummary.h: New file.
29459 * ipa-inline-transform.h: Include ipa-inline.h.
29460 * ipa-inline.c: LIkewise.
29462 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29464 * ipa-inline.c (edge_badness): Use inlined_time instead of
29465 inline_summaries->get.
29467 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29469 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29471 2017-05-22 Nathan Sidwell <nathan@acm.org>
29473 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29474 (fdump-lang): Document 'raw' option.
29475 * dumpfile.h (TDI_tu): Delete.
29476 * dumpfile.c (dump_files): Remove translation-unit.
29477 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29479 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29481 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29482 command option from $(AWK) call.
29483 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29484 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29485 [FORMAT]: Remove handling of variable.
29486 * config/avr/t-multilib: Regenerate.
29488 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29490 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29492 (dump_inline_summary): Do not print self_time.
29493 (estimate_function_body_sizes): Do not set self_time.
29494 (compute_inline_parameters): Likewise.
29495 (inline_read_section, inline_write_summary): Do not stream self_time.
29496 * ipa-inline.h (inline_summary): Drop self_time.
29498 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29500 * ipa-inline-analysis.c (account_size_time): Rename to ...
29501 (inline_summary::account_size_time): ... this one.
29502 (reset_ipa_call_summary): Turn to ...
29503 (ipa_call_summary::reset): ... this one.
29504 (reset_inline_summary): Turn to ...
29505 (inline_summary::reset): ... this one.
29506 (inline_summary_t::remove): Update.
29507 (inline_summary_t::duplicate): Update.
29508 (ipa_call_summary_t::remove): Update.
29509 (dump_inline_summary): Update.
29510 (estimate_function_body_sizes): Update.
29511 (compute_inline_parameters): Update.
29512 (estimate_node_size_and_time): Update.
29513 (inline_merge_summary): Update.
29514 (inline_update_overall_summary): Update.
29515 (inline_read_section): Update.
29516 (inline_write_summary): Update.
29517 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29518 add account_size_time and reset member functions.
29519 (ipa_call_summary): Add reset function.
29520 * ipa-predicate.h (predicate::operator &): Constify.
29522 2017-05-22 Richard Biener <rguenther@suse.de>
29524 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29526 2017-05-19 Jason Merrill <jason@redhat.com>
29528 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29530 2017-05-19 Marek Polacek <polacek@redhat.com>
29533 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29534 TYPE_OVERFLOW_WRAPS checks.
29536 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29538 * tree-core.h (enum omp_clause_default_kind): Add
29539 "OMP_CLAUSE_DEFAULT_PRESENT".
29540 * tree-pretty-print.c (dump_omp_clause): Handle it.
29541 * gimplify.c (enum gimplify_omp_var_data): Add
29542 "GOVD_MAP_FORCE_PRESENT".
29543 (gimplify_adjust_omp_clauses_1): Map it to
29544 "GOMP_MAP_FORCE_PRESENT".
29545 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29547 * gimplify.c (oacc_default_clause): Clarify.
29549 2017-05-19 Nathan Sidwell <nathan@acm.org>
29551 LANG_HOOK_REGISTER_DUMPS
29552 * toplev.c (general_init): Call register dump lang hook.
29553 * doc/invoke.texi: Document -fdump-lang option family.
29554 * dumpfile.c (dump_files): Remove class dump here.
29555 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29556 * dumpfile.h (tree_dump_index): Remove TDI_class.
29557 * langhooks-def.h (lhd_register_dumps): Declare.
29558 (LANG_HOOKS_REGISTER_DUMPS): Define.
29559 (LANG_HOOKS_INITIALIZER): Add it.
29560 * langhooks.c (lhd_register_dumps): Define.
29561 * langhooks.h (struct lang_hooks): Add register_dumps.
29563 2017-05-19 Nathan Sidwell <nathan@acm.org>
29565 * context.h (context::set_passes): New.
29566 * context.c (context::context): Do not create pass manager.
29567 * toplev.c (general_init): Create pass manager here.
29569 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29571 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29572 use this splitter if two add or or instructions would also work for
29573 the constant we want to generate.
29575 2017-05-19 Richard Biener <rguenther@suse.de>
29578 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29579 predicate evaluation.
29581 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29583 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29585 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29588 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29590 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29591 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29592 (inline_edge_summary_vec): Turn into ...
29593 (ipa_call_summaries): ... this one.
29594 (redirect_to_unreachable, edge_set_predicate,
29595 evaluate_properties_for_edge, inline_summary_alloc,
29596 reset_ipa_call_summary, reset_inline_summary,
29597 inline_summary_t::duplicate): Update.
29598 (inline_edge_duplication_hook): Turn to ...
29599 (ipa_call_summary_t::duplicate): ... this one.
29600 (inline_edge_removal_hook): Turn to ...
29601 (ipa_call_summary_t::remove): ... this one.
29602 (dump_inline_edge_summary): Turn to ...
29603 (dump_ipa_call_summary): ... this one.
29604 (estimate_function_body_sizes): Update.
29605 (inline_update_callee_summaries): Update.
29606 (remap_edge_change_prob): Update.
29607 (remap_edge_summaries): Update.
29608 (inline_merge_summary): Update.
29609 (do_estimate_edge_time): Update.
29610 (inline_generate_summary): Update.
29611 (inline_read_section): Update.
29612 (inline_read_summary): Update.
29613 (inline_free_summary): Update.
29614 * ipa-inline.c (can_inline_edge_p): Update.
29615 (compute_inlined_call_time): Update.
29616 (want_inline_small_function_p): Update.
29617 (edge_badness): Update.
29618 (early_inliner): Update.
29619 * ipa-inline.h (inline_edge_summary): Turn to ...
29620 (ipa_call_summary): ... this one.
29621 (ipa_call_summary_t): New class.
29622 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29623 (ipa_call_summaries): New.
29624 (inline_edge_summary): Remove.
29625 (estimate_edge_growth): Update.
29626 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29627 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29628 * ipa-split.c (execute_split_functions): Update.
29629 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29631 2017-05-19 Richard Biener <rguenther@suse.de>
29633 PR middle-end/80764
29634 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29636 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29638 * config/rs6000/rs6000.c (struct machine_function): Add field
29639 fpr_is_wrapped_separately.
29640 (rs6000_get_separate_components): Use 64 components. Handle the
29641 new FPR components.
29642 (rs6000_components_for_bb): Handle the FPR components.
29643 (rs6000_emit_prologue_components): Handle the FPR components.
29644 (rs6000_emit_epilogue_components): Handle the FPR components.
29645 (rs6000_set_handled_components): Handle the FPR components.
29646 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29647 that are already separately shrink-wrapped.
29648 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29649 that are already separately shrink-wrapped.
29651 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29654 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29657 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29658 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29659 cases where the register allocator uses a move and an offsettable
29660 memory operation to/from a FPR register on ISA 2.06/2.07.
29661 (define_peephole2 for Altivec d-form store): Likewise.
29663 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29666 * config/i386/mmx.md (*mov<mode>_internal): Enable
29667 alternatives 11, 12, 13 and 14 also for 32bit targets.
29668 Remove alternatives 15, 16, 17 and 18.
29669 * config/i386/sse.md (vec_concatv2di): Change
29670 alternative (!x, *y) to (x, ?!*Yn).
29672 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29674 * dumpfile.h (enum dump_kind): Remove stray comma.
29676 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29678 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29679 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29680 predicate::num_conditions
29681 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29682 (CHANGED): turn into predicate::changed.
29683 (agg_position_info): Move to ipa-predicate.h
29684 (add_condition, predicate::add_clause, predicate::operator &=,
29685 predicate::or_with, predicate::evaluate, predicate::probability,
29686 dump_condition, dump_clause, predicate::dump,
29687 predicate::remap_after_duplication, predicate::remap_after_inlining,
29688 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29689 (evaluate_conditions_for_known_args): Update.
29690 (set_cond_stmt_execution_predicate): Update.
29691 * ipa-inline.h: Include ipa-predicate.h
29692 (condition, inline_param_summary, conditions, agg_position_info,
29693 predicate): Move to ipa-predicate.h
29694 * ipa-predicate.c: New file.
29695 * ipa-predicate.h: New file.
29697 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29699 * final.c (leaf_function_p): Check we are not in a sequence.
29701 2017-05-18 Martin Liska <mliska@suse.cz>
29703 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29704 * dumpfile.c (dump_register): Use new enum dump_kind.
29705 (get_dump_file_name): Likewise.
29706 (dump_enable_all): Likewise.
29707 (dump_switch_p_1): Likewise.
29708 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29709 * dumpfile.h (enum dump_kind): New enum type.
29710 (struct dump_file_info): Create constructor and
29711 format fields and comments.
29712 * passes.c (pass_manager::register_one_dump_file):
29714 * statistics.c (statistics_early_init): Likewise.
29715 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29716 TDF_TREE with TDF_SLIM.
29717 (gather_memory_references_ref): Likewise.
29719 2017-05-18 Martin Liska <mliska@suse.cz>
29721 * vec.h (struct vnull): Use it.
29723 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29725 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29726 (true_predicate, false_predicate, true_predicate_p,
29727 false_predicate_p): Remove.
29728 (single_cond_predicate, not_inlined_predicate): Turn to member function
29730 (add_condition): Update.
29731 (add_clause): Turn to...
29732 (predicate::add_clause): ... this one; update; allow passing NULL
29734 (and_predicates): Turn to ...
29735 (predicate::operator &=): ... this one.
29736 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29737 (or_predicates): Turn to ...
29738 (predicate::or_with): ... this one.
29739 (evaluate_predicate): Turn to ...
29740 (predicate::evaluate): ... this one.
29741 (predicate_probability): Turn to ...
29742 (predicate::probability): ... this one.
29743 (dump_condition): Update.
29744 (dump_predicate): Turn to ...
29745 (predicate::dump): ... this one.
29746 (account_size_time): Update.
29747 (edge_set_predicate): Update.
29748 (set_hint_predicate): UPdate.
29749 (evaluate_conditions_for_known_args): Update.
29750 (evaluate_properties_for_edge): Update.
29751 (remap_predicate_after_duplication): Turn to...
29752 (predicate::remap_after_duplication): ... this one.
29753 (remap_hint_predicate_after_duplication): Update.
29754 (inline_summary_t::duplicate): UPdate.
29755 (dump_inline_edge_summary): Update.
29756 (dump_inline_summary): Update.
29757 (set_cond_stmt_execution_predicate): Update.
29758 (set_switch_stmt_execution_predicate): Update.
29759 (compute_bb_predicates): Update.
29760 (will_be_nonconstant_expr_predicate): Update.
29761 (will_be_nonconstant_predicate): Update.
29762 (phi_result_unknown_predicate): Update.
29763 (predicate_for_phi_result): Update.
29764 (array_index_predicate): Update.
29765 (estimate_function_body_sizes): Update.
29766 (estimate_node_size_and_time): Update.
29767 (estimate_ipcp_clone_size_and_time): Update.
29768 (remap_predicate): Rename to ...
29769 (predicate::remap_after_inlining): ... this one.
29770 (remap_hint_predicate): Update.
29771 (inline_merge_summary): Update.
29772 (inline_update_overall_summary): Update.
29773 (estimate_size_after_inlining): Update.
29774 (read_predicate): Rename to ...
29775 (predicate::stream_in): ... this one.
29776 (read_inline_edge_summary): Update.
29777 (write_predicate): Rename to ...
29778 (predicate::stream_out): ... this one.
29779 (write_inline_edge_summary): Update.
29780 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29781 (clause_t): Turn to uint32_t
29782 (predicate): Turn to class; implement constructor and operators
29784 (size_time_entry): Update.
29785 (inline_summary): Update.
29786 (inline_edge_summary): Update.
29788 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29790 * fold-const.c (fold_binary_loc): Move transformation...
29791 * match.pd (C - X CMP X): ... here.
29793 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29795 * config/sparc/sparc.c (sparc_option_override): Set function
29796 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29797 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29799 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29801 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29803 2017-05-18 Marek Polacek <polacek@redhat.com>
29806 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29807 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29808 gimple_assign_load_p.
29810 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29812 PR middle-end/80692
29813 * real.c (do_compare): Give decimal_do_compare preference over
29814 comparing just the signs.
29816 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29818 * doc/md.texi (Canonicalization of Instructions): Describe the
29819 canonical form of instructions that inherently set a condition
29822 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29824 PR middle-end/80775
29825 * tree-cfg.c: Move deletion of unreachable case statements to after
29826 the merging of consecutive case labels.
29828 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29830 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29831 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29832 restoring of callee-saved registers.
29834 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29836 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29837 * config/visium/visium.c (single_set_and_flags): Likewise.
29838 * config/visium/visium.md (Substitutions): Likewise.
29840 2017-05-17 Martin Liska <mliska@suse.cz>
29842 * cfg.c: Introduce dump_flags_t type and
29843 use it instead of int type.
29845 * cfghooks.c: Likewise.
29846 * cfghooks.h (struct cfg_hooks): Likewise.
29847 * cfgrtl.c: Likewise.
29848 * cfgrtl.h: Likewise.
29849 * cgraph.c (cgraph_node::get_body): Likewise.
29850 * coretypes.h: Likewise.
29851 * domwalk.c: Likewise.
29852 * domwalk.h: Likewise.
29853 * dumpfile.c (struct dump_option_value_info): Likewise.
29854 (dump_enable_all): Likewise.
29855 (dump_switch_p_1): Likewise.
29856 (opt_info_switch_p): Likewise.
29857 * dumpfile.h (enum tree_dump_index): Likewise.
29858 (struct dump_file_info): Likewise.
29859 * genemit.c: Likewise.
29860 * generic-match-head.c: Likewise.
29861 * gengtype.c (open_base_files): Likewise.
29862 * gimple-pretty-print.c: Likewise.
29863 * gimple-pretty-print.h: Likewise.
29864 * graph.c (print_graph_cfg): Likewise.
29865 * graphite-scop-detection.c (dot_all_sese): Likewise.
29866 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
29867 * loop-unroll.c (report_unroll): Likewise.
29868 * passes.c (pass_manager::register_one_dump_file): Likewise.
29869 * print-tree.c: Likewise.
29870 * statistics.c: Likewise.
29871 * tree-cfg.c: Likewise.
29872 * tree-cfg.h: Likewise.
29873 * tree-dfa.c: Likewise.
29874 * tree-dfa.h: Likewise.
29875 * tree-dump.c (dump_function): Likewise.
29876 * tree-dump.h (struct dump_info): Likewise.
29877 * tree-pretty-print.c: Likewise.
29878 * tree-pretty-print.h: Likewise.
29879 * tree-ssa-live.c: Likewise.
29880 * tree-ssa-live.h: Likewise.
29881 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
29882 * tree-vect-loop.c: Likewise.
29883 * tree-vect-slp.c: Likewise.
29885 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
29886 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29888 PR tree-optimization/80457
29889 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
29890 of all arguments to a statement as scalar_to_vec operations.
29891 (vectorizable_call): Adjust call to vect_model_simple_cost for
29893 (vectorizable_conversion): Likewise.
29894 (vectorizable_assignment): Likewise.
29895 (vectorizable_shift): Likewise.
29896 (vectorizable_operation): Likewise.
29897 (vectorizable_comparison): Likewise.
29898 (vect_is_simple_cond): Record the def types for operands.
29899 (vectorizable_condition): Likewise, call vect_model_simple_cost.
29900 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
29901 for statement argument count.
29903 2017-05-16 Carl Love <cel@us.ibm.com>
29905 * config/rs6000/rs6000-c: Add support for built-in functions
29906 vector unsigned long long vec_bperm (vector unsigned long long,
29907 vector unsigned char)
29908 vector signed long long vec_mule (vector signed int,
29910 vector unsigned long long vec_mule (vector unsigned int,
29911 vector unsigned int)
29912 vector signed long long vec_mulo (vector signed int,
29914 vector unsigned long long vec_mulo (vector unsigned int,
29915 vector unsigned int)
29916 vector signed char vec_sldw (vector signed char,
29917 vector signed char,
29919 vector unsigned char vec_sldw (vector unsigned char,
29920 vector unsigned char,
29922 vector signed short vec_sldw (vector signed short,
29923 vector signed short,
29925 vector unsigned short vec_sldw (vector unsigned short,
29926 vector unsigned short,
29928 vector signed int vec_sldw (vector signed int,
29931 vector unsigned int vec_sldw (vector unsigned int,
29932 vector unsigned int,
29934 vector signed long long vec_sldw (vector signed long long,
29935 vector signed long long,
29937 vector unsigned long long vec_sldw (vector unsigned long long,
29938 vector unsigned long long,
29940 * config/rs6000/rs6000-c: Add support for built-in functions
29941 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
29942 * config/rs6000/altivec.h: Add defintion for vec_sldw.
29943 * doc/extend.texi: Update the built-in documentation for the
29944 new built-in functions.
29946 2017-05-16 Marek Polacek <polacek@redhat.com>
29950 * tree.c (save_expr): Don't fold the expression.
29952 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
29954 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
29955 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
29956 and (?*y,m). Update insn attributes.
29958 2017-05-16 Martin Liska <mliska@suse.cz>
29960 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
29961 flags argument of print_gimple_stmt, print_gimple_expr,
29962 print_generic_stmt and print_generic_expr.
29963 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
29964 * coretypes.h: Likewise.
29965 * except.c (dump_eh_tree): Likewise.
29966 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
29967 * gimple-pretty-print.h: Likewise.
29968 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
29969 (backprop::push_to_worklist): Likewise.
29970 (backprop::pop_from_worklist): Likewise.
29971 (backprop::process_use): Likewise.
29972 (backprop::intersect_uses): Likewise.
29973 (note_replacement): Likewise.
29974 * gimple-ssa-store-merging.c
29975 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
29976 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
29977 (pass_store_merging::execute): Likewise.
29978 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
29979 (ssa_base_cand_dump_callback): Likewise.
29980 (dump_incr_vec): Likewise.
29981 (replace_refs): Likewise.
29982 (replace_mult_candidate): Likewise.
29983 (create_add_on_incoming_edge): Likewise.
29984 (create_phi_basis): Likewise.
29985 (insert_initializers): Likewise.
29986 (all_phi_incrs_profitable): Likewise.
29987 (introduce_cast_before_cand): Likewise.
29988 (replace_one_candidate): Likewise.
29989 * gimplify.c (gimplify_expr): Likewise.
29990 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
29991 (set_rename): Likewise.
29992 (rename_uses): Likewise.
29993 (copy_loop_phi_nodes): Likewise.
29994 (add_close_phis_to_merge_points): Likewise.
29995 (copy_loop_close_phi_args): Likewise.
29996 (copy_cond_phi_args): Likewise.
29997 (graphite_copy_stmts_from_block): Likewise.
29998 (translate_pending_phi_nodes): Likewise.
29999 * graphite-poly.c (print_pdr): Likewise.
30000 (dump_gbb_cases): Likewise.
30001 (dump_gbb_conditions): Likewise.
30002 (print_scop_params): Likewise.
30003 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30004 (build_cross_bb_scalars_use): Likewise.
30005 (gather_bbs::before_dom_children): Likewise.
30006 * hsa-dump.c (dump_hsa_immed): Likewise.
30007 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30008 (get_replacement_map): Likewise.
30009 * ipa-inline-analysis.c (dump_condition): Likewise.
30010 (estimate_function_body_sizes): Likewise.
30011 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30012 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30013 * ipa-prop.c (ipa_dump_param): Likewise.
30014 (ipa_print_node_jump_functions_for_edge): Likewise.
30015 (ipa_modify_call_arguments): Likewise.
30016 (ipa_modify_expr): Likewise.
30017 (ipa_dump_param_adjustments): Likewise.
30018 (ipa_dump_agg_replacement_values): Likewise.
30019 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30020 * ipa-pure-const.c (check_stmt): Likewise.
30021 (pass_nothrow::execute): Likewise.
30022 * ipa-split.c (execute_split_functions): Likewise.
30023 * omp-offload.c (dump_oacc_loop_part): Likewise.
30024 (dump_oacc_loop): Likewise.
30025 * trans-mem.c (tm_log_emit): Likewise.
30026 (tm_memopt_accumulate_memops): Likewise.
30027 (dump_tm_memopt_set): Likewise.
30028 (dump_tm_memopt_transform): Likewise.
30029 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30030 (print_loop): Likewise.
30031 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30032 (chkp_gather_checks_info): Likewise.
30033 (chkp_get_check_result): Likewise.
30034 (chkp_remove_check_if_pass): Likewise.
30035 (chkp_use_outer_bounds_if_possible): Likewise.
30036 (chkp_reduce_bounds_lifetime): Likewise.
30037 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30038 (chkp_mark_completed_bounds): Likewise.
30039 (chkp_register_incomplete_bounds): Likewise.
30040 (chkp_mark_invalid_bounds): Likewise.
30041 (chkp_maybe_copy_and_register_bounds): Likewise.
30042 (chkp_build_returned_bound): Likewise.
30043 (chkp_get_bound_for_parm): Likewise.
30044 (chkp_build_bndldx): Likewise.
30045 (chkp_get_bounds_by_definition): Likewise.
30046 (chkp_generate_extern_var_bounds): Likewise.
30047 (chkp_get_bounds_for_decl_addr): Likewise.
30048 * tree-chrec.c (chrec_apply): Likewise.
30049 * tree-data-ref.c (dump_data_reference): Likewise.
30050 (dump_subscript): Likewise.
30051 (dump_data_dependence_relation): Likewise.
30052 (analyze_overlapping_iterations): Likewise.
30053 * tree-inline.c (expand_call_inline): Likewise.
30054 (tree_function_versioning): Likewise.
30055 * tree-into-ssa.c (dump_defs_stack): Likewise.
30056 (dump_currdefs): Likewise.
30057 (dump_names_replaced_by): Likewise.
30058 (dump_update_ssa): Likewise.
30059 (update_ssa): Likewise.
30060 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30061 * tree-parloops.c (build_new_reduction): Likewise.
30062 (try_create_reduction_list): Likewise.
30063 (ref_conflicts_with_region): Likewise.
30064 (oacc_entry_exit_ok_1): Likewise.
30065 (oacc_entry_exit_single_gang): Likewise.
30066 * tree-pretty-print.h: Likewise.
30067 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30068 (get_scalar_evolution): Likewise.
30069 (add_to_evolution): Likewise.
30070 (get_loop_exit_condition): Likewise.
30071 (analyze_evolution_in_loop): Likewise.
30072 (analyze_initial_condition): Likewise.
30073 (analyze_scalar_evolution): Likewise.
30074 (instantiate_scev): Likewise.
30075 (number_of_latch_executions): Likewise.
30076 (gather_chrec_stats): Likewise.
30077 (final_value_replacement_loop): Likewise.
30078 (scev_const_prop): Likewise.
30079 * tree-sra.c (dump_access): Likewise.
30080 (disqualify_candidate): Likewise.
30081 (create_access): Likewise.
30082 (reject): Likewise.
30083 (maybe_add_sra_candidate): Likewise.
30084 (create_access_replacement): Likewise.
30085 (analyze_access_subtree): Likewise.
30086 (analyze_all_variable_accesses): Likewise.
30087 (sra_modify_assign): Likewise.
30088 (initialize_constant_pool_replacements): Likewise.
30089 (find_param_candidates): Likewise.
30090 (decide_one_param_reduction): Likewise.
30091 (replace_removed_params_ssa_names): Likewise.
30092 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30093 * tree-ssa-copy.c (dump_copy_of): Likewise.
30094 (copy_prop_visit_cond_stmt): Likewise.
30095 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30096 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30097 (record_equivalences_from_stmt): Likewise.
30098 * tree-ssa-dse.c (compute_trims): Likewise.
30099 (delete_dead_call): Likewise.
30100 (delete_dead_assignment): Likewise.
30101 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30102 (forward_propagate_into_cond): Likewise.
30103 (pass_forwprop::execute): Likewise.
30104 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30105 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30107 (move_computations_worker): Likewise.
30108 (execute_sm): Likewise.
30109 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30110 (remove_exits_and_undefined_stmts): Likewise.
30111 (remove_redundant_iv_tests): Likewise.
30112 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30113 (adjust_iv_update_pos): Likewise.
30114 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30115 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30116 (value_replacement): Likewise.
30117 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30118 * tree-ssa-pre.c (print_pre_expr): Likewise.
30119 (get_representative_for): Likewise.
30120 (create_expression_by_pieces): Likewise.
30121 (insert_into_preds_of_block): Likewise.
30122 (eliminate_insert): Likewise.
30123 (eliminate_dom_walker::before_dom_children): Likewise.
30124 (eliminate): Likewise.
30125 (remove_dead_inserted_code): Likewise.
30126 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30127 * tree-ssa-reassoc.c (get_rank): Likewise.
30128 (eliminate_duplicate_pair): Likewise.
30129 (eliminate_plus_minus_pair): Likewise.
30130 (eliminate_not_pairs): Likewise.
30131 (undistribute_ops_list): Likewise.
30132 (eliminate_redundant_comparison): Likewise.
30133 (update_range_test): Likewise.
30134 (optimize_range_tests_var_bound): Likewise.
30135 (optimize_vec_cond_expr): Likewise.
30136 (rewrite_expr_tree): Likewise.
30137 (rewrite_expr_tree_parallel): Likewise.
30138 (linearize_expr): Likewise.
30139 (break_up_subtract): Likewise.
30140 (linearize_expr_tree): Likewise.
30141 (attempt_builtin_powi): Likewise.
30142 (attempt_builtin_copysign): Likewise.
30143 (transform_stmt_to_copy): Likewise.
30144 (transform_stmt_to_multiply): Likewise.
30145 (dump_ops_vector): Likewise.
30146 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30147 (print_scc): Likewise.
30148 (set_ssa_val_to): Likewise.
30149 (visit_reference_op_store): Likewise.
30150 (visit_use): Likewise.
30151 (sccvn_dom_walker::before_dom_children): Likewise.
30152 (run_scc_vn): Likewise.
30153 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30155 (expr_hash_elt::print): Likewise.
30156 (const_and_copies::pop_to_marker): Likewise.
30157 (const_and_copies::record_const_or_copy_raw): Likewise.
30158 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30159 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30160 (dump_predicates): Likewise.
30161 (find_uninit_use): Likewise.
30162 (warn_uninitialized_phi): Likewise.
30163 (pass_late_warn_uninitialized::execute): Likewise.
30164 * tree-ssa.c (verify_vssa): Likewise.
30165 (verify_ssa): Likewise.
30166 (maybe_optimize_var): Likewise.
30167 * tree-vrp.c (dump_value_range): Likewise.
30168 (dump_all_value_ranges): Likewise.
30169 (dump_asserts_for): Likewise.
30170 (register_edge_assert_for_2): Likewise.
30171 (vrp_visit_cond_stmt): Likewise.
30172 (vrp_visit_switch_stmt): Likewise.
30173 (vrp_visit_stmt): Likewise.
30174 (vrp_visit_phi_node): Likewise.
30175 (simplify_cond_using_ranges_1): Likewise.
30176 (fold_predicate_in): Likewise.
30177 (evrp_dom_walker::before_dom_children): Likewise.
30178 (evrp_dom_walker::push_value_range): Likewise.
30179 (evrp_dom_walker::pop_value_range): Likewise.
30180 (execute_early_vrp): Likewise.
30182 2017-05-16 Richard Biener <rguenther@suse.de>
30184 * dwarf2out.c (loc_list_from_tree_1): Do not create
30185 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30187 2017-05-16 Richard Biener <rguenther@suse.de>
30189 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30191 (note_variable_value_in_expr): If we resolved the decl ref
30192 do not push to the stack.
30194 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30196 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30197 operations in fast-math mode.
30198 (vaddq_f16): Likewise.
30199 (vmul_f16): Likewise.
30200 (vmulq_f16): Likewise.
30201 (vsub_f16): Likewise.
30202 (vsubq_f16): Likewise.
30203 * config/arm/neon.md (add<mode>3): New.
30205 (fma:<VH:mode>3): New. Also remove outdated comment.
30208 2017-05-16 Martin Liska <mliska@suse.cz>
30212 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30213 (odr_types_equivalent_p): Likewise.
30215 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30217 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30219 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30222 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30223 non-interunit SSE move alternatives with '?'.
30224 (zero-extendsidi peephole2): New peephole to skip intermediate
30225 general register in SSE zero-extend sequence.
30227 2017-05-15 Jeff Law <law@redhat.com>
30229 * reorg.c (relax_delay_slots): Create a new variable to hold
30230 the temporary target rather than clobbering TARGET_LABEL.
30232 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30233 missing argument to extract_bit_field call.
30234 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30236 2017-05-15 Martin Liska <mliska@suse.cz>
30239 * gcc.c (process_command): Do not allow empty argument of -o option.
30241 2017-05-15 Renlin Li <renlin.li@arm.com>
30243 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30244 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30245 * config/aarch64/constraints.md (Usf): Add long call check.
30246 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30247 (call_value): Likewise.
30248 (sibcall): Likewise.
30249 (sibcall_value): Likewise.
30251 (call_value_insn): New.
30252 (sibcall_insn): Update rtx pattern.
30253 (sibcall_value_insn): Likewise.
30254 (call_internal): Remove.
30255 (call_value_internal): Likewise.
30256 (sibcall_internal): Likewise.
30257 (sibcall_value_internal): Likewise.
30258 (call_reg): Likewise.
30259 (call_symbol): Likewise.
30260 (call_value_reg): Likewise.
30261 (call_value_symbol): Likewise.
30263 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30266 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30268 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30270 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30271 compatible with CCGOCmode and with CCZmode.
30273 2017-05-14 Martin Sebor <msebor@redhat.com>
30275 PR middle-end/77671
30276 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30277 (gimple_fold_builtin_snprintf): Same.
30278 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30279 (gimple_fold_builtin_snprintf): Same.
30280 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30281 of character types.
30282 (is_call_safe): New function.
30283 (try_substitute_return_value): Call it.
30284 (try_simplify_call): New function.
30285 (pass_sprintf_length::handle_gimple_call): Call it.
30287 2017-05-14 Martin Sebor <msebor@redhat.com>
30289 PR middle-end/80669
30290 * builtins.c (expand_builtin_stpncpy): Simplify.
30292 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30294 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30295 * config/i386/i386.h
30296 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30297 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30298 (struct machine_function): Add new members call_ms2sysv,
30299 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30300 (struct machine_frame_state): New fields sp_realigned and
30301 sp_realigned_offset.
30302 * config/i386/i386.c
30303 (enum xlogue_stub): New enum.
30304 (enum xlogue_stub_sets): New enum.
30305 (class xlogue_layout): New class.
30306 (struct ix86_frame): New fields stack_realign_allocate_offset,
30307 stack_realign_offset and outlined_save_offset. Modify comments to
30308 detail stack layout when using out-of-line stubs.
30309 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30310 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30311 -mcall-ms2sysv-xlogues.
30312 (stub_managed_regs): New static variable.
30313 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30314 registers managed by out-of-line stub.
30315 (disable_call_ms2sysv_xlogues): New function.
30316 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30317 m->call_ms2sysv when appropriate and compute frame layout for
30319 (sp_valid_at, fp_valid_at): New inline functions.
30320 (choose_basereg): New function.
30321 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30323 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30324 Use align parameter of choose_baseaddr to generated aligned SSE movs
30326 (pro_epilogue_adjust_stack): Modify to track
30327 machine_frame_state::sp_realigned.
30328 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30329 (ix86_nsaved_sseregs): Likewise.
30330 (ix86_emit_save_regs): Likewise.
30331 (ix86_emit_save_regs_using_mov): Likewise.
30332 (ix86_emit_save_sse_regs_using_mov): Likewise.
30333 (get_scratch_register_on_entry): Likewise.
30334 (gen_frame_set): New function.
30335 (gen_frame_load): Likewise.
30336 (gen_frame_store): Likewise.
30337 (emit_outlined_ms2sysv_save): Likewise.
30338 (emit_outlined_ms2sysv_restore): Likewise.
30339 (ix86_expand_prologue): Modify stack re-alignment code and call
30340 emit_outlined_ms2sysv_save when appropriate.
30341 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30342 parameter rtx_insn *insn, which allows the function to be used to only
30343 generate the notes.
30344 (ix86_expand_epilogue): Modify validity checks of frame and stack
30345 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30346 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30347 * config/i386/predicates.md
30348 (save_multiple): New predicate.
30349 (restore_multiple): Likewise.
30350 * config/i386/sse.md
30351 (save_multiple<mode>): New pattern.
30352 (save_multiple_realign<mode>): Likewise.
30353 (restore_multiple<mode>): Likewise.
30354 (restore_multiple_and_return<mode>): Likewise.
30355 (restore_multiple_leave_return<mode>): Likewise.
30356 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30358 2017-05-14 Julia Koval <julia.koval@intel.com>
30360 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30361 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30362 (__builtin_ia32_xsetbv): New builtins.
30363 * config/i386/i386.c (ix86_expand_special_args_builtin):
30365 (ix86_expand_builtin): Special expand for new intrinsics.
30366 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30367 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30368 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30370 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30372 * cfganal.c (inverted_post_order_compute): Change argument type
30374 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30375 * df-core.c (rest_of_handle_df_initialize): Adjust.
30376 (rest_of_handle_df_finish): Likewise.
30377 (df_analyze_1): Likewise.
30378 (df_analyze): Likewise.
30379 (loop_inverted_post_order_compute): Change argument to be a vec *.
30380 (df_analyze_loop): Adjust.
30381 (df_get_n_blocks): Likewise.
30382 (df_get_postorder): Likewise.
30383 * df.h (struct df_d): Change field to be a vec.
30384 * lcm.c (compute_laterin): Adjust.
30385 (compute_available): Likewise.
30386 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30387 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30388 * tree-ssa-pre.c (compute_antic): Likewise.
30390 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30392 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30393 (depth_first_search::depth_first_search): Change structure init
30394 function to this constructor.
30395 (depth_first_search::add_bb): Rename function to this member.
30396 (depth_first_search::execute): Likewise.
30397 (flow_dfs_compute_reverse_finish): Adjust.
30399 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30401 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30402 (longest_simple_path): Likewise.
30403 * shrink-wrap.c (spread_components): Likewise.
30404 (disqualify_problematic_components): Likewise.
30405 (emit_common_heads_for_components): Likewise.
30406 (emit_common_tails_for_components): Likewise.
30407 (insert_prologue_epilogue_for_components): Likewise.
30409 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30411 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30414 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30416 * df-core.c (df_set_blocks): Start using auto_bitmap.
30417 (df_compact_blocks): Likewise.
30418 * df-problems.c (df_rd_confluence_n): Likewise.
30419 * df-scan.c (df_insn_rescan_all): Likewise.
30420 (df_process_deferred_rescans): Likewise.
30421 (df_update_entry_block_defs): Likewise.
30422 (df_update_exit_block_uses): Likewise.
30423 (df_entry_block_bitmap_verify): Likewise.
30424 (df_exit_block_bitmap_verify): Likewise.
30425 (df_scan_verify): Likewise.
30426 * lra-constraints.c (lra_constraints): Likewise.
30427 (undo_optional_reloads): Likewise.
30428 (lra_undo_inheritance): Likewise.
30429 * lra-remat.c (calculate_gen_cands): Likewise.
30430 (do_remat): Likewise.
30431 * lra-spills.c (assign_spill_hard_regs): Likewise.
30432 (spill_pseudos): Likewise.
30433 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30434 (bitmap_set_subtract_values): Likewise.
30436 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30438 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30439 management with auto_bitmap.
30440 (fix_inter_tick): Likewise.
30441 (fix_recovery_deps): Likewise.
30442 * ira.c (add_store_equivs): Likewise.
30443 (find_moveable_pseudos): Likewise.
30444 (split_live_ranges_for_shrink_wrap): Likewise.
30445 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30446 (rtx_reuse_manager::seen_def_p): Likewise.
30447 (rtx_reuse_manager::set_seen_def): Likewise.
30448 * print-rtl.h (class rtx_reuse_manager): Likewise.
30450 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30452 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30454 (migrate_btr_def): Likewise.
30455 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30456 * df-core.c (loop_post_order_compute): Likewise.
30457 (loop_inverted_post_order_compute): Likewise.
30458 * hsa-common.h: Likewise.
30459 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30460 * init-regs.c (initialize_uninitialized_regs): Likewise.
30461 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30462 (inline_small_functions): Likewise.
30463 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30464 * ira.c (combine_and_move_insns): Likewise.
30465 (build_insn_chain): Likewise.
30466 * loop-invariant.c (find_invariants): Likewise.
30467 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30468 * predict.c (tree_predict_by_opcode): Likewise.
30469 (predict_paths_leading_to): Likewise.
30470 (predict_paths_leading_to_edge): Likewise.
30471 (estimate_loops_at_level): Likewise.
30472 (estimate_loops): Likewise.
30473 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30474 (spread_components): Likewise.
30475 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30476 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30477 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30478 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30479 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30480 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30481 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30482 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30483 (mark_threaded_blocks): Likewise.
30484 (thread_through_all_blocks): Likewise.
30485 * tree-ssa.c (verify_ssa): Likewise.
30486 (execute_update_addresses_taken): Likewise.
30487 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30489 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30491 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30493 (post_order_compute): Likewise.
30494 (inverted_post_order_compute): Likewise.
30495 (pre_and_rev_post_order_compute_fn): Likewise.
30497 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30499 * genrecog.c (int_set::int_set): Explicitly construct our
30500 auto_vec base class.
30501 * vec.h (auto_vec::auto_vec): New constructor.
30503 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30505 * bitmap.h (class auto_bitmap): New constructor taking
30506 bitmap_obstack * argument.
30508 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30510 * bitmap.h (class auto_bitmap): Change type of m_bits to
30511 bitmap_head, and adjust ctor / dtor and member operators.
30513 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30515 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30516 when returned register mode doesn't match original mode.
30518 2017-05-12 Jeff Law <law@redhat.com>
30519 Jakub Jelinek <jakub@redhat.com>
30521 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30522 we look for cc setter after the compare-elim changes.
30523 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30524 within the vector to match what compare-elim now expects.
30525 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30526 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30528 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30529 after the compare-elim changes.
30530 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30531 the vector to match what compare-elim now expects.
30532 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30533 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30534 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30535 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30536 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30538 * config/visium/visium.c (single_set_and_flags): Fix where
30539 we look for cc setter after the compare-elim changes.
30540 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30541 with the vector to match what compare-elim now expects.
30542 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30543 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30544 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30545 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30546 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30547 (neg<mode>2_insn_set_overflow): Likewise.
30549 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30551 PR middle-end/79794
30552 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30553 maybe_expand_insn call, set ops[0].target. If still set after call,
30554 set alt_rtl. Add extra arg to recursive calls.
30555 (extract_bit_field): Add alt_rtl argument. Pass to
30557 * expmed.h (extract_bit_field): Fix prototype.
30558 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30559 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30560 to extract_bit_field_calls.
30561 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30562 Pass alt_rtl to extract_bit_field calls.
30563 * calls.c (store_unaligned_arguments_into_psuedos)
30564 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30565 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30567 * optabs.h (struct expand_operand): Add target bitfield.
30569 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30571 * compare-elim.c (try_eliminate_compare): Canonicalize
30572 operation with embedded compare to
30573 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30574 (set (reg) (operation)].
30576 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30578 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30581 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30582 cost of adding a carry flag for ADC instruction.
30583 [case MINUS]: Ignore the cost of subtracting a carry flag
30584 for SBB instruction.
30586 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30588 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30590 * config/rs6000/bmiintrin.h: New file.
30591 * config/rs6000/bmi2intrin.h: New file.
30592 * config/rs6000/x86intrin.h: New file.
30594 2017-05-12 Jeff Law <law@redhat.com>
30596 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30599 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30601 PR middle-end/80707
30602 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30604 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30606 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30607 early expansion of vector divide builtins.
30608 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30609 builtins identified as having unsigned arguments.
30611 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30613 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30614 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30615 expansion of vector logical operations (and, andc, or, xor,
30618 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30620 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30621 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30623 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30625 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30626 early GIMPLE expansion of vector multiplies.
30628 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30630 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30631 TARGET_HAVE_MOVT conditional.
30632 (movt splitter): Likewise.
30634 2017-05-12 Richard Biener <rguenther@suse.de>
30636 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30637 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30638 Fold all stmts not inplace.
30640 2017-05-12 Richard Biener <rguenther@suse.de>
30642 PR tree-optimization/80713
30643 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30644 inserted_exprs bit for not removed stmts.
30646 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30648 PR middle-end/69921
30649 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30650 parallelized" attribute for parallelized OpenACC kernels.
30651 * omp-offload.c (execute_oacc_device_lower): Use it.
30653 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30654 Set "oacc kernels" attribute.
30655 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30656 parameter. Adjust all users.
30657 (oacc_fn_attrib_kernels_p): Remove function.
30658 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30659 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30660 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30661 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30662 assert "oacc kernels" attribute is set.
30664 2017-05-11 Carl Love <cel@us.ibm.com>
30666 * config/rs6000/rs6000-c: Add support for built-in functions
30667 vector unsigned char vec_popcnt (vector signed char)
30668 vector unsigned char vec_popcnt (vector unsigned char)
30669 vector unsigned short vec_popcnt (vector signed short)
30670 vector unsigned short vec_popcnt (vector unsigned short)
30671 vector unsigned int vec_popcnt (vector signed int)
30672 vector unsigned int vec_popcnt (vector unsigned int)
30673 vector unsigned long long vec_popcnt (vector signed long long)
30674 vector unsigned long long vec_popcnt (vector unsigned long long)
30675 vector signed long long vec_slo (vector signed long long,
30676 vector signed char)
30677 vector signed long long vec_slo (vector signed long long,
30678 vector unsigned char)
30679 vector unsigned long long vec_slo (vector unsigned long long,
30680 vector signed char)
30681 vector unsigned long long vec_slo (vector unsigned long long,
30682 vector unsigned char)
30683 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30684 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30685 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30686 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30687 * doc/extend.texi: Update the built-in documentation file for the
30688 new built-in functions.
30690 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30692 * attribs.h (sorted_attr_string): Move machine independent
30693 functions for target clone support from the i386 port to common
30694 code. Rename ix86_function_versions to common_function_versions.
30695 Rename make_name to make_unique_name.
30696 (common_function_versions): Likewise.
30697 (make_unique_name): Likewise.
30698 (make_dispatcher_decl): Likewise.
30699 (is_function_default_version): Likewise.
30700 * attribs.c (attr_strcmp): Likewise.
30701 (sorted_attr_string): Likewise.
30702 (common_function_versions): Likewise.
30703 (make_unique_name): Likewise.
30704 (make_dispatcher_decl): Likewise.
30705 (is_function_default_version): Likewise.
30706 * config/i386/i386.c (attr_strcmp): Likewise.
30707 (sorted_attr_string): Likewise.
30708 (ix86_function_versions): Likewise.
30709 (make_name): Likewise.
30710 (make_dispatcher_decl): Likewise.
30711 (is_function_default_version): Likewise.
30712 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30714 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30717 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30718 Account for direct move costs for vec_construct of integer
30721 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30724 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30725 (UNSPEC_STX_ATOMIC): Ditto.
30726 (loaddi_via_sse): New insn.
30727 (storedi_via_sse): Ditto.
30728 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30729 Update corresponding peephole2 patterns.
30730 (atomic_storedi_fpu): Ditto.
30732 2017-05-11 Julia Koval <julia.koval@intel.com>
30734 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30735 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30737 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30738 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30739 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30741 2017-05-11 Nathan Sidwell <nathan@acm.org>
30743 * graphite-poly.c: Include dumpfile.h.
30745 * dumpfle.h (dump_function): Declare here ...
30746 * tree-dump.h (dump_function): ... not here.
30747 * dumpfile.c: #include tree-cfg.h.
30748 (dump_function): Move here from ...
30749 * tree-dump.c (dump_function): ... here.
30750 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30751 * graphite-poly.c: Don't include tree-dump.h.
30752 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30753 * print-tree.c: Likewise.
30754 * stor-layout.c: Likewise.
30755 * tree-nested.c: Likewise.
30757 * dumpfile.c (dump_start): Use TDF_FLAGS.
30758 (dump_enable_all): Fix TDF_KIND check thinko.
30760 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30762 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30763 array entries to represent two legal parameterizations of the
30764 overloaded __builtin_cmpb function, as represented by the
30765 P6_OV_BUILTIN_CMPB constant.
30766 (altivec_resolve_overloaded_builtin): Add special case handling
30767 for the __builtin_cmpb function, as represented by the
30768 P6_OV_BUILTIN_CMPB constant.
30769 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30770 (BU_P6_64BIT_2): New macro.
30771 (BU_P6_OVERLOAD_2): New macro
30772 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30773 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30774 (CMPB): Add overload support to represent both 32-bit and 64-bit
30775 compare-bytes function.
30776 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30777 support for TARGET_CMPB.
30778 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30779 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30780 documentation of the __builtin_cmpb overloaded built-in function.
30782 2017-05-11 Richard Biener <rguenther@suse.de>
30784 PR tree-optimization/80705
30785 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30786 bases are not vectorizable.
30788 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30790 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30791 when counting register pressure.
30793 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30795 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30796 (iv_ca_more_deps): Renamed to ...
30797 (iv_ca_compare_deps): ... this.
30798 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30800 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30802 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30804 (determine_group_iv_costs): ... here.
30805 (find_inv_vars_cb): Record inv var if it's not recorded before.
30807 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30809 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30810 (get_shiftadd_cost): Ditto.
30812 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30814 * tree-ssa-address.c: Include header file.
30815 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30817 (add_to_parts): Refactor.
30818 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30819 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30822 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30824 PR tree-optimization/53090
30825 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30827 (extract_cond_operands): Detect condition with IV on both sides
30828 and return COMP_IV_EXPR_2.
30829 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30830 (rewrite_use_compare): Simplify by removing call to function
30831 extract_cond_operands.
30833 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30835 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30836 (extract_cond_operands): Detect condition comparing against non-
30837 invariant bound and return appropriate enum value.
30838 (find_interesting_uses_cond): Update use of extract_cond_operands.
30839 Handle its return value accordingly.
30840 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30842 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30844 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30845 nonlinear iv_use computation in loop invariant sensitive way.
30847 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30849 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30850 (find_iv_candidates): Call relate_compare_use_with_all_cands.
30852 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30854 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
30855 (dump_cand): Support iv_cand.inv_exprs.
30856 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
30858 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
30861 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30863 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
30865 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
30866 as local function. Include necessary header files.
30867 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
30869 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30871 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
30873 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30875 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
30876 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
30877 RSHIFT_EXPR and BIT_NOT_EXPR.
30879 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30881 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
30882 (adjust_setup_cost): New parameter supporting round up adjustment.
30883 (struct address_cost_data): Delete.
30884 (force_expr_to_var_cost): Don't bound cost with spill_cost.
30885 (split_address_cost, ptr_difference_cost): Delete.
30886 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
30887 (struct ainc_cost_data): New struct.
30888 (get_address_cost_ainc): New function.
30889 (get_address_cost, get_computation_cost): Reimplement.
30890 (determine_group_iv_cost_address): Record inv_expr for all uses of
30892 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
30893 (iv_ca_has_deps): Reimplemented to ...
30894 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
30896 (iv_ca_extend): Call iv_ca_more_deps.
30898 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30900 * tree-ssa-address.c (struct mem_address): Move to header file.
30901 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
30902 * tree-ssa-address.h (struct mem_address): Move from C file.
30903 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
30905 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30907 * tree-affine.h (aff_combination_type): New interface.
30908 (aff_combination_zero_p): Remove static.
30909 (aff_combination_const_p): New interface.
30910 (aff_combination_singleton_var_p): New interfaces.
30912 2017-05-11 Richard Biener <rguenther@suse.de>
30914 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30915 Skip unreachable blocks and destinations.
30916 (eliminate): Move stmt removal and fixup ...
30917 (fini_eliminate): ... here. Skip inserted exprs.
30918 (pass_pre::execute): Move fini_pre after fini_eliminate.
30919 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
30920 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
30921 PRE to get rid of dead code that has invalid SSA form and
30922 split critical edges again.
30924 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30926 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30928 2017-05-11 Richard Biener <rguenther@suse.de>
30930 * passes.c (execute_function_todo): Verify loops if they are
30931 said to be up-to-date.
30932 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
30933 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
30935 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
30938 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
30939 handle calling assemble_external ourself.
30942 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
30943 modes with zero size. Enhance comment.
30945 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30947 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
30948 built-ins for vec_xl and vec_xst with short and char pointer
30951 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30953 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
30954 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
30955 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
30956 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
30957 (_mm_maskz_min_round_ss): New intrinsics.
30958 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
30959 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
30960 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
30961 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
30962 (__builtin_ia32_minss_mask_round): New builtins.
30963 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30964 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30965 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
30967 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
30968 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
30970 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
30973 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30975 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
30976 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
30977 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
30978 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
30979 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
30980 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
30981 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
30982 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30983 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30984 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
30985 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
30986 (__builtin_ia32_mulss_mask_round): New builtins.
30987 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30988 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30989 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
30991 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
30992 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30994 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30997 2017-05-10 Julia Koval <julia.koval@intel.com>
30999 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31000 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31001 (_mm256_setr_m128i): New intrinsics.
31003 2017-05-10 Julia Koval <julia.koval@intel.com>
31005 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31006 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31007 (_mm_maskz_rcp14_ss): New intrinsics.
31008 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31009 (__builtin_ia32_rcp14ss_mask): New builtins.
31010 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31012 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31014 PR tree-optimization/51513
31015 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31016 (assert_unreachable_fallthru_edge_p): Use it.
31017 (group_case_labels_stmt): Likewise.
31018 * tree-cfg.h: Prototype it.
31019 * stmt.c: Include cfghooks.h and tree-cfg.h.
31020 (emit_case_dispatch_table) <gap_label>: New local variable.
31021 Use it to fill dispatch table gaps.
31022 Test for default_label before updating probabilities.
31023 (expand_case) <default_label>: Remove unneeded initialization.
31024 Test for unreachable default case statement and remove its edge.
31025 Set default_label accordingly.
31026 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31028 2017-05-10 Carl Love <cel@us.ibm.com>
31030 * config/rs6000/rs6000-c: Add support for built-in functions
31031 vector signed char vec_neg (vector signed char)
31032 vector signed short int vec_neg (vector short int)
31033 vector signed int vec_neg (vector signed int)
31034 vector signed long long vec_neg (vector signed long long)
31035 vector float vec_neg (vector float)
31036 vector double vec_neg (vector double)
31037 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31039 * config/rs6000/altivec.h: Add define for vec_neg
31040 * doc/extend.texi: Update the built-in documentation for the
31041 new built-in functions.
31043 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31045 PR tree-optimization/77644
31046 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31048 2017-05-10 Nathan Sidwell <nathan@acm.org>
31050 * dumpfile.h (TDI_lang_all): New.
31051 (TDF_KIND): New. Renumber others
31052 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31054 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31056 (get_dump_file_name): Adjust suffix generation.
31057 (dump_enable_all): Use TDF_KIND.
31058 * doc/invoke.texi (-fdump-lang-all): Document.
31060 * dumpfile.h: Tabify.
31062 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31065 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31066 Move member access before delete.
31068 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31070 * tree-inline.c (expand_call_inline): Split block at stmt
31073 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31076 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31077 are now unused after splitting mov{sf,sd}_hardfloat.
31078 (f32_lr2): Likewise.
31079 (f32_lm): Likewise.
31080 (f32_lm2): Likewise.
31081 (f32_li): Likewise.
31082 (f32_li2): Likewise.
31083 (f32_lv): Likewise.
31084 (f32_sr): Likewise.
31085 (f32_sr2): Likewise.
31086 (f32_sm): Likewise.
31087 (f32_sm2): Likewise.
31088 (f32_si): Likewise.
31089 (f32_si2): Likewise.
31090 (f32_sv): Likewise.
31091 (f32_dm): Likewise.
31092 (f32_vsx): Likewise.
31093 (f32_av): Likewise.
31094 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31095 For movsf, order stores so the VSX stores occur before the GPR
31096 store which encourages the register allocator to use a traditional
31097 FPR instead of a GPR. For movsd, order the stores so that the GPR
31098 store comes before the VSX stores to allow the power6 to work.
31099 This is due to the power6 not having a 32-bit integer store
31100 instruction from a FPR.
31101 (movsf_hardfloat): Likewise.
31102 (movsd_hardfloat): Likewise.
31104 2017-05-09 Martin Sebor <msebor@redhat.com>
31106 PR translation/80280
31107 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31110 PR translation/80280
31111 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31112 data member added in r247778.
31113 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31115 2017-05-09 Nathan Sidwell <nathan@acm.org>
31117 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31119 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31122 2017-05-09 Marek Polacek <polacek@redhat.com>
31124 * doc/invoke.texi: Fix typo.
31126 2017-05-09 Richard Biener <rguenther@suse.de>
31128 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31129 (vrp_val_is_min): Likewise.
31130 (set_value_range_to_value): Likewise.
31131 (set_value_range_to_nonnegative): Likewise.
31132 (gimple_assign_nonzero_p): Likewise.
31133 (gimple_stmt_nonzero_p): Likewise.
31134 (vrp_int_const_binop): Likewise. Remove unreachable case.
31135 (adjust_range_with_scev): Adjust comments.
31136 (compare_range_with_value): Likewise.
31137 (extract_range_from_phi_node): Likewise.
31138 (test_for_singularity): Likewise.
31140 2017-05-09 Richard Biener <rguenther@suse.de>
31142 * tree-vrp.c (get_single_symbol): Add assert that we don't
31143 get overflowed constants as invariant part.
31144 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31145 checks. Use wi::cmp instead of recursing for integer constants.
31146 (compare_values): Just ignore whether we assumed undefined
31147 overflow instead of failing the compare.
31148 (extract_range_for_var_from_comparison_expr): Add comment before the
31149 TREE_NO_WARNING sets.
31150 (test_for_singularity): Likewise.
31151 (extract_range_from_comparison): Do not disable optimization
31152 when we assumed undefined overflow.
31153 (extract_range_basic): Remove init of unused var.
31155 2017-05-09 Richard Biener <rguenther@suse.de>
31157 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31158 (extract_range_from_multiplicative_op_1): Adjust.
31159 (extract_range_from_binary_expr_1): Use int_const_binop.
31161 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31164 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31165 rs6000_store_data_bypass_p in seven define_bypass directives and
31166 in several comments.
31167 * config/rs6000/rs6000-protos.h: Add prototype for
31168 rs6000_store_data_bypass_p function.
31169 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31170 function implements slightly different (rs6000-specific) semantics
31171 than store_data_bypass_p, returning false rather than aborting
31172 with assertion error when arguments do not satisfy the
31173 requirements of store data bypass.
31174 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31175 rs6000_store_data_bypass_p.
31177 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31179 * config/xtensa/xtensa-protos.h
31180 (xtensa_initial_elimination_offset): New declaration.
31181 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31182 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31183 macro definition, add case for FRAME_POINTER_REGNUM when
31184 FRAME_GROWS_DOWNWARD.
31185 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31186 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31187 xtensa_initial_elimination_offset.
31189 2017-05-08 Nathan Sidwell <nathan@acm.org>
31191 * doc/invoke.texi: Alphabetize -fdump options.
31193 2017-05-08 Martin Sebor <msebor@redhat.com>
31195 PR translation/80280
31196 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31198 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31200 * target.def (compute_frame_layout): New optional target hook.
31201 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31202 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31203 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31205 * reload1.c (verify_initial_elim_offsets): Likewise.
31206 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31207 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31208 (arm_get_frame_offsets): Split up into this ...
31209 (arm_compute_frame_layout): ... and this function.
31211 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31213 * config/aarch64/constraints.md (Usa): New constraint.
31214 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31216 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31218 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31219 with_multilib_list after it has been checked.
31221 2017-05-08 Richard Biener <rguenther@suse.de>
31223 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31224 (bitmap_set_subtract_values): Likewise.
31226 2017-05-08 Richard Biener <rguenther@suse.de>
31228 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31229 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31231 (gimple_stmt_nonzero_warnv_p): Rename to ...
31232 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31234 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31235 (extract_range_basic): Adjust, do not disable propagation on
31236 strict overflow sensitive simplification.
31237 (vrp_visit_cond_stmt): Likewise.
31239 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31241 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31242 body size unconditionally.
31244 2017-05-07 Jeff Law <law@redhat.com>
31247 2017-05-06 Jeff Law <law@redhat.com>
31248 PR tree-optimization/78496
31249 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31252 PR tree-optimization/78496
31253 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31254 (simplify_stmt_using_ranges): Call it.
31255 (vrp_dom_walker::before_dom_children): Extract equivalences
31256 from an ASSERT_EXPR with an equality comparison against a
31259 2017-05-06 Jeff Law <law@redhat.com>
31261 PR tree-optimization/78496
31262 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31265 PR tree-optimization/78496
31266 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31267 (simplify_stmt_using_ranges): Call it.
31268 (vrp_dom_walker::before_dom_children): Extract equivalences
31269 from an ASSERT_EXPR with an equality comparison against a
31272 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31274 * lra-constraints.c (lra_copy_reg_equiv): New function.
31275 (split_reg): Use it to copy equivalence information from the
31276 original register to the spill register.
31278 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31280 PR rtl-optimization/75964
31281 * simplify-rtx.c (simplify_const_relational_operation): Remove
31282 invalid handling of comparisons of integer ABS.
31284 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31286 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31287 initialize to zero.
31288 (init_regs): Remove declaration.
31289 (function_arg_advance_32): Initialize error_p as boolean variable.
31291 2017-05-05 Nathan Sidwell <nathan@acm.org>
31293 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31294 lines. Use for (;;).
31296 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31298 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31299 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31300 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31301 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31302 VF=2 that require versioning.
31304 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31306 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31309 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31311 * diagnostic.h (diagnostic_override_option_index): Convert from
31312 macro to inline function.
31314 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31316 * diagnostic.c (last_module_changed_p): New function.
31317 (set_last_module): New function.
31318 (diagnostic_report_current_module): Convert macro usage to
31319 the above functions.
31320 * diagnostic.h (diagnostic_context::last_module): Strengthen
31321 from const line_map * to const line_map_ordinary *.
31322 (diagnostic_last_module_changed): Delete macro.
31323 (diagnostic_set_last_module): Delete macro.
31325 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31327 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31328 with diagnostic_report_diagnostic.
31329 (diagnostic_n_impl_richloc): Likewise.
31330 * diagnostic.h (report_diagnostic): Delete macro.
31331 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31332 with diagnostic_report_diagnostic.
31333 * substring-locations.c (format_warning_va): Likewise.
31335 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31337 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31338 save/restor of format_spec. Move option-printing code to...
31339 (print_option_information): ...this new function, and
31340 reimplement by simply printing to the pretty_printer,
31341 rather than appending to the format string.
31343 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31345 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31346 handling logic into...
31347 (update_effective_level_from_pragmas): ...this new function.
31349 2017-05-04 Andrew Waterman <andrew@sifive.com>
31351 * config/riscv/riscv.opt (mstrict-align): New option.
31352 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31353 (SLOW_UNALIGNED_ACCESS): Define.
31354 (riscv_slow_unaligned_access): Declare.
31355 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31357 (riscv_slow_unaligned_access): New variable.
31358 (rocket_tune_info): Set slow_unaligned_access to true.
31359 (optimize_size_tune_info): Set slow_unaligned_access to false.
31360 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31361 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31362 (riscv_option_override): Set riscv_slow_unaligned_access.
31363 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31365 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31367 * config/riscv/riscv.md: Unify indentation.
31369 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31374 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31376 (extendsi<mode>2): Add support for doing sign extension via
31377 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31378 don't have ISA 3.0 instructions.
31379 (extendsi<mode>2 splitter): Likewise.
31380 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31381 generate the normal insns since SImode can now go in vector
31382 registers. Disallow the special UNSPECs needed for previous
31383 machines to hide SImode being used. Add new insns
31384 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31385 (fix_trunc<mode>si2_stfiwx): Likewise.
31386 (fix_trunc<mode>si2_internal): Likewise.
31387 (fixuns_trunc<mode>si2): Likewise.
31388 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31389 (fctiw<u>z_<mode>_smallint): Likewise.
31390 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31391 of floating point to 32-bit integer from doing a direct move to
31392 the GPR registers to do a store.
31393 (fctiwz_<mode>): Break long line.
31395 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31397 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31398 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31399 (addr_list, addr_offset_valid_p): New.
31400 (split_address_groups): Check offset validity with above function.
31401 (gt-tree-ssa-loop-ivopts.h): Include header file.
31403 2017-05-05 Nathan Sidwell <nathan@acm.org>
31405 * config.gcc (arm*-*-*): Add missing 'fi'.
31407 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31409 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31410 included in -fopt-info does not matter.
31411 * doc/optinfo.texi (-fopt-info): Fix description of default
31412 behavour. Explicitly say order of options included in -fopt-info
31415 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31417 * config.gcc: Allow combinations of aprofile and rmprofile values for
31418 --with-multilib-list.
31419 * config/arm/t-multilib: New file.
31420 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31421 variables. Remove setting of ISA and floating-point ABI in
31422 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31423 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31424 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31425 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31427 * config/arm/t-rmprofile: Likewise except for the matches changes.
31428 * doc/install.texi (--with-multilib-list): Document the combination of
31429 aprofile and rmprofile values and warn about pitfalls in doing that.
31431 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31433 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31434 (movdi_aarch64): Likewise.
31436 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31438 PR tree-optimization/80632
31439 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31441 (build_arrays): Initialize it for virtual phis.
31442 (fix_phi_nodes): Use it for virtual phis.
31444 PR tree-optimization/80558
31445 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31446 [x, y] op z into [x op, y op z] for op & or | if conditions
31449 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31450 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31453 * config/arm/arm.md (use_literal_pool): Remove.
31454 (64-bit immediate split): No longer takes cost into consideration
31455 if arm_disable_literal_pool is enabled.
31456 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31457 used when arm_disable_literal_pool is enabled.
31458 (arm_max_const_double_inline_cost): Remove use of
31459 arm_disable_literal_pool.
31460 (push_minipool_fix): Add assert.
31461 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31462 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31463 (no_literal_pool_sf_immediate): New.
31465 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31467 PR tree-optimization/80613
31468 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31469 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31471 2017-05-05 Richard Biener <rguenther@suse.de>
31473 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31475 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31477 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31478 of this flag from insn conditions due to removal from r247495.
31480 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31482 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31484 (arm_early_store_addr_dep_ptr): Likewise.
31485 * config/arm/aarch-common-protos.h
31486 (arm_early_load_addr_dep_ptr): Add prototype.
31487 (arm_early_store_addr_dep_ptr): Likewise.
31488 * config/arm/cortex-a53.md: Add new bypasses.
31490 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31492 * tree.c (next_type_uid): Change type to unsigned.
31493 (type_hash_canon): Decrement back next_type_uid if
31494 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31495 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31498 2017-05-04 Martin Sebor <msebor@redhat.com>
31500 * builtins.c: Fix a trivial typo in a comment.
31502 PR middle-end/79234
31503 * builtins.c (check_sizes): Adjust to handle reading past the end.
31504 Avoid printing excessive upper bound of ranges. Use %E to print
31505 tree nodes instead of converting them to %wu.
31506 (expand_builtin_memchr): New function.
31507 (compute_dest_size): Rename...
31508 (compute_objsize): ...to this.
31509 (expand_builtin_memcpy): Adjust.
31510 (expand_builtin_mempcpy): Adjust.
31511 (expand_builtin_strcat): Adjust.
31512 (expand_builtin_strcpy): Adjust.
31513 (check_strncat_sizes): Adjust.
31514 (expand_builtin_strncat): Adjust.
31515 (expand_builtin_strncpy): Adjust and simplify.
31516 (expand_builtin_memset): Adjust.
31517 (expand_builtin_bzero): Adjust.
31518 (expand_builtin_memcmp): Adjust.
31519 (expand_builtin): Handle memcmp.
31520 (maybe_emit_chk_warning): Check strncat just once.
31522 2017-05-04 Martin Sebor <msebor@redhat.com>
31524 PR preprocessor/79214
31525 PR middle-end/79222
31526 PR middle-end/79223
31527 * builtins.c (check_sizes): Add inlining context and issue
31528 warnings even when -Wno-system-headers is set.
31529 (check_strncat_sizes): Same.
31530 (expand_builtin_strncat): Same.
31531 (expand_builtin_memmove): New function.
31532 (expand_builtin_stpncpy): Same.
31533 (expand_builtin): Handle memmove and stpncpy.
31535 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31537 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31538 which is not used any more.
31540 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31542 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31544 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31546 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31547 (cortexa53_tunings): Likewise.
31548 (cortexa57_tunings): Likewise.
31549 (cortexa72_tunings): Likewise.
31550 (cortexa73_tunings): Likewise.
31552 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31554 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31555 Set loop alignment to 8.
31557 2017-05-04 Martin Sebor <msebor@redhat.com>
31559 PR translation/80280
31560 * builtins.c (expand_builtin_object_size): Add missing quoting to
31561 %D and like directives.
31562 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31563 (hsa_type_for_tree_type): Same.
31564 (verify_function_arguments): Same.
31565 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31566 * varasm.c (get_section): Same.
31569 2017-05-04 Martin Sebor <msebor@redhat.com>
31571 PR translation/80280
31572 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31574 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31576 * config/aarch64/aarch64.c (generic_addrcost_table):
31577 Change HI/TI mode setting.
31579 2017-05-04 Martin Jambor <mjambor@suse.cz>
31581 PR tree-optimization/80622
31582 * tree-sra.c (comes_initialized_p): New function.
31583 (build_accesses_from_assign): Only set write lazily when
31584 comes_initialized_p is false.
31585 (analyze_access_subtree): Use comes_initialized_p.
31586 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31587 instead of testing for PARM_DECL.
31589 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31591 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31592 constraint on operand 0 to allow more general addressing modes.
31593 Adjust output template.
31594 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31596 * config/aarch64/aarch64-protos.h
31597 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31598 * config/aarch64/constraints.md (Dp): New address constraint.
31599 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31602 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31604 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31605 update use of estimate_ipcp_clone_size_and_time.
31606 (estimate_local_effects): Update use of
31607 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31608 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31609 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31610 Return nonspecialized time.
31612 2017-05-04 Richard Biener <rguenther@suse.de>
31614 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31615 for the last VUSE which def dominates the PHI. Directly call
31617 (get_continuation_for_phi_1): Remove.
31619 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31621 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31622 to explain the use of truncating division. Cap the number of
31623 iterations to the maximum given by nb_iterations_upper_bound,
31626 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31628 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31629 * configure: Regenerate.
31630 * config.in: Regenerate.
31631 * config/i386/driver-mingw32.c: new file.
31632 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31633 * config.host: Link driver-mingw32.o on MinGW host.
31634 * doc/install.texi: Document new --enable-mingw-wildcard configure
31637 2017-05-04 Marek Polacek <polacek@redhat.com>
31639 PR tree-optimization/80612
31640 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31642 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31643 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31645 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31646 (movt splitter): Likewise.
31647 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31648 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31649 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31650 block for Thumb-1 with MOVT.
31651 (thumb2_legitimate_address_p): Move code block ...
31652 (can_avoid_literal_pool_for_label_p): ... into this new function.
31653 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31655 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31656 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31657 "M-profile targets with the MOVT instruction".
31659 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31661 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31662 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31663 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31665 2017-05-04 Martin Liska <mliska@suse.cz>
31667 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31668 variable cond_code.
31670 2017-05-04 Richard Biener <rguenther@suse.de>
31672 * tree.c (array_at_struct_end_p): Handle arrays at struct
31673 end with flexarrays more conservatively. Refactor and treat
31674 arrays of arrays or aggregates more strict. Fix
31675 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31676 * tree.c (array_at_struct_end_p): Adjust prototype.
31677 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31678 * gimple-fold.c (get_range_strlen): Likewise.
31679 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31681 2017-05-04 Richard Biener <rguenther@suse.de>
31683 PR tree-optimization/31130
31684 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31686 (supports_overflow_infinity): Likewise.
31687 (is_negative_overflow_infinity): Likewise.
31688 (is_positive_overflow_infinity): Likewise.
31689 (is_overflow_infinity): Likewise.
31690 (stmt_overflow_infinity): Likewise.
31691 (overflow_infinity_range_p): Likewise.
31692 (usable_range_p): Remove as always returning true.
31693 (make_overflow_infinity): Remove.
31694 (negative_overflow_infinity): Likewise.
31695 (positive_overflow_infinity): Likewise.
31696 (avoid_overflow_infinity): Likewise.
31697 (set_value_range): Adjust accordingly.
31698 (set_value_range_to_nonnegative): Likewise, remove now unused
31699 overflow_infinity arg.
31700 (vrp_operand_equal_p): Adjust.
31701 (update_value_range): Likewise.
31702 (range_int_cst_singleton_p): Likewise.
31703 (operand_less_p): Likewise.
31704 (compare_values_warnv): Likewise.
31705 (extract_range_for_var_from_comparison_expr): Likewise.
31706 (vrp_int_const_binop): Likewise.
31707 (zero_nonzero_bits_from_vr): Likewise.
31708 (extract_range_from_multiplicative_op_1): Likewise.
31709 (extract_range_from_binary_expr_1): Likewise.
31710 (extract_range_from_unary_expr): Likewise.
31711 (extract_range_from_comparison): Likewise.
31712 (extract_range_basic): Likewise.
31713 (adjust_range_with_scev): Likewise.
31714 (compare_ranges): Likewise.
31715 (compare_range_with_value): Likewise.
31716 (dump_value_range): Likewise.
31717 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31719 (simplify_cond_using_ranges): Adjust.
31721 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31723 * brig-builtins.def: Added a builtin for class_f64.
31724 * builtin-types.def: Added a builtin type needed by class_f64.
31726 2017-05-03 Jason Merrill <jason@redhat.com>
31728 * timevar.def: Add TV_CONSTEXPR.
31730 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31732 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31734 2017-05-03 Martin Jambor <mjambor@suse.cz>
31736 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31737 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31738 * ipa-cp.c (ipcp_propagate_stage): Do not call
31739 ipa_update_after_lto_read.
31740 * ipa-inline.c (ipa_inline): Likewise.
31742 2017-05-03 Martin Jambor <mjambor@suse.cz>
31744 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31745 tag. Added a default constructor and a destructor.
31746 (ipa_edge_args_sum_t): New class;
31747 (ipa_edge_args_sum): Declare.
31748 (ipa_edge_args_vector): Remove declaration.
31749 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31750 (ipa_free_edge_args_substructures): Remove declaration.
31751 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31752 (ipa_edge_args_info_available_for_edge_p): Likewise.
31753 * ipa-prop.c (ipa_edge_args_vector): Removed.
31754 (edge_removal_hook_holder): Likewise.
31755 (edge_duplication_hook_holder): Likewise.
31756 (ipa_edge_args_sum): New variable.
31757 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31758 ipa_edge_args_vector.
31759 (ipa_free_edge_args_substructures): Likewise.
31760 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31761 ipa_edge_args_vector.
31762 (ipa_edge_removal_hook): Turned into method
31763 ipa_edge_args_sum_t::remove.
31764 (ipa_edge_duplication_hook): Turned into method
31765 ipa_edge_args_sum_t::duplicate.
31766 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31767 registering edge hooks.
31768 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31769 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31770 ipa_edge_args_sum instead of ipa_edge_args_vector.
31771 * ipa-profile.c (ipa_profile): Likewise.
31773 2017-05-03 Martin Jambor <mjambor@suse.cz>
31775 * symbol-summary.h (function_summary): New method exists.
31776 (function_summary::symtab_removal): Deallocate through release.
31777 (call_summary): New class.
31778 (gt_ggc_mx): New overload.
31779 (gt_pch_nx): Likewise.
31780 (gt_pch_nx): Likewise.
31782 2017-05-03 Jeff Law <law@redhat.com>
31784 PR tree-optimization/78496
31785 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31786 from simplify_cond_using_ranges. Split off code to walk
31787 backwards through casts into ...
31788 (simplify_cond_using_ranges_2): New function.
31789 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31790 (execute_vrp): After identifying jump threads, call
31791 simplify_cond_using_ranges_2.
31793 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31796 * ipa-inline.h (inline_summary): Add ctor.
31797 (create_ggc): Do not use ggc_cleared_alloc.
31799 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31800 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31802 * gcc.c (handle_braces): Support escaping in switch matching
31804 * doc/invoke.texi (Spec Files): Document it.
31805 Remove superfluous @code markup in items.
31807 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31809 * diagnostic-show-locus.c (struct column_range): New struct.
31810 (get_affected_columns): New function.
31811 (get_printed_columns): New function.
31812 (struct correction): New struct.
31813 (correction::ensure_capacity): New function.
31814 (correction::ensure_terminated): New function.
31815 (struct line_corrections): New struct.
31816 (line_corrections::~line_corrections): New dtor.
31817 (line_corrections::add_hint): New function.
31818 (layout::print_trailing_fixits): Reimplement in terms of the new
31820 (selftest::test_overlapped_fixit_printing): New function.
31821 (selftest::diagnostic_show_locus_c_tests): Call it.
31823 2017-05-03 Nathan Sidwell <nathan@acm.org>
31825 Canonicalize canonical type hashing
31826 * tree.h (type_hash_canon_hash): Declare.
31827 * tree.c (type_hash_list, attribute_hash_list): Move into
31828 type_hash_canon_hash.
31829 (build_type_attribute_qual_variant): Break out hash code calc into
31830 type_hash_canon_hash.
31831 (type_hash_canon_hash): New. Generic type hash computation.
31832 (build_range_type_1, build_array_type_1, build_function_type,
31833 build_method_type_directly, build_offset_type, build_complex_type,
31834 make_vector_type): Call it.
31836 2017-05-03 Richard Biener <rguenther@suse.de>
31838 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31839 When all DRs have unknown misaligned do not always peel
31840 when there is a store but apply the same costing model as if
31841 there were only loads.
31843 2017-05-03 Richard Biener <rguenther@suse.de>
31846 PR tree-optimization/80492
31847 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31848 compare_base_decls returning dont-know properly.
31850 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
31852 * config/arm/iterators.md (CCSI): New mode iterator.
31853 (arch): New mode attribute.
31854 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
31855 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
31856 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
31857 code iterator for success result mode.
31858 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
31859 the corresponding new insn generators.
31861 2017-05-03 Bin Cheng <bin.cheng@arm.com>
31864 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31865 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31867 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
31869 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
31870 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
31871 (DDR_A): Wrap DDR argument in brackets.
31872 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
31873 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
31874 (DDR_REVERSED_P): Likewise.
31876 2017-05-03 Jakub Jelinek <jakub@redhat.com>
31878 PR tree-optimization/79472
31879 * tree-switch-conversion.c (struct switch_conv_info): Add
31880 contiguous_range and default_case_nonstandard fields.
31881 (collect_switch_conv_info): Compute contiguous_range and
31882 default_case_nonstandard fields, don't clear final_bb if
31883 contiguous_range and only the default case doesn't have the required
31885 (check_all_empty_except_final): Set default_case_nonstandard instead
31886 of failing if contiguous_range and the default case doesn't have empty
31888 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
31889 and only the default case doesn't have the required constants. Skip
31891 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
31892 if default_case_nonstandard.
31893 (build_constructors): Build constant 1 just once. Assert that default
31894 values aren't inserted in between cases if contiguous_range. Skip
31896 (build_arrays): Skip virtual phis.
31897 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
31898 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
31899 Handle virtual phis.
31900 (gen_inbound_check): Handle default_case_nonstandard case.
31901 (process_switch): Adjust check_final_bb caller. Call
31902 gather_default_values with the first non-default case instead of
31903 default case if default_case_nonstandard.
31905 2017-05-02 Nathan Sidwell <nathan@acm.org>
31907 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
31908 check. Fix formatting.
31910 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
31912 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
31913 errors when comparing specialized and unspecialized times.
31915 2017-05-02 David Malcolm <dmalcolm@redhat.com>
31917 * diagnostic-show-locus.c
31918 (layout::should_print_annotation_line_p): Make private.
31919 (layout::print_annotation_line): Make private.
31920 (layout::annotation_line_showed_range_p): Make private.
31921 (layout::show_ruler): Make private.
31922 (layout::print_source_line): Make private. Pass in line and
31923 line_width, rather than calling location_get_source_line. Drop
31925 (layout::print_leading_fixits): New method.
31926 (layout::print_any_fixits): Rename to...
31927 (layout::print_trailing_fixits): ...this, and make private.
31928 Don't print newline fixits.
31929 (diagnostic_show_locus): Move logic for printing one row into...
31930 (layout::print_line): ...this new function. Move the
31931 location_get_source_line call and error-handling from
31932 print_source_line to here. Call print_leading_fixits, and rename
31933 print_any_fixits to print_trailing_fixits.
31934 (selftest::test_fixit_insert_containing_newline): Update now that
31935 newlines are partially supported.
31936 (selftest::test_fixit_insert_containing_newline_2): New test.
31937 (selftest::test_fixit_replace_containing_newline): Update comments.
31938 (selftest::diagnostic_show_locus_c_tests): Call the new test.
31939 * edit-context.c (class added_line): New class.
31940 (class edited_line): Describe newline handling in comment.
31941 (edited_line::actually_edited_p): New method.
31942 (edited_line::print_content): Delete redundant decl.
31943 (edited_line::m_predecessors): New field.
31944 (edited_file::print_content): Call edited_line::print_content.
31945 (edited_file::print_diff): Update to support newlines.
31946 (edited_file::print_diff_hunk): Likewise.
31947 (edited_file::print_run_of_changed_lines): New function.
31948 (edited_file::print_diff_line): Convert to...
31949 (print_diff_line): ...this.
31950 (edited_file::get_effective_line_count): New function.
31951 (edited_line::edited_line): Initialize new field m_predecessors.
31952 (edited_line::~edited_line): Clean up m_predecessors.
31953 (edited_line::apply_fixit): Handle newlines.
31954 (edited_line::get_effective_line_count): New function.
31955 (edited_line::print_content): New function.
31956 (edited_line::print_diff_lines): New function.
31957 (selftest::test_applying_fixits_insert_containing_newline): New
31959 (selftest::test_applying_fixits_replace_containing_newline): New
31961 (selftest::insert_line): New function.
31962 (selftest::test_applying_fixits_multiple_lines): Add example of
31964 (selftest::edit_context_c_tests): Call the new tests.
31966 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31968 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
31969 parameter cand. Update dump information.
31970 (get_computation_cost): Update uses.
31972 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31974 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
31975 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
31976 (get_computation_at, rewrite_use_address): Update use of
31977 get_computation_aff.
31979 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31981 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
31982 (get_computation): Delete.
31983 (get_computation_cost): Implement like get_computation_cost_at.
31984 Use get_computation_at.
31985 (get_computation_cost_at): Delete.
31986 (rewrite_use_nonlinear_expr): Use get_computation_at.
31987 (rewrite_use_compare, remove_unused_ivs): Ditto.
31989 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31991 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
31993 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31995 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
31996 (ivopts_global_cost_for_size): Rename parameter and update uses.
31997 (iv_ca_recount_cost): Update uses.
31998 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
31999 candidates seperately in n_invs and n_cands.
32000 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32002 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32004 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32005 (find_inv_vars_cb): New.
32006 (find_depends): Renamed to ...
32007 (find_inv_vars): ... this.
32008 (add_candidate_1, force_var_cost): Call find_inv_vars.
32009 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32011 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32013 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32014 inv_vars. Add inv_exprs.
32015 (struct iv_cand): Rename depends_on to inv_vars.
32016 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32017 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32018 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32019 (dump_cand): Dump inv_vars.
32020 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32021 (record_invariant, find_depends, add_candidate_1): Ditto.
32022 (set_group_iv_cost, force_var_cost): Ditto.
32023 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32024 (get_computation_cost_at, get_computation_cost): Ditto.
32025 (determine_group_iv_cost_generic): Ditto.
32026 (determine_group_iv_cost_address): Ditto.
32027 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32028 (determine_group_iv_costs): Ditto.
32029 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32030 (iv_ca_set_remove_invariants): Renamed to ...
32031 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32032 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32033 (iv_ca_set_add_invariants): Renamed to ...
32034 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32035 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32036 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32037 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32038 (create_new_ivs): Remove useless dump.
32040 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32042 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32044 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32045 (iv_ca_set_no_cp, create_new_iv): Ditto.
32047 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32049 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32051 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32053 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32054 function tree_check2.
32056 2017-05-02 Martin Liska <mliska@suse.cz>
32058 * doc/gcov.texi: Add missing preposition.
32059 * gcov.c (function_info::function_info): Properly fill up
32060 all member variables.
32062 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32064 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32066 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32068 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32070 2017-05-02 Martin Liska <mliska@suse.cz>
32073 * lto-streamer-in.c (lto_read_tree_1): Remove
32074 LTO_STREAMER_DEBUG.
32075 * lto-streamer.c (struct tree_hash_entry): Likewise.
32076 (struct tree_entry_hasher): Likewise.
32077 (tree_entry_hasher::hash): Likewise.
32078 (tree_entry_hasher::equal): Likewise.
32079 (lto_streamer_init): Likewise.
32080 (lto_orig_address_map): Likewise.
32081 (lto_orig_address_get): Likewise.
32082 (lto_orig_address_remove): Likewise.
32083 * lto-streamer.h: Likewise.
32084 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32085 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32087 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32089 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32090 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32091 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32092 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32093 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32094 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32095 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32096 (mm_maskz_sub_ss): New intrinsics.
32097 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32098 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32099 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32100 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32101 (__builtin_ia32_subss_mask_round): New builtins.
32102 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32103 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32104 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32106 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32107 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32109 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32112 2017-05-02 Martin Jambor <mjambor@suse.cz>
32114 PR tree-optimization/78687
32115 * tree-sra.c (access): New field parent.
32116 (process_subtree_disqualification): New function.
32117 (disqualify_candidate): Call it.
32118 (build_accesses_from_assign): Reset write flag if creating an
32120 (build_access_subtree): Fill in parent field and also prpagate
32121 down grp_write flag.
32122 (create_artificial_child_access): New parameter set_grp_write, set
32123 grp_write to its value.
32124 (propagate_subaccesses_across_link): Also propagate grp_write flag
32126 (propagate_all_subaccesses): Push the closest parent back to work
32127 queue if add_access_to_work_queue returned true.
32129 2017-05-02 Richard Biener <rguenther@suse.de>
32131 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32132 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32133 -fstrict-overflow documentation.
32134 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32135 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32136 flag_strict_overflow.
32137 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32138 * lto-opts.c (lto_write_options): Do not stream it.
32139 * lto-wrapper.c (merge_and_complain): Do not handle it.
32140 * opts.c (default_options_table): Do not set -fstrict-overflow.
32141 (finish_options): Likewise do not clear it when sanitizing.
32142 * simplify-rtx.c (simplify_const_relational_operation): Do not
32143 test flag_strict_overflow.
32145 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32147 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32148 using enabled attribute.
32149 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32150 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32151 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32152 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32153 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32154 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32155 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32156 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32157 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32158 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32160 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32162 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32164 2017-05-02 Richard Biener <rguenther@suse.de>
32166 PR tree-optimization/80591
32168 2017-04-10 Richard Biener <rguenther@suse.de>
32170 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32173 2017-05-02 Richard Biener <rguenther@suse.de>
32175 PR tree-optimization/80549
32176 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32177 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32178 headers if they do not have a preheader.
32180 2017-05-02 Martin Liska <mliska@suse.cz>
32183 * common.opt: Fix typo.
32184 * doc/invoke.texi: Likewise.
32186 2017-05-01 Jan Beulich <jbeulich@suse.com>
32188 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32189 swapping, add (x,x,m,x,n) alternative.
32191 2017-05-01 Nathan Sidwell <nathan@acm.org>
32193 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32194 unnecessary unadjusted_alignment check.
32196 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32199 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32201 * gimplify.c (gimplify_cilk_detach): New function.
32202 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32203 * tree-core.h: Document EXPR_CILK_SPAWN.
32204 * tree.h (EXPR_CILK_SPAWN): Define.
32206 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32208 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32209 to use new fixit_hint representation, using the "replace" logic.
32210 (get_line_span_for_fixit_hint): Likewise.
32211 (layout::print_any_fixits): Likewise.
32212 (selftest::test_one_liner_many_fixits): Rename to...
32213 (selftest::test_one_liner_many_fixits_1): ...this, and update
32214 comment and expected output to reflect that the multiple fix-it
32215 hints are now consolidated into one insertion.
32216 (selftest::test_one_liner_many_fixits_2): New test.
32217 (selftest::test_diagnostic_show_locus_one_liner): Update for
32219 (selftest::test_fixit_consolidation): Update for fix-it API
32221 * diagnostic.c (print_parseable_fixits): Likewise.
32222 * edit-context.c (edited_line::m_line_events): Convert from
32223 auto_vec <line_event *> to auto_vec <line_event>.
32224 (class line_event): Convert from abstract base class to a concrete
32225 class, taking over the role of replace_event.
32226 (class insert_event): Delete.
32227 (class replace_event): Rename to class line_event. Convert to
32229 (edit_context::add_fixits): Reimplement.
32230 (edit_context::apply_insert): Delete.
32231 (edit_context::apply_replace): Rename to...
32232 (edit_context::apply_fixit): ...this. Convert to half-open range.
32233 (edited_file::apply_insert): Delete.
32234 (edited_file::apply_replace): Rename to...
32235 (edited_file::apply_fixit): ...this.
32236 (edited_line::~edited_line): Drop deletion of events.
32237 (edited_line::apply_insert): Delete.
32238 (edited_line::apply_replace): Rename to...
32239 (edited_line::apply_fixit): ...this. Convert to half-open range.
32240 Update for change to type of m_line_events.
32241 * edit-context.h (edit_context::apply_insert): Delete.
32242 (edit_context::apply_replace): Rename to...
32243 (edit_context::apply_fixit): ...this.
32245 2017-05-01 Martin Sebor <msebor@redhat.com>
32247 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32250 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32253 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32254 __get_cpuid_max returns 0.
32255 (__get_cpuid_count): Ditto.
32257 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32259 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32260 replacement expression is another instance of one of its arguments.
32262 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32265 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32266 check for stack push/pop autoinc.
32267 * config/i386/i386.c (ix86_agi_dependent): Return false
32268 if the only reason why modified_in_p returned true is that
32269 addr is SP based and set_insn is a push or pop.
32271 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32273 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32276 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32279 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32280 (account_size_time): Use two predicates - exec_pred and
32282 (evaluate_conditions_for_known_args): Compute both clause and
32284 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32285 (inline_summary_t::duplicate): Update.
32286 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32288 (compute_inline_parameters): Likewise.
32289 (estimate_edge_size_and_time): Update caluclation of time.
32290 (estimate_node_size_and_time): Compute both time and nonspecialized
32292 (estimate_ipcp_clone_size_and_time): Update.
32293 (inline_merge_summary): Update.
32294 (do_estimate_edge_time): Update.
32295 (do_estimate_edge_size): Update.
32296 (do_estimate_edge_hints): Update.
32297 (inline_read_section, inline_write_summary): Stream both new predicates.
32298 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32300 (compute_inlined_call_time): Cleanup.
32301 (big_speedup_p): Update.
32302 (edge_badness): Update.
32303 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32304 (size_time_entry): Replace predicate by exec_predicate and
32305 nonconst_predicate.
32306 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32307 (estimate_edge_time): Return also nonspec_time.
32308 (reset_edge_growth_cache): Update.
32310 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32312 PR rtl-optimization/80491
32313 * ifcvt.c (noce_process_if_block): When looking for x setter
32314 with missing else_bb, don't check only the insn right before
32315 cond_earliest, but look for the last insn that x is modified in
32316 within the same bb.
32318 PR rtl-optimization/80491
32319 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32321 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32323 PR tree-optimization/80487
32324 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32326 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32328 PR tree-optimization/79697
32329 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32330 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32331 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32333 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32334 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32336 2017-04-28 Martin Sebor <msebor@redhat.com>
32338 PR tree-optimization/80523
32339 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32340 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32342 (maybe_warn, format_directive, parse_directive): Use new functions.
32343 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32345 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32347 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32349 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32351 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32352 target_header_dir): Set correctly.
32353 * configure: Regenerated.
32354 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32355 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32356 instead of SYSTEM_HEADER_DIR.
32358 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32360 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32361 (estimate_local_effects): Likewise.
32362 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32363 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32364 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32365 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32366 do_estimate_edge_time, estimate_edge_time): Likewise.
32367 * ipa-inline-analysis.c (estimate_node_size_and_time,
32368 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32369 (estimate_time_after_inlining): Remove.
32371 2017-04-28 Martin Liska <mliska@suse.cz>
32373 * doc/gcov.texi: Enhance documentation of gcov.
32375 2017-04-28 Martin Liska <mliska@suse.cz>
32377 * doc/gcov.texi: Sort options in alphabetic order.
32378 * doc/gcov-dump.texi: Likewise.
32379 * doc/gcov-tool.texi: Likewise.
32380 * gcov.c (print_usage): Likewise.
32381 * gcov-dump.c (print_usage): Likewise.
32382 * gcov-tool.c (print_merge_usage_message): Likewise.
32383 (print_rewrite_usage_message): Likewise.
32384 (print_overlap_usage_message): Likewise.
32386 2017-04-28 Martin Liska <mliska@suse.cz>
32388 PR gcov-profile/53915
32389 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32391 2017-04-28 Martin Liska <mliska@suse.cz>
32393 PR gcov-profile/79891
32394 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32395 is marked by compiler as living on a line.
32396 (get_cycles_count): Remove usage of the union.
32397 (output_intermediate_file): Likewise.
32398 (find_source): Fix GNU coding style.
32399 (accumulate_line_counts): Remove old non-all block mode.
32400 (output_lines): Remove usage of the union.
32401 * profile.c (output_location): Include all BBs, even if
32402 belonging to a same line (and file) as a previous BB.
32404 2017-04-28 Martin Liska <mliska@suse.cz>
32406 * gcov.c (process_args): Handle new argument 'w'.
32407 (read_graph_file): Assign ID to BBs.
32408 (output_branch_count): Display BB # if verbose flag is set.
32409 (output_lines): Likewise for arcs.
32410 (print_usage): Add '--verbose' option help.
32411 * doc/gcov.texi: Document --verbose (-w) option.
32413 2017-04-28 Martin Liska <mliska@suse.cz>
32415 * gcov.c (struct block_location_info): New struct.
32416 (process_file): Fill up the new structure.
32417 (read_graph_file): Replace usage of encoding by the newly added
32419 (add_line_counts): Likewise.
32420 (accumulate_line_counts): Remove usage of the union.
32421 (function_info::function_info): New function.
32422 (function_info::~function_info): Likewise.
32423 (process_file): Call delete instead of release_function.
32424 (release_function): Release the function.
32425 (release_structures): Call delete instead of release_function.
32426 (solve_flow_graph): Replace usage of num_blocks.
32427 (find_exception_blocks): Likewise.
32428 (output_lines): Fix GNU coding style.
32430 2017-04-28 Martin Liska <mliska@suse.cz>
32433 * coverage.c (coverage_remove_note_file): New function.
32434 * coverage.h: Declare the function.
32435 * toplev.c (finalize): Clean if an error has been seen.
32437 2017-04-28 Martin Liska <mliska@suse.cz>
32439 PR gcov-profile/80031
32440 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32441 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32442 * gcov.c (read_graph_file): Read just number of blocks.
32443 * profile.c (branch_prob): Do not stream 0 flags per a basic
32446 2017-04-28 Martin Liska <mliska@suse.cz>
32448 * gcov-dump.c (tag_*): Add new argument to declarations.
32449 (dump_gcov_file): Likewise.
32450 (tag_blocks): Add and use new argument depth.
32451 (tag_arcs): Likewise.
32452 (tag_lines): Likewise.
32453 (tag_counters): Likewise.
32454 (tag_summary): Likewise.
32455 (dump_working_sets): Use depth to do a proper indentation.
32457 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32460 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32461 * symtab.c (symtab_node::debug_symtab): Move definition here.
32463 2017-04-28 Richard Biener <rguenther@suse.de>
32465 * lto-streamer.h (LTO_major_version): Bump to 7.
32467 2017-04-28 Richard Biener <rguenther@suse.de>
32469 * tree-vrp.c (assert_info): New struct.
32470 (add_assert_info): New helper.
32471 (register_edge_assert_for_2): Refactor to add asserts to a vector
32473 (register_edge_assert_for_1): Likewise.
32474 (register_edge_assert_for): Likewise.
32475 (finish_register_edge_assert_for): New helper actually registering
32476 asserts where live on edge.
32477 (find_conditional_asserts): Adjust.
32478 (find_switch_asserts): Likewise.
32479 (evrp_dom_walker::try_find_new_range): Generalize.
32480 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32482 2017-04-27 Marek Polacek <polacek@redhat.com>
32485 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32486 arg10 and arg11 to itype.
32488 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32490 * doc/extend.texi (Object Size Checking): Improve grammar.
32492 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32495 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32496 that the logic for permitting reciprocal estimates matches that
32499 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32502 * tree.c (type_cache_hasher::equal): Only compare
32503 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32504 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32505 non-aggregate element types.
32506 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32507 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32509 2017-04-27 Richard Biener <rguenther@suse.de>
32511 PR middle-end/80533
32512 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32513 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32514 keeping a reference to a trailing array.
32516 2017-04-27 Richard Biener <rguenther@suse.de>
32518 PR middle-end/80539
32519 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32520 being in loop-closed SSA form conservatively.
32521 (chrec_fold_multiply_poly_poly): Likewise.
32523 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32525 PR middle-end/79665
32526 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32527 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32529 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32532 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32533 (aarch64_function_arg_alignment): Return unsigned int again, but still
32534 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32535 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32536 Don't emit -Wpsabi note.
32537 (aarch64_function_arg_boundary): Likewise.
32538 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32541 2017-04-26 Nathan Sidwell <nathan@acm.org>
32543 * tree.h (crc32_unsigned_n): Declare.
32544 (crc32_unsigned, crc32_unsigned): Make inline.
32545 * tree.c (crc32_unsigned_bits): Replace with ...
32546 (crc32_unsigned_n): ... this.
32547 (crc32_unsigned, crc32_byte): Remove.
32548 (crc32_string): Remove unnecessary braces.
32550 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32552 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32553 * ipa-inline-analysis.c (MAX_TIME): Remove.
32554 (account_size_time): Use sreal for time.
32555 (dump_inline_summary): Update.
32556 (estimate_function_body_sizes): Update.
32557 (estimate_edge_size_and_time): Update.
32558 (estimate_calls_size_and_time): Update.
32559 (estimate_node_size_and_time): Update.
32560 (inline_merge_summary): Update.
32561 (inline_update_overall_summary): Update.
32562 (estimate_time_after_inlining): Update.
32563 (inline_read_section): Update.
32564 (inline_write_summary): Update.
32565 * ipa-inline.c (compute_uninlined_call_time): Update.
32566 (compute_inlined_call_time): Update.
32567 (recursive_inlining): Update.
32568 (inline_small_functions): Update.
32569 (dump_overall_stats): Update.
32570 * ipa-inline.h: Include sreal.h.
32571 (size_time_entry): Turn time to sreal.
32572 (inline_summary): Turn self_time nad time to sreal.
32574 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32576 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32578 (sreal::stream_out, sreal::stream_in): New.
32579 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32581 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32583 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32586 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32589 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32591 (dimode_scalar_chain::compute_convert_gain): Ditto.
32592 (dimode_scalar_chain::make_vector_copies): Ditto.
32593 (dimode_scalar_chain::convert_reg): Ditto.
32594 (dimode_scalar_chain::convert_insn): Ditto.
32595 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32596 (VI248_AVX512BW_1): New mode iterator.
32597 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32598 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32601 2017-04-25 Martin Sebor <msebor@redhat.com>
32603 PR tree-optimization/80497
32604 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32605 constants are representable in HOST_WIDE_INT.
32606 (parse_directive): Ditto.
32608 2017-04-25 Martin Sebor <msebor@redhat.com>
32611 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32612 (new_zero_array): Adjust signature.
32613 (dom_info::dom_init): Used unsigned rather that size_t.
32614 (dom_info::dom_info): Same.
32616 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32617 Jakub Jelinek <jakub@redhat.com>
32620 * config/arm/arm.c: Include gimple.h.
32621 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32622 returns negative, increment ncrn only if it returned positive.
32623 (arm_needs_doubleword_align): Return int instead of bool,
32624 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32625 members, but if there is any such non-FIELD_DECL
32626 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32627 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32628 returns negative, increment nregs only if it returned positive.
32629 (arm_setup_incoming_varargs): Likewise.
32630 (arm_function_arg_boundary): Emit -Wpsabi note if
32631 arm_needs_doubleword_align returns negative, return
32632 DOUBLEWORD_ALIGNMENT only if it returned positive.
32634 2017-04-25 Marek Polacek <polacek@redhat.com>
32637 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32638 first argument to type.
32640 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32643 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32644 type checks to test for compatibility instead of equality.
32646 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32647 Jakub Jelinek <jakub@redhat.com>
32650 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32652 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32653 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32654 the alignment computation, but return their maximum in warn_alignment.
32655 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32656 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32658 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32659 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32662 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32664 * config/arc/simdext.md (dmpyh): Fix typo.
32666 2017-04-25 Richard Biener <rguenther@suse.de>
32668 PR tree-optimization/80492
32669 * alias.c (compare_base_decls): Handle registers with asm
32670 specification conservatively.
32671 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32672 compare_base_decls returning dont-know properly.
32674 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32676 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32677 (legitimate_offset_address_p): New function.
32678 (arc_legitimate_address_p): Use above function.
32680 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32682 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32684 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32686 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32687 ACCH registers whenever they are available.
32689 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32691 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32692 double regs fix when not used.
32694 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32696 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32698 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32699 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32701 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32703 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32704 when emitting short ADD instructions.
32706 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32708 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32710 (cmpsi_cc_c_insn): Likewise.
32711 (cbranchsi4_scratch): Compute proper instruction length using
32712 compact_hreg_operand.
32713 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32715 2017-04-25 Richard Biener <rguenther@suse.de>
32717 PR middle-end/80509
32718 * passes.c (pass_manager::pass_manager): Initialize
32719 m_name_to_pass_map.
32721 2017-04-25 Richard Biener <rguenther@suse.de>
32723 PR tree-optimization/79201
32724 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32726 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32729 * config/s390/vector.md: Split MEM->GPR vector moves for
32730 non-s_operand addresses.
32732 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32735 * config/s390/predicates.md (reload_const_wide_int_operand): New
32737 * config/s390/s390.md ("movti"): Remove d/P alternative.
32738 ("movti_bigconst"): New pattern definition.
32740 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32743 * s390-protos.h (s390_expand_cs_hqi): Removed.
32744 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32745 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32746 modes as well as CCZ1mode and CCZmode.
32747 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32748 signature of s390_emit_compare_and_swap.
32749 (s390_expand_cs_hqi): Likewise, make static.
32750 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32751 compare-and-swap, in some cases.
32752 (s390_expand_cs): Wrapper function.
32753 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32755 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32756 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32757 patterns for small and large integers. Forbid symref memory operands.
32758 Move expander to s390.c. Require cc register.
32759 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32760 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32761 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32762 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32763 symref memory operands. Remove CC mode and call s390_match_ccmode
32765 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32767 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32769 * config/s390/s390.md (define_peephole2): New peephole to help
32770 combining the load-and-test pattern with volatile memory.
32772 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32774 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32775 with CCZmode for TARGET_Z196.
32777 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32779 PR rtl-optimization/80501
32780 * combine.c (make_compound_operation_int): Set subreg_code to SET
32781 even for AND with mask of the sign bit of mode.
32783 PR rtl-optimization/80500
32784 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32785 sum's initial value.
32787 2017-04-25 Julian Brown <julian@codesourcery.com>
32788 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32790 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32792 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32794 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32796 2017-04-25 Julian Brown <julian@codesourcery.com>
32797 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32799 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32800 (thunderx2t99_sha): New Reservation.
32802 2017-04-25 Julian Brown <julian@codesourcery.com>
32803 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32805 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32806 type for 1-element load.
32808 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32810 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32812 2017-04-24 Martin Jambor <mjambor@suse.cz>
32814 PR tree-optimization/80293
32815 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32816 char arrays not totally scalarizable if it is false.
32817 (analyze_all_variable_accesses): Pass correct value in the new
32818 parameter. Add a statistics counter.
32820 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32822 PR middle-end/79931
32823 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32825 2017-04-24 Richard Biener <rguenther@suse.de>
32827 PR tree-optimization/80494
32828 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32829 out for complex types.
32831 2017-04-24 Richard Biener <rguenther@suse.de>
32833 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32834 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32835 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32836 (DFS): Adjust and never fail.
32837 (sccvn_dom_walker::fail): Remove.
32838 (sccvn_dom_walker::before_dom_children): Adjust.
32839 (run_scc_vn): Likewise and never fail.
32840 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32841 (pass_fre::execute): Likewise.
32843 2017-04-24 Richard Biener <rguenther@suse.de>
32845 PR tree-optimization/79725
32846 * tree-ssa-sink.c (statement_sink_location): Return whether
32847 failure reason was zero uses. Move that check later.
32848 (sink_code_in_bb): Deal with zero uses by removing the stmt
32851 2017-04-24 Richard Biener <rguenther@suse.de>
32854 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
32855 pointer-based references.
32857 2017-04-24 Richard Biener <rguenther@suse.de>
32860 * pass_manager.h (pass_manager::operator new): Remove.
32861 (pass_manager::operator delete): Likewise.
32862 * passes.c (pass_manager::operator new): Remove.
32863 (pass_manager::operator delete): Likewise.
32864 (pass_manager::pass_manager): Zero individual pass members.
32866 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
32869 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
32870 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
32871 Check "XEXP (src, 1)" operand here.
32872 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
32873 Check "XEXP (src, 1)" operand here.
32874 (dimode_scalar_chain::make_vector_copies): Detect count register
32875 of a shift instruction. Zero extend count register from QImode
32876 to DImode to satisfy vector shift pattern count operand predicate.
32877 Substitute vector shift count operand with a DImode copy.
32878 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
32881 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
32883 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
32884 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32885 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
32886 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32887 (UNSPEC_NOREX_MEM): Remove definition.
32889 2017-04-21 Richard Biener <rguenther@suse.de>
32891 PR tree-optimization/79547
32892 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
32893 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
32894 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
32895 without any constraints.
32897 2017-04-21 Richard Biener <rguenther@suse.de>
32899 PR tree-optimization/78847
32900 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
32902 2017-04-21 Richard Biener <rguenther@suse.de>
32904 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
32905 (build_distinct_type_copy): Likewise.
32906 (build_variant_type_copy): Likewise.
32907 * tree.c (build_qualified_type): Pass down mem-stat info.
32908 (build_distinct_type_copy): Likewise.
32909 (build_variant_type_copy): Likewise.
32911 2017-04-21 Richard Biener <rguenther@suse.de>
32913 PR tree-optimization/80237
32914 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
32916 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
32917 for a simplified result.
32919 2016-04-21 Richard Biener <rguenther@suse.de>
32921 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
32922 sth as strict as a simple_iv but a chrec without symbols and an
32923 operand defined in the loop we are peeling (and not some subloop).
32924 (propagate_constants_for_unrolling): Propagate all constants.
32926 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32929 * config/i386/i386.c (print_reg): Remove assert for disalowed
32930 regno values, call output_operand_lossage instead.
32932 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32935 * config/i386/constraints.md (Yc): New register constraint.
32936 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
32937 Use Yc constraint for alternative 2 of operand 0. Remove
32938 preferred_for_speed attribute.
32940 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32942 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
32943 lastprivate clauses in SIMT case.
32945 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
32947 * doc/invoke.texi (-Wextra-semi): Document new warning option.
32949 2017-04-20 Richard Biener <rguenther@suse.de>
32951 PR tree-optimization/57796
32952 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
32953 as N scalar stores.
32954 (vect_model_load_cost): Cost gathers as N scalar loads.
32956 2017-04-20 Richard Biener <rguenther@suse.de>
32958 * ggc-page.c (ggc_allocated_p): Rename to ...
32959 (safe_lookup_page_table_entry): ... this and return the lookup
32961 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
32963 2017-04-20 Richard Biener <rguenther@suse.de>
32965 PR tree-optimization/80453
32966 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
32967 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
32968 from the conditions.
32969 (vn_phi_eq): Pass them down.
32970 (vn_phi_lookup): Record them.
32971 (vn_phi_insert): Likewise.
32973 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
32975 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
32976 uninitialized variable warning to avoid buffer overrun.
32978 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32981 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
32982 is suppressed for '{ 0 }' in C.
32984 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32986 * BASE-VER: Set to 8.0.0.
32988 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
32990 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
32991 priority .init_array and .fini_array section with SECTION_NOTYPE
32994 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32996 PR middle-end/80423
32997 * tree.h (build_array_type): Add typeless_storage default argument.
32998 * tree.c (type_cache_hasher::equal): Also compare
32999 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33000 (build_array_type): Add typeless_storage argument, set
33001 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33003 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33004 (build_array_type): Likewise. Add typeless_storage argument.
33006 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33007 Jakub Jelinek <jakub@redhat.com>
33009 PR tree-optimization/80426
33010 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33011 operation on symbolic operands, also compute the overflow for the
33012 invariant part when the operation degenerates into a negation.
33014 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33017 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33018 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33021 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33023 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33026 * config/avr/avr.c (tree.h): Include it.
33027 (cgraph.h): Include it.
33028 (avr_encode_section_info): Don't warn for uninitialized progmem
33029 variable if it's just an alias.
33031 2017-04-19 Richard Biener <rguenther@suse.de>
33034 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33035 when needed by AutoPGO.
33037 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33040 * doc/lto.texi: Remove an extra 'that'.
33042 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33044 PR rtl-optimization/80429
33045 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33046 are only used in debug insns.
33048 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33049 Vladimir Makarov <vmakarov@redhat.com>
33051 * config/sparc/predicates.md (input_operand): Add comment. Return
33052 true for any memory operand when LRA is in progress.
33053 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33055 2017-04-18 Jeff Law <law@redhat.com>
33058 * mips.md ({return,simple_return}_internal): Do not overwrite
33061 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33063 PR tree-optimization/80443
33064 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33065 instead of adding 1, subtract -1 and similarly instead of subtracting
33068 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33070 PR rtl-optimization/80357
33071 * haifa-sched.c (tmp_bitmap): New variable.
33072 (model_recompute): Handle duplicate use records.
33073 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33074 (free_global_sched_pressure_data): Free it.
33076 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33079 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33080 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33081 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33082 instead of SYSTEM_HEADER_DIR.
33084 2017-04-18 Jeff Law <law@redhat.com>
33086 PR middle-end/80422
33087 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33088 predecessors after walking up the insn chain.
33090 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33093 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33094 sizetype type into debug info.
33096 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33099 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33100 unneeded test for TARGET_UPPER_REGS_SF.
33101 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33103 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33106 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33107 instead of gsi_after_labels.
33109 2017-04-18 Jeff Law <law@redhat.com>
33111 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33115 2017-04-13 Jeff Law <law@redhat.com>
33116 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33117 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33119 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33122 * config/avr/avr.c (intl.h): Include it.
33123 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33125 2017-04-18 Martin Liska <mliska@suse.cz>
33127 PR gcov-profile/78783
33128 * gcov-tool.c (gcov_output_files): Validate that destination
33129 file is either removed by the tool or by a user.
33131 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33132 Guy Benyei <guybe@mellanox.com>
33134 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33135 block, and do not negate it, the stored id is already negative.
33137 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33139 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33141 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33144 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33145 masks of options that should be turned off if the VSX vector
33146 options are turned off.
33147 (OTHER_P8_VECTOR_MASKS): Likewise.
33148 (OTHER_VSX_VECTOR_MASKS): Likewise.
33149 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33150 rs6000_disable_incompatible_switches to validate no type switches
33152 (rs6000_incompatible_switch): New function to disallow turning on
33153 other vector options if -mno-vsx, -mno-power8-vector, or
33154 -mno-power9-vector are specified.
33156 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33158 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33160 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33162 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33163 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33164 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33165 (ARG_POINTER_CFA_OFFSET): Likewise.
33167 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33169 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33170 conditions to take advantage of various optimizations.
33172 2017-04-13 Jeff Law <law@redhat.com>
33174 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33175 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33176 (zero_extendsidi2_dext): Likewise.
33178 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33181 * fold-const.c (fold_ternary_loc): Revert
33182 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33185 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33187 PR rtl-optimization/80343
33188 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33189 new scratch pseudo.
33191 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33194 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33195 to ubsan_encode_value.
33197 2017-04-13 Jeff Law <law@redhat.com>
33199 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33200 appearing in DEBUG_INSNs.
33202 2017-04-13 Martin Liska <mliska@suse.cz>
33204 PR gcov-profile/80413
33205 * gcov-io.c (gcov_write_string): Copy to buffer just when
33206 allocated size is greater than zero.
33208 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33211 * dwarf2out.c (decls_for_scope): Ignore declarations of
33212 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33214 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33217 * ipa-visibility.c (non_local_p): Fix typos.
33218 (localize_node): When localizing symbol in same comdat group,
33219 dissolve the group only when we know external symbols are going
33221 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33223 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33225 PR tree-optimization/79390
33226 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33227 order does not result in usable sequence, retry with reversed operand
33233 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33234 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33235 op0 instead of fold_convert_loc (loc, type, arg0).
33237 2017-04-12 Jeff Law <law@redhat.com>
33239 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33240 has a delay slot in the generated code.
33242 * config/cris/cris.md (cris_preferred_reload_class): Return
33243 GENNONACR_REGS rather than GENERAL_REGS.
33245 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33248 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33249 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33250 signedness of the result type.
33252 2017-04-12 Richard Biener <rguenther@suse.de>
33253 Jeff Law <law@redhat.com>
33255 PR tree-optimization/80359
33256 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33257 trim stores to TARGET_MEM_REFs.
33259 2017-04-12 Richard Biener <rguenther@suse.de>
33261 PR tree-optimization/79390
33262 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33263 threading case even more.
33265 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33268 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33269 for quad_address_p for TImode, instead of just not indexed_address.
33271 2017-04-12 Richard Biener <rguenther@suse.de>
33272 Bernd Edlinger <bernd.edlinger@hotmail.de>
33274 PR middle-end/79671
33275 * alias.c (component_uses_parent_alias_set_from): Handle
33276 TYPE_TYPELESS_STORAGE.
33277 (get_alias_set): Likewise.
33278 * tree-core.h (tree_type_common): Add typeless_storage flag.
33279 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33280 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33281 for types containing members with TYPE_TYPELESS_STORAGE.
33282 (place_field): Likewise.
33283 (layout_type): Likewise for ARRAY_TYPE.
33284 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33285 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33286 TYPE_TYPELESS_STORAGE.
33287 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33289 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33292 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33293 first argument to type.
33295 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33299 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33300 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33301 (rs6000_expand_binop_builtin): Likewise.
33302 (rs6000_expand_ternop_builtin): Likewise; also add missing
33303 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33304 vshasigma built-ins.
33305 * doc/extend.texi: Document that vec_xxpermdi's third argument
33306 must be a constant.
33308 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33310 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33311 Use shift_const cost parameter when calculating gain of STV shifts.
33313 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33315 PR rtl-optimization/70478
33316 * lra-constraints.c (process_alt_operands): Check memory for
33317 disfavoring memory insn operand.
33319 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33321 PR middle-end/80100
33322 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33323 left shift in unsigned HOST_WIDE_INT type.
33325 PR rtl-optimization/80385
33326 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33327 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33330 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33331 if they have any depend clauses.
33333 2017-04-11 Martin Liska <mliska@suse.cz>
33336 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33337 * ipa-split.c (split_function): Create a local comdat symbol
33338 if caller is in a comdat group.
33340 2017-04-11 Martin Liska <mliska@suse.cz>
33343 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33346 2017-04-11 Martin Sebor <msebor@redhat.com>
33348 PR middle-end/80364
33349 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33350 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33352 (directive::set_width, directive::set_precision, format_character):
33354 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33357 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33360 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33361 conflict, set target->arch_name instead of target->cpu_name.
33363 2017-04-11 Richard Biener <rguenther@suse.de>
33365 PR tree-optimization/80374
33366 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33367 build_zero_cst, remove fold_convertible_p check again.
33369 2017-04-11 Martin Liska <mliska@suse.cz>
33372 * ubsan.c (instrument_object_size): Do not instrument register
33375 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33378 * config/i386/i386-builtin-types.def
33379 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33380 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33381 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33382 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33383 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33384 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33385 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33386 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33387 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33388 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33389 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33390 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33391 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33392 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33393 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33394 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33395 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33396 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33397 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33398 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33399 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33400 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33401 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33402 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33403 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33404 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33405 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33406 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33407 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33408 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33409 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33410 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33411 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33412 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33413 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33414 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33415 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33416 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33417 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33418 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33419 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33420 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33421 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33422 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33423 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33424 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33426 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33427 flag to second_arg_count, handle 4 argument function type _COUNT
33428 aliases, handle second_arg_count on second argument rather than last.
33430 2017-04-10 Jeff Law <law@redhat.com>
33432 PR tree-optimization/80374
33433 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33434 record anything if we can not convert integer_zero_node to the
33437 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33440 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33441 Enhance special handling given to the TARGET_P9_MINMAX option in
33442 relation to certain other options.
33444 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33446 PR tree-optimization/80153
33447 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33448 remove POINTER_PLUS_EXPR's base part directly, rather than through
33451 2017-04-10 Richard Biener <rguenther@suse.de>
33452 Bin Cheng <bin.cheng@arm.com>
33454 PR tree-optimization/80153
33455 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33456 the first element of pointer type aff_tree. Build result expr in
33458 (add_elt_to_tree): Convert to type unconditionally. Remove other
33459 fold_convert calls.
33460 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33461 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33463 2017-04-10 Richard Biener <rguenther@suse.de>
33465 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33468 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33470 PR rtl-optimization/70478
33471 * lra-constraints.c (curr_small_class_check): New.
33472 (update_and_check_small_class_inputs): New.
33473 (process_alt_operands): Update curr_small_class_check. Disfavor
33474 alternative insn memory operands. Check available regs for small
33477 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33480 * config/mips/mips.opt (-mvirt): Update description.
33481 * doc/invoke.texi (-mvirt): Likewise.
33483 2017-04-10 Richard Biener <rguenther@suse.de>
33485 PR middle-end/80362
33486 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33487 looking for NEGATE_EXPR in -A / -B to A / B folding.
33489 2017-04-10 Martin Liska <mliska@suse.cz>
33491 PR gcov-profile/80224
33492 * gcov.c (print_usage): Fix usage string.
33493 (get_gcov_intermediate_filename): Remove.
33494 (output_gcov_file): Use both for normal and intermediate format.
33495 (generate_results): Do not initialize special file for
33496 intermediate format.
33498 2017-04-10 Richard Biener <rguenther@suse.de>
33500 PR tree-optimization/80304
33501 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33504 2017-04-10 Nathan Sidwell <nathan@acm.org>
33507 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33508 (rs6000_init_builtins): Use it.
33510 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33512 * config/arm/arm.md (<mrc>): Add mode to SET source.
33513 (<mrrc>): Likewise.
33515 2017-04-10 Richard Biener <rguenther@suse.de>
33517 PR middle-end/80344
33518 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33520 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33523 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33524 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33525 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33526 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33527 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33528 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33529 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33530 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33531 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33532 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33533 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33534 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33535 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33536 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33537 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33538 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33539 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33540 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33541 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33542 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33543 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33544 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33545 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33547 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33549 PR rtl-optimization/70478
33550 * lra-constraints.c: Reverse the last patch.
33552 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33554 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33555 Add comment for WCHAR_T.
33557 2017-04-08 Martin Liska <mliska@suse.cz>
33560 2017-04-07 Martin Liska <mliska@suse.cz>
33563 * ipa-split.c (split_function): Add function part to a same comdat
33566 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33569 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33571 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33573 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33574 (vec_pairing): ...to this.
33575 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33576 (rs6000_sched_init): Adjust for name change.
33577 (struct rs6000_sched_context): Likewise.
33578 (rs6000_init_sched_context): Likewise.
33579 (rs6000_set_sched_context): Likewise.
33581 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33587 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33589 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33590 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33591 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33593 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33595 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33597 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33599 PR rtl-optimization/70703
33600 * ira-color.c (update_conflict_hard_regno_costs): Use
33601 int64_t instead of HOST_WIDE_INT.
33603 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33605 PR rtl-optimization/70478
33606 * lra-constraints.c (process_alt_operands): Disfavor alternative
33607 insn memory operands.
33609 2017-04-07 Jeff Law <law@redhat.com>
33611 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33612 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33614 2017-04-07 Martin Liska <mliska@suse.cz>
33617 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33618 Show error message instead of an ICE.
33620 2017-04-07 Martin Liska <mliska@suse.cz>
33623 * ipa-split.c (split_function): Add function part to a same comdat
33626 2017-04-07 Richard Biener <rguenther@suse.de>
33628 PR middle-end/80341
33629 * tree.c (get_unwidened): Also handle ! for_type case for
33631 * convert.c (do_narrow): Split out from ...
33632 (convert_to_integer_1): ... here. Do not pass final truncation
33633 type to get_unwidened for TRUNC_DIV_EXPR.
33635 2017-04-07 Richard Biener <rguenther@suse.de>
33637 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33639 (aff_combination_const): Adjust.
33640 (aff_combination_scale): Likewise.
33641 (aff_combination_add_elt): Likewise.
33642 (aff_combination_add_cst): Likewise.
33643 (aff_combination_convert): Likewise.
33644 (add_elt_to_tree): Likewise. Remove unused argument.
33645 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33647 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33649 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33651 * config/arm/arm.c (arm_default_short_enums): Use
33652 ARM_DEFAULT_SHORT_ENUMS.
33653 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33655 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33658 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33659 members with redundant out-of-class redeclaration.
33661 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33664 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33665 * config/i386/i386.md (*zero_extendsidi2):
33666 Add (?*x,*x) and (?*v,*v) alternatives.
33668 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33671 * config/i386/i386.c (ix86_expand_builtin)
33672 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33673 mode from insn data. Convert operands to insn operand mode.
33674 Copy operands that don't satisfy insn predicate to a register.
33676 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33678 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33681 2017-04-06 Richard Biener <rguenther@suse.de>
33683 PR tree-optimization/80334
33684 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33685 preserve alignment of accesses.
33687 2017-04-06 Richard Biener <rguenther@suse.de>
33689 PR tree-optimization/80262
33690 * tree-sra.c (build_ref_for_offset): Preserve address-space
33692 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33693 Drop useless address-space information on MEM_REF offsets.
33695 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33697 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33699 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33701 PR rtl-optimization/70703
33702 * ira-color.c (update_conflict_hard_regno_costs): Use
33703 HOST_WIDE_INT instead of long.
33705 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33708 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33709 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33711 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33712 for x86_64 target. Handle -m3dnowa option.
33714 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33716 PR rtl-optimization/70703
33717 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33718 (update_conflict_hard_regno_costs): Use long instead of unsigned
33719 arithmetic for cost calculation.
33721 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33722 Bernd Edlinger <bernd.edlinger@hotmail.de>
33725 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33728 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33731 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33732 ptr_mode with Pmode throughout.
33733 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33734 into probe_stack_range and use DImode.
33736 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33739 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33740 call_eh_return is true.
33742 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33744 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33745 Initialize last_match_fntype_index.
33747 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33750 * tree-nvr.c: Include internal-fn.h.
33751 (pass_return_slot::execute): Ignore internal calls without
33754 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33755 Richard Biener <rguenther@suse.de>
33758 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33759 captures used multiple times, except for the last use.
33760 * generic-match-head.c: Include gimplify.h.
33762 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33764 PR tree-optimization/79390
33765 * target.h (struct noce_if_info): Declare.
33766 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33767 * target.def (noce_conversion_profitable_p): New target hook.
33768 * ifcvt.h (struct noce_if_info): New type, moved from ...
33769 * ifcvt.c (struct noce_if_info): ... here.
33770 (noce_conversion_profitable_p): Renamed to ...
33771 (default_noce_conversion_profitable_p): ... this. No longer
33773 (noce_try_store_flag_constants, noce_try_addcc,
33774 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33775 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33776 instead of noce_conversion_profitable_p.
33777 * config/i386/i386.c: Include ifcvt.h.
33778 (ix86_option_override_internal): Don't override
33779 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33780 (ix86_noce_conversion_profitable_p): New function.
33781 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33782 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33783 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33784 * doc/tm.texi: Regenerated.
33786 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33788 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33791 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33794 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33795 instructions for small multiply cores.
33797 2017-04-04 Jeff Law <law@redhat.com>
33799 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33801 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33802 that are not set by the loop over the elements.
33804 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33807 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33808 int mode, convert_modes it to mode as unsigned, otherwise use
33809 lowpart_subreg to mode rather than SImode.
33810 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33811 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33812 Use DImode instead of SImode for the shift count operand.
33813 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33816 2017-04-04 Richard Biener <rguenther@suse.de>
33818 PR middle-end/80281
33819 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33820 arithmetic done for the negate or the plus. Simplify.
33821 (A - (-B) -> A + B): Likewise.
33822 * fold-const.c (split_tree): Make sure to not negate pointers.
33824 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33826 PR rtl-optimization/60818
33827 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33828 a compare of comparisons with the thing compared if this results
33829 in a different machine mode.
33831 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33833 * alias.c (base_alias_check): Fix typo in comment.
33834 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33835 * cgraphunit.c (symbol_table::compile): Likewise.
33836 * collect2.c (maybe_run_lto_and_relink): Likewise.
33837 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33838 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33839 * config/avr/avr.c (avr_map_op_t): Likewise.
33840 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33841 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33842 * config/epiphany/epiphany.md (movcc): Likewise.
33843 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33844 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33846 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33847 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33848 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33849 * config/sh/sh.c (sh_rtx_costs): Likewise.
33850 * fold-const.c (fold_truth_andor): Likewise.
33851 * genautomata.c (collapse_flag): Likewise.
33852 * gengtype.h (struct type::u::s): Likewise.
33853 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
33854 * input.c (FORMAT_AMOUNT): Likewise.
33855 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
33856 (known_aggs_to_agg_replacement_list): Likewise.
33857 * ipa-inline-analysis.c: Likewise.
33858 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
33859 * ipa-polymorphic-call.c
33860 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
33861 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
33862 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
33864 * modulo-sched.c (apply_reg_moves): Likewise.
33865 * omp-expand.c (build_omp_regions_1): Likewise.
33866 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
33867 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
33868 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
33869 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33870 * value-prof.c: Likewise.
33871 * var-tracking.c (val_reset): Likewise.
33873 2017-04-03 Richard Biener <rguenther@suse.de>
33875 PR tree-optimization/80275
33876 * fold-const.c (split_address_to_core_and_offset): Handle
33879 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
33881 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
33882 descriptors is at least equal to that of functions.
33884 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33886 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
33888 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33891 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
33892 (mov<IMOD4:mode>): New expander.
33893 (*mov<IMOD4:mode>_internal): New insn and split pattern.
33895 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
33897 PR rtl-optimization/79405
33898 * fwprop.c (propagations_left): New variable.
33899 (forward_propagate_into): Decrement it.
33900 (fwprop_init): Initialize it.
33901 (fw_prop): If the variable has reached zero, stop propagating.
33902 (fwprop_addr): Ditto.
33904 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33907 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
33908 a FUNCTION_DECL, pass it as decl instead of origin to
33911 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
33913 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
33916 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
33919 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
33920 TARGET_VSX_SMALL_INTEGER.
33922 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33924 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
33925 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
33927 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33929 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
33930 extraction from odd-numbered MSA register.
33932 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33934 PR middle-end/80173
33935 * expmed.c (store_bit_field_1): Don't attempt to create
33936 a word subreg out of hard registers wider than word if they
33937 have HARD_REGNO_NREGS of 1 for their mode.
33939 PR middle-end/80163
33940 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
33941 conversions to integer types wider than word and pointer.
33944 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
33945 (rtx_equal_for_cselib_p): Pass 0 to it.
33946 * cselib.c (cselib_hasher::equal): Likewise.
33947 (rtx_equal_for_cselib_1): Add depth argument. If depth
33948 is 128, don't look up VALUE locs and punt. Increment
33949 depth in recursive calls when walking VALUE locs.
33951 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
33953 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
33954 (make_gcov_file_name): Use the canonical path name for generating
33956 (read_line): Fix handling of files with ascii null bytes.
33958 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
33960 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
33961 to initialise a vector register instead
33962 of using a const_int.
33964 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33966 PR translation/80189
33967 * gimplify.c (omp_default_clause): Use %qs instead of %s in
33968 diagnostic messages.
33970 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
33973 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
33974 (dfp_diex_<mode>): Update mode of operand 1.
33975 * doc/extend.texi (dxex, dxexq): Document change to return type.
33976 (diex, diexq): Document change to argument type.
33978 2017-03-30 Martin Jambor <mjambor@suse.cz>
33981 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
33982 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
33983 it reflects the signature changes performed at the callee side.
33984 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
33985 to cgraph_build_function_type_skip_args.
33986 (build_function_decl_skip_args): Adjust call to the above function.
33988 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33991 * config/i386/sse.md
33992 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
33993 register as dest whenever it is a MEM not rtx_equal_p to the
33994 corresponding dup operand, and when forcing into reg move the
33995 reg into the memory afterwards.
33996 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
33997 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
33998 for the force_reg mode.
33999 (avx512vl_vextractf128<mode>): Use register as dest either
34000 always when a MEM, or when it is a MEM not rtx_equal_p to the
34001 corresponding dup operand, or even not when it is a CONST_VECTOR
34002 depending on the mode and lo vs. hi.
34003 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34005 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34006 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34007 Likewise. Require that operands[2] is even.
34008 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34009 Remove extraneous parens. Require that operands[2] is a multiple
34011 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34012 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34013 disallow memory then.
34015 2017-03-30 Richard Biener <rguenther@suse.de>
34017 PR tree-optimization/77498
34018 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34019 to non-constants over backedges.
34021 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34023 PR rtl-optimization/80233
34024 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34025 as last_combined_insn. Do not test for BARRIER_P separately.
34027 2017-03-29 Andreas Schwab <schwab@suse.de>
34030 * calls.c (prepare_call_address): Convert funexp to Pmode before
34031 copying to temp reg.
34033 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34035 PR tree-optimization/80158
34036 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34037 Handle possible future case of more than one alternate
34039 (replace_rhs_if_not_dup): Likewise.
34040 (replace_one_candidate): Likewise.
34042 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34044 PR rtl-optimization/80193
34045 * ira.c (ira): Do not check allocation for LRA.
34047 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34049 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34050 (nvptx_output_simt_exit): Declare.
34051 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34052 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34053 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34054 (nvptx_declare_function_name): ...here. Emit declaration of local
34055 memory space buffer for omp_simt_enter insn.
34056 (nvptx_output_unisimt_switch): New.
34057 (nvptx_output_softstack_switch): New.
34058 (nvptx_output_simt_enter): New.
34059 (nvptx_output_simt_exit): New.
34060 * config/nvptx/nvptx.h (struct machine_function): New fields
34061 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34062 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34063 (UNSPECV_SIMT_EXIT): Ditto.
34064 (omp_simt_enter_insn): New insn.
34065 (omp_simt_enter): New expansion.
34066 (omp_simt_exit): New insn.
34067 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34069 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34070 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34071 (expand_GOMP_SIMT_EXIT): New.
34072 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34073 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34074 (GOMP_SIMT_EXIT): Ditto.
34075 * target-insns.def (omp_simt_enter): New insn.
34076 (omp_simt_exit): Ditto.
34077 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34079 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34080 (lower_rec_input_clauses): Likewise.
34081 (lower_lastprivate_clauses): Handle SIMT privatization.
34083 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34084 (ompdevlow_adjust_simt_enter): New.
34085 (find_simtpriv_var_op): New.
34086 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34087 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34089 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34090 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34091 (copy_decl_for_dup_finish): Ditto.
34093 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34095 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34098 * config/i386/i386.c (ix86_option_override_internal): Always
34099 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34101 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34103 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34105 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34107 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34108 mark new edge's irreducible flag accordign to it.
34109 (vect_do_peeling): Check loop preheader edge's irreducible flag
34110 and pass it to function slpeel_add_loop_guard.
34112 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34114 PR tree-optimization/80218
34115 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34116 Update block frequencies and counts.
34118 2017-03-28 Richard Biener <rguenther@suse.de>
34120 PR tree-optimization/78644
34121 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34122 of a simplification result we may not use it at all.
34124 2017-03-28 Richard Biener <rguenther@suse.de>
34127 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34128 without arguments, generate default definition of a SSA name.
34130 2017-03-28 Richard Biener <rguenther@suse.de>
34132 PR middle-end/80222
34133 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34134 TYPE_REF_CAN_ALIAS_ALL references.
34135 * fold-const.c (fold_indirect_ref_1): Likewise.
34137 2017-03-28 Martin Liska <mliska@suse.cz>
34140 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34141 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34143 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34144 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34146 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34147 (EXTRA_SPECS): Define.
34148 (SUBTARGET_EXTRA_SPECS): Likewise.
34149 (SUBTARGET_CPP_SPEC): Likewise.
34150 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34151 SUBTARGET_EXTRA_SPECS.
34152 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34154 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34156 * config/arc/simdext.md (vst64_insn): Update pattern.
34157 (vld32wh_insn): Likewise.
34158 (vld32wl_insn): Likewise.
34159 (vld64_insn): Likewise.
34160 (vld32_insn): Likewise.
34162 2017-03-28 Marek Polacek <polacek@redhat.com>
34165 * fold-const.c (fold_comparison): Use protected_set_expr_location
34166 instead of SET_EXPR_LOCATION.
34168 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34170 * tree.c (add_expr): Avoid name lookup warning.
34172 2017-03-27 Jeff Law <law@redhat.com>
34174 PR tree-optimization/80216
34175 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34176 function name. Limit recursion depth.
34177 (record_temporary_equivalences): Corresponding changes.
34179 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34181 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34184 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34187 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34189 * cfgcleanup.c (reg_note_cfa_p): New array.
34190 (insns_have_identical_cfa_notes): New function.
34191 (old_insns_match_p): Don't cross-jump in between /f
34192 and non-/f instructions. If both i1 and i2 are frame related,
34193 verify all CFA notes, their order and content.
34195 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34198 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34199 HImode and SImode with zero extend to DImode to one insn.
34200 (bswap<mode>2_extenddi): Likewise.
34201 (bswapsi2_extenddi): Likewise.
34202 (bswaphi2_extendsi): Likewise.
34203 (bswaphi2): Combine bswap HImode and SImode into one insn.
34204 Separate memory insns from swapping register.
34205 (bswapsi2): Likewise.
34206 (bswap<mode>2): Likewise.
34207 (bswaphi2_internal): Delete, no longer used.
34208 (bswapsi2_internal): Likewise.
34209 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34210 store, and gpr<-gpr swap insns.
34211 (bswap<mode>2_store): Likewise.
34212 (bswaphi2_reg): Register only splitter, combine with the splitter.
34213 (bswaphi2 splitter): Likewise.
34214 (bswapsi2_reg): Likewise.
34215 (bswapsi2 splitter): Likewise.
34216 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34217 the insns into load, store, and register/register insns.
34218 (bswapdi2_ldbrx): Likewise.
34219 (bswapdi2_load): Likewise.
34220 (bswapdi2_store): Likewise.
34221 (bswapdi2_reg): Likewise.
34223 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34225 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34226 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34228 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34231 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34233 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34234 special handling for target option conflicts between dform
34235 options (-mpower9-dform, -mpower9-dform-vector,
34236 -mpower9-dform-scalar) and -mno-direct-move.
34238 2017-03-27 Richard Biener <rguenther@suse.de>
34240 PR tree-optimization/80181
34241 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34243 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34245 * config/arc/predicates.md (move_double_src_operand): Replace the
34246 call to move_double_src_operand with a call to address_operand.
34248 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34250 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34251 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34252 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34254 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34256 * config/arc/predicates.md (long_immediate_loadstore_operand):
34257 Consider scaled addresses cases.
34259 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34261 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34262 restored when in interrupt.
34263 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34264 doesn't have delay slot.
34266 2017-03-27 Richard Biener <rguenther@suse.de>
34269 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34270 inlined thunk clones.
34272 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34275 * asan.c (instrument_derefs): Copy over last operand from
34276 original COMPONENT_REF to the new COMPONENT_REF with
34277 DECL_BIT_FIELD_REPRESENTATIVE.
34278 * ubsan.c (instrument_object_size): Likewise.
34280 2017-03-27 Richard Biener <rguenther@suse.de>
34282 PR tree-optimization/80170
34283 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34284 sure DR/SCEV didnt fold in constants we do not see when looking
34285 at the reference base alignment.
34287 2017-03-27 Richard Biener <rguenther@suse.de>
34289 PR middle-end/80171
34290 * gimple-fold.c (fold_ctor_reference): Properly guard against
34291 NULL return value from canonicalize_constructor_val.
34293 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34296 * config/i386/i386.c (ix86_expand_builtin)
34297 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34298 flags reg setting and flags reg using instructions.
34299 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34300 clobbering instructions to zero extend op2.
34302 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34304 * doc/install.texi (Configuration) <--with-aix-soname>:
34305 Update link to AIX ld.
34307 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34309 PR rtl-optimization/80160
34310 PR rtl-optimization/80159
34311 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34312 reg_alternate_class into account.
34314 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34317 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34318 to consider in curr_insn_transform.
34320 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34322 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34323 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34324 and emit_mode_inner.
34326 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34328 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34329 argument to the overloaded builtin variants. Use the new flag to
34330 deprecate certain builtin variants.
34331 * config/s390/s390-builtin-types.def: Add new builtin types.
34332 * config/s390/s390-builtins.h: Support new flags field for
34333 overloaded builtins.
34334 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34335 (s390_macro_to_expand): Enable vector float data type.
34336 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34337 builtins by incrementing the __VEC__ version number.
34338 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34340 (s390_resolve_overloaded_builtin): Emit error messages depending
34341 on the builtin flags.
34342 * config/s390/s390.c (s390_expand_builtin): Support additional
34343 flags argument. Change error message to match the messages
34344 emitted in s390-c.c.
34345 * config/s390/s390.md: New UNSPEC_* constants.
34346 (op_type): Add new instruction types.
34347 * config/s390/vecintrin.h: Add new builtins and test data class
34349 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34350 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34351 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34352 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34353 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34354 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34356 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34357 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34358 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34359 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34361 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34362 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34363 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34364 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34366 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34367 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34368 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34369 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34370 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34371 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34372 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34374 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34375 ("vec_scatter_element<V_HW_4:mode>_DI")
34376 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34377 ("vec_fpint<mode>", "vflls")
34378 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34379 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34380 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34381 ("*vec_cmphe<mode>_cc"): ... these.
34383 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34384 mode constant instead of magic value.
34386 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34388 * config/s390/s390.c (s390_expand_vec_compare): Support other
34389 vector floating point modes than just V2DF.
34390 (s390_expand_vcond): Likewise.
34391 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34392 (s390_cannot_change_mode_class): Prevent mode changes between TF
34393 and V1TF in vector registers.
34394 * config/s390/s390.md (DF, SF): New mode attributes.
34395 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34396 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34397 SFmode support for VRs.
34398 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34400 (VFT, VF_HW): New mode iterators.
34401 (vw, sdx): New mode attributes.
34402 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34403 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34404 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34405 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34406 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34407 also the new vector floating point modes. Renaming to ...
34409 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34410 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34411 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34412 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34413 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34414 ("vec_unordered<mode>"): ... these.
34416 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34417 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34418 ("*vec_extendv2df"): New insn definitions.
34420 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34422 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34423 ("mulditi3_2", "*muldi3_sign"): New patterns.
34424 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34425 rename the pattern definition.
34427 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34429 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34431 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34433 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34435 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34436 instruction if possible.
34437 * config/s390/vector.md (vec_halfnumelts): New mode
34439 ("*vec_vllezlf<mode>"): New pattern.
34441 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34443 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34444 ("popcountv4si2", "popcountv2di2"): Rename to ...
34445 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34446 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34448 ("popcount<mode>2_vxe"): New pattern.
34450 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34452 * common/config/s390/s390-common.c (processor_flags_table): Add
34454 * config.gcc: Add arch12.
34455 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34456 Default to arch12 for unknown CPU model numbers.
34457 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34458 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34459 PROCESSOR_max sanity check.
34460 * config/s390/s390-opts.h (enum processor_type): Add
34462 * config/s390/s390.c (processor_table): Add arch12.
34463 (s390_expand_builtin): Add check for B_VXE flag.
34464 (s390_issue_rate): Add PROCESSOR_ARCH12.
34465 (s390_get_sched_attrmask): Likewise.
34466 (s390_get_unit_mask): Likewise.
34467 (s390_sched_score): Enable z13 scheduling for arch12.
34468 (s390_sched_reorder): Likewise.
34469 (s390_sched_variable_issue): Likewise.
34470 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34472 (s390_tune_attr): Use z13 scheduling also for arch12.
34473 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34474 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34475 (TARGET_VXE_P): New macros.
34476 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34477 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34478 * config/s390/s390.opt: Add arch12 as processor_type.
34480 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34482 * config/s390/s390.md
34483 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34484 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34485 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34487 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34488 Rename expanders to ...
34490 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34491 ("fixuns_truncdddi2_emu"): ... these.
34493 ("fixuns_trunc<mode>si2_emu"): New expander.
34495 ("*fixuns_truncdfdi2_z13"): Rename to ...
34496 ("*fixuns_truncdfdi2_vx"): ... this.
34498 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34500 * config/s390/2964.md: Remove the single element vector compare
34501 instructions which are no longer used.
34502 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34504 (s390_canonicalize_comparison): Remove handling of DFmode
34506 (s390_expand_vec_compare_scalar): Remove function.
34507 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34508 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34510 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34512 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34514 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34516 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34517 will anyway by matched by mov<mode>_64dfp.
34519 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34521 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34522 vlef/vstef. Add missing operand to vleif.
34524 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34526 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34527 pair for all vector types with 64 bit elements.
34528 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34529 * config/s390/vector.md (V_HW_64): ... here.
34530 (V_128_NOSINGLE): New mode iterator.
34531 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34532 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34533 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34534 ("*vec_load_pairv2di"): Change to ...
34535 ("*vec_load_pair<mode>"): ... this one.
34537 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34539 * config/s390/constraints.md: Add comments.
34540 (jKK): Reject element sizes > 8 bytes.
34541 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34543 * config/s390/s390.md: Add the s_operand checks formerly in
34544 s390_split_ok_p to various splitters where they are still
34546 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34547 for 128 bit vectors. Plus two splitters.
34549 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34551 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34554 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34557 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34558 error if the boundary argument is not constant.
34560 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34562 PR rtl-optimization/80112
34563 * loop-doloop.c (doloop_condition_get): Don't check condition
34564 if cmp isn't SET with IF_THEN_ELSE src.
34566 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34568 PR tree-optimization/80158
34569 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34570 replacing a candidate statement, also replace it for the
34571 candidate's alternate interpretation.
34572 (replace_rhs_if_not_dup): Likewise.
34573 (replace_one_candidate): Likewise.
34575 2017-03-24 Richard Biener <rguenther@suse.de>
34577 PR tree-optimization/80167
34578 * graphite-isl-ast-to-gimple.c
34579 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34581 (translate_isl_ast_to_gimple::get_rename): Likewise.
34583 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34585 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34586 handling of certain combinations of target options, including the
34587 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34588 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34590 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34593 * config/arm/arm.md (*load_multiple): Add reload_completed to
34594 matching condition.
34596 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34597 Richard Biener <rguenth@suse.de>
34599 PR tree-optimization/79908
34600 PR tree-optimization/80136
34601 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34602 been cast away, gimplify_and_add suffices.
34604 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34606 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34608 2017-03-23 Richard Biener <rguenther@suse.de>
34610 PR tree-optimization/80032
34611 * gimplify.c (gimple_push_cleanup): Forced unconditional
34612 cleanups still have to go to the conditional_cleanups
34615 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34617 PR tree-optimization/80072
34618 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34620 (next_operand_entry_id): Change type to unsigned int.
34621 (sort_by_operand_rank): Make sure to return the right return value
34622 even if unsigned fields are bigger than INT_MAX.
34623 (struct oecount): Change cnt and id type to unsigned int.
34624 (oecount_hasher::equal): Formatting fix.
34625 (oecount_cmp): Make sure to return the right return value
34626 even if unsigned fields are bigger than INT_MAX.
34627 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34630 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34631 TREE_READONLY on result if writing it more than once.
34634 * doc/invoke.texi (-fsanitize=thread): Document that with
34635 -fnon-call-exceptions atomics are not able to throw
34639 * tsan.c: Include tree-eh.h.
34640 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34641 maybe_clean_or_replace_eh_stmt where needed.
34642 (instrument_memory_accesses): Add cfg_changed argument.
34643 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34644 if it returned true.
34645 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34647 PR rtl-optimization/63191
34648 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34649 wrapper function, moved the whole old content into ...
34650 (ix86_delegitimize_address_1): ... this. New inline function.
34651 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34652 true as last argument instead of ix86_delegitimize_address.
34654 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34656 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34657 cortexa57_branch_cost.
34659 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34661 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34663 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34666 * doc/md.texi (Constraints): Document wA constraint.
34667 * config/rs6000/constraints.md (wA): New.
34668 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34669 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34670 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34671 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34673 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34676 * gimplify.c (is_oacc_declared): New function.
34677 (oacc_default_clause): Use it to set default flags for acc declared
34678 variables inside parallel regions.
34679 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34680 declared variables.
34681 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34682 declare attribute to any decl as necessary.
34684 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34687 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34688 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34689 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34690 (arm_arch_lpae): This.
34691 * config/arm/arm.c (arm_arch7ve): Rename into ...
34692 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34693 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34696 2017-03-22 Martin Liska <mliska@suse.cz>
34699 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34700 error message instead of an ICE.
34702 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34704 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34706 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34708 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34710 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34713 2017-03-21 Martin Sebor <msebor@redhat.com>
34715 * doc/extend.texi: Use "cannot" instead of "can't."
34716 * doc/hostconfig.texi: Same.
34717 * doc/install.texi: Same.
34718 * doc/invoke.texi: Same.
34719 * doc/loop.texi: Same.
34720 * doc/md.texi: Same.
34721 * doc/objc.texi: Same.
34722 * doc/rtl.texi: Same.
34723 * doc/tm.texi: Same.
34724 * doc/tm.texi.in: Same.
34725 * doc/trouble.texi: Same.
34727 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34730 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34731 (collect_checksum_attributes): Set it.
34732 (die_checksum_ordered): Use it.
34734 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34736 PR tree-optimization/79908
34737 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34738 change: For a VA_ARG whose LHS has been cast away, use
34739 force_gimple_operand to construct the side effects.
34741 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34743 PR translation/80001
34744 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34745 more amenable to translation.
34746 (oacc_loop_auto_partitions): Likewise.
34748 2017-03-21 Marek Polacek <polacek@redhat.com>
34749 Martin Sebor <msebor@redhat.com>
34751 PR tree-optimization/80109
34752 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34753 on INTEGRAL_TYPE_P.
34755 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34756 Segher Boessenkool <segher@kernel.crashing.org>
34759 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34760 check reg_used_between_p between insn and one of succ or succ2
34761 depending on if succ is artificial insn not inserted into insn
34764 2017-03-21 Martin Liska <mliska@suse.cz>
34766 PR gcov-profile/80081
34767 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34768 * doc/gcc.texi: Include gcov-dump stuff.
34769 * doc/gcov-dump.texi: New file.
34771 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34773 PR rtl-optimization/79150
34774 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34775 conditional jump, if the jump is the last insn of the loop.
34777 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34778 Richard Biener <rguenth@suse.de>
34780 PR tree-optimization/79908
34781 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34782 been cast away, use force_gimple_operand to construct the side
34785 2017-03-21 Martin Liska <mliska@suse.cz>
34787 PR libfortran/79956
34788 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34791 2017-03-21 Brad Spengler <spender@grsecurity.net>
34794 * plugin.c (htab_hash_plugin): New function.
34795 (add_new_plugin): Use it and adjust.
34796 (parse_plugin_arg_opt): Adjust.
34797 (init_one_plugin): Likewise.
34799 2017-03-21 Richard Biener <rguenther@suse.de>
34801 PR tree-optimization/80032
34802 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34803 if set force the cleanup to happen unconditionally.
34804 (gimplify_target_expr): Push inserted clobbers with force_uncond
34805 to avoid them being removed by control-dependent DCE.
34807 2017-03-21 Richard Biener <rguenther@suse.de>
34809 PR tree-optimization/80122
34810 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34811 va_arg_pack_len when the inlined call stmt requires pack
34813 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34815 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34818 * tsan.c (instrument_builtin_call): If the memory model argument
34819 is not a constant, assume it is valid.
34822 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34825 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34827 PR rtl-optimization/79910
34828 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34829 if its dest is used by an insn before I2 (other than the combined
34830 insns themselves, which are properly handled already).
34832 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34835 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34837 * combine.c (record_used_regs): New static function.
34838 (try_combine): Handle situations where there is an additional
34839 instruction between I2 and I3 which needs to have a LOG_LINK
34843 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34845 * combine.c (try_combine): Delete redundant i1 test. Call
34846 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34848 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34851 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
34852 alternatives 13/14.
34854 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34856 PR tree-optimization/80054
34857 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
34858 the optimization if a PHI or any of its arguments is not dominated
34859 by the candidate's basis. Use gphi* rather than gimple* as
34861 (replace_profitable_candidates): Clean up a gimple* variable that
34862 should be a gphi* variable.
34864 2017-03-20 Martin Sebor <msebor@redhat.com>
34867 * doc/extend.texi (attribute constructor): Document present limitation.
34869 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
34872 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
34873 __POWER9_VECTOR__ #ifdef control, change template definition to
34874 use Power9-specific built-in function.
34875 (vec_any_eq): Likewise.
34876 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
34877 to control outcomes from this test.
34878 (vector_ae_<mode>p): For VEC_F modes, likewise.
34880 2017-03-20 Ian Lance Taylor <iant@google.com>
34882 * config/i386/i386.c (ix86_function_regparm): Save an extra
34883 register for -fsplit-stack with DECL_STATIC_CHAIN.
34885 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34888 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
34889 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
34891 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34893 * config/riscv/riscv.c (riscv_print_operand): Use "fence
34895 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
34898 2017-03-20 Marek Polacek <polacek@redhat.com>
34901 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
34903 2017-03-20 Richard Biener <rguenther@suse.de>
34905 PR tree-optimization/80113
34906 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
34907 allocate extra SSA name for PHI def.
34908 (add_close_phis_to_outer_loops): Likewise.
34909 (add_close_phis_to_merge_points): Likewise.
34910 (copy_loop_close_phi_args): Likewise.
34911 (copy_cond_phi_nodes): Likewise.
34913 2017-03-20 Martin Liska <mliska@suse.cz>
34915 PR middle-end/79753
34916 * tree-chkp.c (chkp_build_returned_bound): Do not build
34917 returned bounds for a LHS that's not a BOUNDED_P type.
34919 2017-03-20 Martin Liska <mliska@suse.cz>
34923 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
34924 COMPLEX_CST and VECTOR_CST.
34926 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34929 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
34930 target operand. A new splitter adds the clobber statement in case
34931 the target operand is dead anyway.
34933 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
34935 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
34936 to age-old versions of binutils and glibc.
34938 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
34940 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
34942 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34944 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
34946 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34948 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
34949 requirement for binutils 2.13.
34951 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34953 * combine.c (try_combine): Delete redundant i1 test. Call
34954 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34956 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
34958 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
34959 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
34961 <riscv64-*-elf>: Re-arrange section
34962 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
34963 <riscv32-*-linux>: Likewise.
34964 <riscv64-*-elf>: Likewise
34965 <riscv64-*-linux>: Likewise.
34967 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
34970 * aarch64.opt(verbose-cost-dump): Fix typo.
34972 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
34975 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
34976 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
34978 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34980 * reload.c (find_reloads): When reloading a nonoffsettable address,
34981 use RELOAD_OTHER for it and its address reloads.
34983 PR rtl-optimization/79910
34984 * combine.c (record_used_regs): New static function.
34985 (try_combine): Handle situations where there is an additional
34986 instruction between I2 and I3 which needs to have a LOG_LINK
34989 2017-03-17 Jeff Law <law@redhat.com>
34991 PR tree-optimization/71437
34992 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
34993 conditional in the hash table first.
34994 (vrp_dom_walker::before_dom_children): Extract condition from
34995 ASSERT_EXPR. Record condition, its inverion and any implied
34996 conditions as well.
34998 2017-03-17 Marek Polacek <polacek@redhat.com>
34999 Markus Trippelsdorf <markus@trippelsdorf.de>
35001 PR tree-optimization/80079
35002 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35005 2017-03-17 Richard Biener <rguenther@suse.de>
35007 PR middle-end/80075
35008 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35009 Properly verify the LHS before the RHS possibly claims to be
35011 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35014 2017-03-17 Martin Jambor <mjambor@suse.cz>
35016 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35017 (List of -O2 options): Likewise.
35018 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35021 2017-03-17 Tom de Vries <tom@codesourcery.com>
35023 * gcov-dump.c (print_usage): Print bug_report_url.
35025 2017-03-17 Richard Biener <rguenther@suse.de>
35027 PR middle-end/80050
35028 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35029 (parser::peek): Likewise.
35031 2017-03-17 Richard Biener <rguenther@suse.de>
35033 PR tree-optimization/80048
35034 * sese.c (free_sese_info): Properly release rename_map and
35035 copied_bb_map elements.
35037 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35039 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35040 Add linked-list forward and backlinks. Insert on
35041 construction, remove on destruction.
35042 (class pass_store_merging): Add m_stores_head field.
35043 (pass_store_merging::terminate_and_process_all_chains):
35044 Iterate over m_stores_head list.
35045 (pass_store_merging::terminate_all_aliasing_chains):
35047 (pass_store_merging::execute): Check for debug stmts first.
35048 Push new chains onto the m_stores_head stack.
35050 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35053 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35054 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35055 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35057 2017-03-16 Jeff Law <law@redhat.com>
35059 PR tree-optimization/71437
35060 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35061 member function. Implementation moved into after_dom_children
35062 member function and into the threader's thread_outgoing_edges
35064 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35065 some code into new thread_outgoing_edges.
35066 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35067 definition. Simplify marker handling (do it here). Assume we always
35068 have the available expression and the const/copies tables.
35069 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35071 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35072 * tree-vrp.c (equiv_stack): No longer file scoped.
35073 (vrp_dom_walker): New class.
35074 (vrp_dom_walker::before_dom_children): New member function.
35075 (vrp_dom_walker::after_dom_children): Likewise.
35076 (identify_jump_threads): Setup domwalker. Use it rather than
35077 walking edges in a random order by hand. Simplify setup/finalization.
35078 (finalize_jump_threads): Remove.
35079 (vrp_finalize): Do not call identify_jump_threads here.
35080 (execute_vrp): Do it here instead and call thread_through_all_blocks
35083 PR tree-optimization/71437
35084 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35086 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35088 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35089 (dom_opt_dom_walker::thread_across_edge): Remove
35090 handle_dominating_asserts argument. All callers changed.
35091 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35092 changes. Remove calls to lhs_of_dominating_assert. Other
35093 uses of handle_dominating_asserts turn into unconditional code
35094 (simplify_control_stmt_condition_1): Likewise.
35095 (simplify_control_stmt_condition): Likewise.
35096 (thread_through_normal_block, thread_across_edge): Likewise.
35097 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35098 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35099 object if it is not an SSA_NAME.
35100 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35101 before calling into the VRP specific simplifiers.
35102 (identify_jump_threads): Remove handle_dominating_asserts
35105 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35108 * tree-diagnostic.c (default_tree_printer): No longer static.
35109 * tree-diagnostic.h (default_tree_printer): New prototype.
35111 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35113 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35114 Change ins into fmov.
35116 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35118 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35119 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35120 Use h_con constraint for operand 1.
35121 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35122 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35124 2017-03-15 Jeff Law <law@redhat.com>
35126 PR tree-optimization/71437
35127 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35128 (record_temporary_equivalences): Use it.
35130 PR tree-optimization/71437
35131 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35132 tree-ssa-scopedtables.
35133 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35134 (record_conditions, record_cond, vuse_eq): Likewise.
35135 (record_edge_info): Adjust to API tweak of record_conditions.
35136 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35137 (record_temporary_equivalences, optimize_stmt): Likewise.
35138 (eliminate_redundant_computations): Likewise.
35139 (record_equivalences_from_stmt): Likewise.
35140 * tree-ssa-scopedtables.c: Include options.h and params.h.
35141 (vuse_eq): New function, moved from tree-ssa-dom.c
35142 (build_and_record_new_cond): Likewise.
35143 (record_conditions): Likewise. Accept vector of conditions rather
35144 than edge_equivalence structure for first argument.
35145 for the first argument.
35146 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35147 from tree-ssa-dom.c.
35148 (avail_exprs_stack::record_cond): Likewise.
35149 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35150 from tree-ssa-dom.c.
35151 (avail_exprs_stack): Add new member functions lookup_avail_expr
35153 (record_conditions): Declare.
35155 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35158 * lra-constraints.c (process_alt_operands): Increase reject for
35159 reloading an input/output operand.
35161 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35164 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35165 insns to convert from signed/unsigned char/short to IEEE 128-bit
35167 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35169 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35172 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35173 subreg of inner mode for values already in registers.
35175 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35177 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35178 iteration reg is used after the loop.
35180 2017-03-14 Martin Sebor <msebor@redhat.com>
35182 PR tree-optimization/79800
35183 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35184 precision in negative-positive range.
35185 (format_floating): Call non-const overload with adjusted precision.
35187 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35190 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35193 2017-03-14 Martin Sebor <msebor@redhat.com>
35195 PR middle-end/80020
35196 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35197 * builtins.def (aligned_alloc): Use it.
35200 * Makefile.in (GTFILES): Add calls.c.
35201 * calls.c: Include "gt-calls.h".
35203 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35205 PR rtl-optimization/79728
35206 * regs.h (struct target_regs): New field
35207 x_contains_allocatable_regs_of_mode.
35208 (contains_allocatable_regs_of_mode): New macro.
35209 * reginfo.c (init_reg_sets_1): Initialize it, and change
35210 contains_reg_of_mode so it includes global regs as well.
35211 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35212 rather than contains_regs_of_mode.
35214 2017-03-14 Martin Liska <mliska@suse.cz>
35216 * doc/invoke.texi: Document options that can't be combined with
35217 -fcheck-pointer-bounds.
35219 2017-03-14 Martin Liska <mliska@suse.cz>
35221 PR middle-end/79831
35222 * doc/invoke.texi (-Wchkp): Document the option.
35224 2017-03-14 Martin Liska <mliska@suse.cz>
35226 * Makefile.in: Install gcov-dump.
35228 2017-03-14 Martin Liska <mliska@suse.cz>
35230 * multiple_target.c (expand_target_clones): Bail out for
35231 an invalid attribute.
35233 2017-03-14 Richard Biener <rguenther@suse.de>
35235 * alias.c (struct alias_set_entry): Pack properly.
35236 * cfgloop.h (struct loop): Likewise.
35237 * cse.c (struct set): Likewise.
35238 * ipa-utils.c (struct searchc_env): Likewise.
35239 * loop-invariant.c (struct invariant): Likewise.
35240 * lra-remat.c (struct cand): Likewise.
35241 * recog.c (struct change_t): Likewise.
35242 * rtl.h (struct address_info): Likewise.
35243 * symbol-summary.h (function_summary): Likewise.
35244 * tree-loop-distribution.c (struct partition): Likewise.
35245 * tree-object-size.c (struct object_size_info): Likewise.
35246 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35247 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35248 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35249 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35250 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35251 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35252 (struct _stmt_vec_info): Likewise.
35254 2017-03-14 Martin Liska <mliska@suse.cz>
35257 * multiple_target.c (create_dispatcher_calls): Check that
35258 a target can create a function dispatcher.
35260 2017-03-14 Martin Liska <mliska@suse.cz>
35263 * multiple_target.c (expand_target_clones): Drop local.local
35264 flag for default implementation.
35266 2017-03-14 Richard Biener <rguenther@suse.de>
35268 PR tree-optimization/80030
35269 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35271 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35273 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35274 gcc_fallthrough() instead of __attribute__((fallthrough));
35276 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35278 * doc/gcc.texi: Remove "up" link to (DIR).
35279 * doc/gccint.texi: Ditto.
35281 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35283 * doc/install.texi (Specific) <avr>: Remove reference to
35286 2017-03-13 Jeff Law <law@redhat.com>
35288 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35289 attribute rather than comments.
35291 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35292 match_scratch operand is highest.
35294 2017-03-13 Martin Liska <mliska@suse.cz>
35296 PR middle-end/78339
35297 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35298 is a CHKP clone, use original declaration.
35300 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35302 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35303 (arc_conditional_register_usage): Use a different allocation order
35304 when optimizing for size.
35305 * common/config/arc/arc-common.c (arc_option_optimization_table):
35306 Section anchors default on when optimizing for size.
35308 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35310 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35312 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35314 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35315 * config/arc/arc.md (cpu_facility): Add cd variant.
35316 (*movqi_insn): Add code density variant.
35317 (*movhi_insn): Likewise.
35318 (*movqi_insn): Likewise.
35319 (*addsi3_mixed): Likewise.
35320 (subsi3_insn): Likewise.
35322 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35324 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35326 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35328 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35329 expressions with MINUS and UNARY ops.
35331 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35334 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35336 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35337 between vec_select and vector argument.
35338 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35339 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35340 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35341 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35342 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35343 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35345 2017-03-13 Richard Biener <rguenther@suse.de>
35348 * params.def (vect-max-peeling-for-alignment): Fix typo.
35350 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35352 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35353 issue that only occurred with binutils below 2.18.
35355 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35357 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35358 refer to binutils 2.11/2.12 minimum.
35360 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35362 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35363 ftp.kernel.org and simplify binutils requirement.
35365 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35367 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35369 (Optimize Options): Ditto. Also remove redundancy.
35371 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35373 PR translation/79848
35374 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35376 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35377 to G_ to avoid double translation.
35379 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35381 PR translation/79923
35382 * auto-profile.c (get_combined_location): Convert leading
35383 character of diagnostics to lower case and remove trailing period.
35384 (read_profile): Likewise for various diagnostics.
35385 * config/arm/arm.c (arm_option_override): Remove trailing period
35386 from various diagnostics.
35387 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35388 (msp430_expand_delay_cycles): Likewise.
35390 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35393 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35394 full command-line argument, rather than just "str".
35395 (aarch64_validate_march): Likewise.
35396 (aarch64_validate_mtune): Likewise.
35398 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35400 PR rtl-optimization/78911
35401 * lra-assigns.c (must_not_spill_p): New function.
35402 (spill_for): Use it.
35404 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35406 PR tree-optimization/79981
35407 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35408 ATOMIC_COMPARE_EXCHANGE ifn result.
35409 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35410 IFN_ATOMIC_COMPARE_EXCHANGE.
35412 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35415 * opts.c (parse_sanitizer_options): Add missing question mark to
35416 "did you mean" message.
35418 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35420 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35422 (VMULEUH_UNS): Likewise.
35423 (VMULOUB_UNS): Likewise.
35424 (VMULOUH_UNS): Likewise.
35425 * config/rs6000/rs6000.c (builtin_function_type): Remove
35426 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35428 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35431 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35432 x with result of extra_parsing_for_operand_code_0.
35433 (function_reader::extra_parsing_for_operand_code_0): Convert
35434 return type from void to rtx, returning x. When reading
35435 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35436 larger size containing struct block_symbol.
35438 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35440 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35441 -mfloat128-hardware without -m64.
35443 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35446 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35447 entries to the case statement that marks unsigned arguments to
35448 overloaded functions.
35450 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35452 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35453 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35455 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35458 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35459 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35461 2017-03-10 Martin Liska <mliska@suse.cz>
35465 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35466 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35469 2017-03-10 Olivier Hainque <hainque@adacore.com>
35471 * tree-switch-conversion (array_value_type): Start by resetting
35472 candidate type to it's main variant.
35474 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35476 PR rtl-optimization/79909
35477 * combine.c (try_combine): Use simplify_replace_rtx on individual
35478 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35479 of the whole CALL_INSN_FUNCTION_USAGE.
35481 PR tree-optimization/79972
35482 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35483 get_range_info on SSA_NAMEs. Formatting fixes.
35485 2017-03-10 Richard Biener <rguenther@suse.de>
35486 Jakub Jelinek <jakub@redhat.com>
35488 PR tree-optimization/77975
35489 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35490 edge to be constant.
35491 (get_val_for): For constant x return it. Formatting fix.
35492 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35493 would use the same bases as the current one.
35495 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35497 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35498 instead of vec_select for V1TImode.
35499 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35501 (VSX_LE_128): Add V1TI to this mode iterator.
35502 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35503 (*vsx_le_perm_store_<mode>): Likewise.
35504 (pre-reload splitter for VSX stores): Likewise.
35505 (post-reload splitter for VSX stores): Likewise.
35506 (*vsx_xxpermdi2_le_<mode>): Likewise.
35507 (*vsx_lxvd2x2_le_<mode>): Likewise.
35508 (*vsx_stxvd2x2_le_<mode>): Likewise.
35510 2017-03-09 Michael Eager <eager@eagercon.com>
35512 Correct failures with --enable-checking=yes,rtl.
35514 * config/microblaze/microblaze.c (microblaze_expand_shift):
35515 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35516 test for const0_rtx.
35517 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35518 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35520 2017-03-09 Richard Biener <rguenther@suse.de>
35522 PR tree-optimization/79977
35523 * graphite-scop-detection.c (scop_detection::merge_sese):
35524 Handle the case of extra exits to blocks dominating the entry.
35526 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35528 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35531 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35533 PR rtl-optimization/79949
35534 * lra-constraints.c (process_alt_operands): Check memory when
35535 trying to predict a cycle. Print about the overall increase.
35537 2017-03-09 Richard Biener <rguenther@suse.de>
35539 PR middle-end/79971
35540 * gimple-expr.c (useless_type_conversion_p): Preserve
35541 TYPE_SATURATING for fixed-point types.
35543 2017-03-09 Richard Biener <rguenther@suse.de>
35546 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35547 alignment of BLKmode params.
35549 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35552 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35553 (VALL_NO_V2Q): Likewise.
35555 * config/aarch64/aarch64-simd.md
35556 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35558 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35559 VALL_NO_V2Q mode iterator.
35560 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35562 2017-03-09 Martin Liska <mliska@suse.cz>
35564 PR tree-optimization/79631
35565 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35566 tree_int_cst_sign_bit just for INTEGER constants.
35568 2017-03-09 Martin Liska <mliska@suse.cz>
35572 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35575 2017-03-09 Marek Polacek <polacek@redhat.com>
35578 * tree.c (inchash::add_expr): Handle TREE_VEC.
35580 2017-03-09 Martin Liska <mliska@suse.cz>
35583 (chkp_narrow_size_and_offset): New function.
35584 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35585 (void chkp_parse_bit_field_ref): New function.
35586 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35587 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35589 2017-03-09 Martin Liska <mliska@suse.cz>
35592 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35593 (chkp_find_bounds_1): Remove gcc_unreachable.
35595 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35598 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35599 BUILT_IN_SYNC*, determine the access type from the size suffix and
35600 always build a MEM_REF with that type. Handle forgotten
35601 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35604 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35605 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35606 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35607 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35608 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35609 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35610 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35611 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35612 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35613 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35614 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35615 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35616 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35617 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35618 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35619 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35620 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35621 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35622 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35623 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35624 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35625 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35626 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35627 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35628 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35629 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35630 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35631 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35632 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35633 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35634 definitions outside of __OPTIMIZE__ guarded section.
35637 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35638 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35639 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35640 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35643 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35645 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35646 ("vfenez<mode>"): Add missing constraints.
35648 2017-03-08 Martin Sebor <msebor@redhat.com>
35651 * config/nds32/nds32.c (nds32_option_override):
35652 Fix misspelled diagnostic.
35654 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35657 * gimplify.c (gimplify_omp_for): Replace index var in outer
35658 taskloop statement with an artificial variable and add
35659 OMP_CLAUSE_PRIVATE clause for it.
35661 2017-03-08 Richard Biener <rguenther@suse.de>
35663 PR tree-optimization/79955
35664 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35665 for accesses that are completely outside of the variable.
35667 2017-03-08 Andrew Haley <aph@redhat.com>
35669 PR tree-optimization/79943
35670 * tree-ssa-loop-split.c (compute_new_first_bound): When
35671 calculating the new upper bound, (END-BEG) should be added, not
35674 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35676 * config/avr/avr.md (setmemhi): Make sure match_dup
35677 operand number comes before match_scratch.
35679 2017-03-08 Richard Biener <rguenther@suse.de>
35681 PR tree-optimization/79920
35682 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35683 with ncopies == 1 to ...
35684 (vect_transform_slp_perm_load): ... here. Properly compute
35685 all element loads by iterating VF times over the group. Do
35686 not handle ncopies (computed in a broken way) in
35687 vect_create_mask_and_perm.
35689 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35692 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35693 is a uniform vector, use uniform_vector_p return value instead of
35694 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35696 2017-03-07 Marek Polacek <polacek@redhat.com>
35698 PR middle-end/79809
35699 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35700 (alloca_call_type): Likewise.
35702 2017-03-07 Martin Liska <mliska@suse.cz>
35704 * gcov.c (process_args): Put comment to correct location.
35706 2017-03-07 Martin Liska <mliska@suse.cz>
35708 PR middle-end/68270
35709 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35710 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35711 (chkp_narrow_bounds_for_field): Likewise.
35712 (chkp_parse_array_and_component_ref): Pass one more argument to
35715 2017-03-07 Richard Biener <rguenther@suse.de>
35717 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35720 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35722 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35723 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35725 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35728 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35729 to end of description.
35730 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35732 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35734 PR rtl-optimization/79901
35735 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35737 (*avx512f_<code><mode>3<mask_name>): ... this.
35738 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35739 iterator instead of VI8_AVX2_AVX512BW.
35741 PR rtl-optimization/79901
35742 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35743 min/max expander, expand it using expand_vec_cond_expr.
35746 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35749 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35752 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35753 to void * for PCH reasons.
35754 * dwarf2out.c (output_loc_operands, output_die): Cast
35755 v.val_vec.array to unsigned char *.
35757 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35760 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35763 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35765 PR rtl-optimization/79571
35766 * lra-constraints.c (process_alt_operands): Calculate static
35767 reject and subtract it from overall when only addresses will be
35770 2017-03-06 Julia Koval <julia.koval@intel.com>
35773 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35774 incoming stack boundary to 128 for 64-bit targets.
35776 2017-03-06 Richard Biener <rguenther@suse.de>
35778 PR tree-optimization/79894
35779 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35780 to NULL after folding it.
35782 2017-03-06 Richard Biener <rguenther@suse.de>
35784 PR tree-optimization/79824
35785 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35786 check disabling peeling for gaps.
35788 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35790 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35791 attributes): Document gettimeofday.
35793 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35795 * config/s390/s390.c (s390_option_override_internal): Set
35796 PARAM_MIN_VECT_LOOP_BOUND
35798 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35800 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35801 * config/s390/s390.md: Likewise.
35803 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35806 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35807 (<avx2_avx512>_perm<mode>): Rename to ...
35808 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35810 (<avx512>_perm<mode>_mask): Rename to ...
35811 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35813 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35814 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35815 instead of VI8F_256_512.
35816 (avx512f_perm<mode>): New define_expand.
35817 (avx512f_perm<mode>_mask): Likewise.
35818 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35819 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35821 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35823 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35824 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35826 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35828 2017-03-06 Martin Liska <mliska@suse.cz>
35831 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35832 when having a SSA NAME w/o VAR_DECL assigned to it.
35834 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35836 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35837 msa_dpsub_<su>_d): Fix MODE for vec_select.
35839 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35841 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35843 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35845 2017-03-06 Richard Biener <rguenther@suse.de>
35847 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35848 * plugin.c (register_plugin_info): Likewise.
35849 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
35851 2017-03-05 Jakub Jelinek <jakub@redhat.com>
35853 * config/i386/sse.md (sse_storehps, sse_storelps,
35854 avx_<castmode><avxsizesuffix>_<castmode>,
35855 avx512f_<castmode><avxsizesuffix>_<castmode>,
35856 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
35857 in condition that at least one operand is not a MEM.
35859 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35861 PR middle-end/79805
35862 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
35863 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
35865 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
35866 gimple_call_nothrow_p flag based on whether original builtin can throw.
35867 If it can, emit following stmts on the fallthrough edge.
35868 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
35869 don't create new bb if inserting just debug stmts on the edge, try to
35870 insert them on the fallthru bb or just reset debug stmts.
35872 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
35875 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
35876 restore recog_data (including the operand rtxes inside it) around
35877 the call to get_insn_template.
35879 2017-03-03 Martin Sebor <msebor@redhat.com>
35881 PR tree-optimization/79699
35882 * context.c (context::~context): Free MPFR caches to avoid
35883 a memory leak on program exit.
35885 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35887 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
35888 Use wide_int::ulow () instead of .elt (0).
35890 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35892 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
35893 (*pushxf): Limit oF constraint to 32bit targets and add oC
35894 constraint for 64bit targets.
35895 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
35896 (*pushdf): Change rmF constraint to rmC.
35898 2017-03-03 Martin Liska <mliska@suse.cz>
35900 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
35901 Remove unused variable.
35903 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35906 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
35907 is a memory operand, increase num_memory.
35908 (ix86_expand_args_builtin): Likewise.
35910 2017-03-03 Jan Hubicka <jh@suse.cz>
35913 * ipa-devirt.c (maybe_record_node): Properly handle
35914 __cxa_pure_virtual visibility.
35916 2017-03-03 Martin Liska <mliska@suse.cz>
35918 PR tree-optimization/79803
35919 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
35921 (pass_loop_prefetch::execute): Disabled optimization if an
35922 assumption about L1 cache size is not met.
35924 2017-03-03 Martin Liska <mliska@suse.cz>
35926 PR rtl-optimization/79574
35927 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
35928 (hash_scan_set): Likewise.
35929 (dump_hash_table): Likewise.
35930 (hoist_code): Likewise.
35932 2017-03-03 Richard Biener <rguenther@suse.de>
35934 * fixed-value.c (fixed_from_string): Restore use of elt (1)
35935 in place of uhigh ().
35936 (fixed_convert_from_real): Likewise.
35938 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35941 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
35943 2017-03-03 Richard Biener <rguenther@suse.de>
35945 PR middle-end/79818
35946 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
35947 TYPE_OVERFLOW_UNDEFINED check.
35949 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35951 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
35953 (vector_ae_<mode>_p): Likewise.
35954 (vector_nez_<mode>_p): Likewise.
35955 (vector_ne_v2di_p): Likewise.
35956 (vector_ae_v2di_p): Likewise.
35957 (vector_ne_<mode>_p): Likewise.
35958 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
35960 (vsx_tsqrt<mode>2_fe): Likewise.
35962 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
35965 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
35967 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35969 PR rtl-optimization/79780
35970 * cprop.c (one_cprop_pass): When second and further conditional trap
35971 in a single basic block is turned into an unconditional trap, turn it
35972 into a deleted note to avoid RTL verification failures.
35974 2017-03-02 Richard Biener <rguenther@suse.de>
35976 * fold-const.c (const_binop): Use ulow () instead of elt (0).
35978 2017-03-02 Richard Biener <rguenther@suse.de>
35980 PR tree-optimization/79345
35982 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
35983 param and abort the walk, returning -1 if it is hit.
35984 (walk_aliased_vdefs): Take a limit param and pass it on.
35985 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
35986 defaulting to 0 and return a signed int.
35987 * tree-ssa-uninit.c (struct check_defs_data): New struct.
35988 (check_defs): New helper.
35989 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
35990 about uninitialized memory.
35991 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
35992 bogus uninitialized warning.
35993 (fixed_convert_from_real): Likewise.
35995 2017-03-02 Bin Cheng <bin.cheng@arm.com>
35997 PR tree-optimization/66768
35998 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
35999 iv_use if base object can't be determined.
36001 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36003 PR tree-optimization/79345
36004 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36005 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36006 (get_pattern_stats): Initialize it.
36007 * genemit.c (gen_expand): Verify match_scratch numbers come after
36008 match_operand/match_dup numbers.
36009 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36010 match_scratch numbers.
36011 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36013 * config/s390/s390.md (trunctdsd2): Likewise.
36015 2017-03-02 Richard Biener <rguenther@suse.de>
36017 * wide-int.h (wide_int_storage::operator=): Implement in terms
36020 2017-03-02 Richard Biener <rguenther@suse.de>
36022 PR tree-optimization/79777
36023 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36024 the to insert expression to sth existing.
36026 2017-03-01 Martin Sebor <msebor@redhat.com>
36028 PR middle-end/79692
36029 * gimple-ssa-sprintf.c
36030 (directive::known_width_and_precision): New function.
36031 (format_integer): Use it.
36032 (get_mpfr_format_length): Consider the full range of precision
36033 when computing %g output with the # flag. Set the likely byte
36034 count to 3 rather than 1 when precision is indeterminate.
36035 (format_floating): Correct the lower bound of precision.
36037 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36039 * doc/invoke.texi: Document default code model for 64-bit Linux.
36041 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36044 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36045 udiv rather than div since input pattern is unsigned.
36047 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36049 * config/i386/i386.c (print_reg): Warn for values of
36050 unsupported size in integer register.
36052 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36055 * config/rs6000/predicates.md (current_file_function_operand): Do
36056 not allow self calls to be local if the function is replaceable.
36058 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36061 * config/rs6000/altivec.h (vec_ctz and others): Change the
36062 preprocessor macro that controls conditional compilation from
36063 _ARCH_PWR9 to __POWER9_VECTOR__.
36064 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36065 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36066 control (instead of _ARCH_PWR9 control) so that template
36067 definition uses power9-specific function.
36068 (vec_any_eq): Likewise.
36069 (vec_all_ne): Change macro definition to use a power9-specific
36070 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36071 _ARCH_PWR9 control).
36072 (vec_any_eq) Likewise.
36073 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36074 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36075 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36076 support for xvcmpnedp instruction.
36077 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36078 macro expansion so that Power9 implementation of vec_all_ne does
36079 not use the AltiVec predicate framework.
36080 (VCMPNEH_P): Likewise.
36081 (VCMPNEW_P): Likewise.
36082 (VCMPNED_P): Likewise.
36083 (VCMPNEFP_P): Likewise.
36084 (VCMPNEDP_P): Likewise.
36085 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36086 implementation of vec_any_eq to not use AltiVec predicate
36088 (VCMPAEH_P): Likewise.
36089 (VCMPAEW_P): Likewise.
36090 (VCMPAED_P): Likewise.
36091 (VCMPAEFP_P): Likewise.
36092 (VCMPAEDP_P): Likewise.
36093 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36094 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36095 not use the AltiVec predicate framework.
36096 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36097 of vec_any_eq to not use AltiVec predicate framework.
36098 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36099 support for predefined __POWER9_VECTOR__ macro to indicate that
36100 Power9 instruction selection is enabled.
36101 (altivec_overloaded_builtins): Remove extraneous
36102 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36103 function argument types RS6000_BTI_bool_V16QI and
36104 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36105 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36106 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36107 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36108 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36109 Power9 for implementations of vec_cmpne. Change the signature for
36110 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36111 (representing vec_all_ne) to remove the previously described first
36112 argument of type RS6000_BTI_INTSI, as this was an artifact of
36113 reliance on the AltiVec predicate framework, which is no longer
36114 used in the implementation of these functions. Add
36115 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36116 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36117 since, unlike the AltiVec predicate framework implementation, we
36118 do not share function descriptors between vec_alle and vec_anyeq.
36119 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36120 set of modes that receive special treatment even when
36121 TARGET_P9_VECTOR is true. The special treatment emits code that
36122 does not depend on Power9 instructions.
36123 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36124 define_expand to not rely on AltiVec predicate framework.
36125 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36127 (vector_ne_v2di_p): Change this define_expand to not rely on
36128 AltiVec predicate framework.
36129 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36131 (vector_ne_<mode>_p): Change this define_expand to not rely on
36132 AltiVec predicate framework.
36133 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36135 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36136 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36137 define_insn pattern.
36138 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36139 define_insn pattern because the xvcmpne<VSs>. instruction is not
36141 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36142 instruction is not supported.
36144 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36146 * config/nvptx/nvptx.c: Include intl.h.
36148 2017-03-01 Martin Jambor <mjambor@suse.cz>
36151 * ipa-prop.h (ipa_bits): Removed field known.
36152 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36153 to pointers. Adjusted their comments to warn about their sharing.
36154 (ipcp_transformation_summary): Change bits to a vector of pointers.
36155 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36156 (ipa_get_ipa_bits_for_value): Declare.
36157 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36158 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36159 (ipa_bits_hash_table): Likewise.
36160 (ipa_vr_ggc_hash_traits): Likewise.
36161 (ipa_vr_hash_table): Likewise.
36162 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36163 being pointers and vr_known being removed.
36164 (ipa_set_jf_unknown): Likewise.
36165 (ipa_get_ipa_bits_for_value): New function.
36166 (ipa_set_jfunc_bits): Likewise.
36167 (ipa_get_value_range): New overloaded functions.
36168 (ipa_set_jfunc_vr): Likewise.
36169 (ipa_compute_jump_functions_for_edge): Use the above functions to
36170 construct bits and vr parts of jump functions.
36171 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36172 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36174 (ipcp_grow_transformations_if_necessary): Also allocate
36175 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36177 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36178 them. Fix too long lines.
36179 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36180 vr_known being removed.
36181 (ipa_read_jump_function): Use new setter functions to construct bits
36182 and vr parts of jump functions or set them to NULL.
36183 (write_ipcp_transformation_info): Adjust for bits being pointers.
36184 (read_ipcp_transformation_info): Likewise.
36185 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36187 Include gt-ipa-prop.h.
36188 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36190 (ipcp_store_bits_results): Likewise.
36191 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36192 Do not write to existing jump functions but use a temporary instead.
36194 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36197 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36198 attempt to use its first operand as BIT_FIELD_REF base.
36200 2017-03-01 Richard Biener <rguenther@suse.de>
36202 PR middle-end/79721
36203 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36204 interpolating formula in wrapping arithmetic.
36205 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36207 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36209 PR tree-optimization/79734
36210 * tree-vect-generic.c (expand_vector_condition): Optimize
36211 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36212 Handle VEC_COND_EXPR where comparison has different inner width from
36213 type's inner width.
36215 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36217 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36218 markup, and similar issues. Remove @opindex entries for things
36219 that aren't options. Add missing -mmpy-option entries.
36221 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36223 PR tree-optimization/79737
36224 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36225 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36226 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36227 instead of byte_size. Formatting fix.
36228 (shift_bytes_in_array_right): Formatting fix.
36230 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36233 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36234 condition on optimize for the leaf function test.
36236 2017-02-28 Martin Liska <mliska@suse.cz>
36239 * read-rtl-function.c (function_reader::handle_unknown_directive):
36240 Bail out when one uses -flto.
36242 2017-02-28 Martin Liska <mliska@suse.cz>
36244 * common.opt: Replace space with tabular for options of <number>
36246 * config/i386/i386.opt: Show <number> value for
36247 -mlarge-data-threshold.
36248 * opts.c (print_filtered_help): Do not display number in hexadecimal
36251 2017-02-28 Martin Liska <mliska@suse.cz>
36253 * common.opt: Fix --help=option -Q for options which are of
36256 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36258 * config/i386/i386.c (print_reg): Error out for values
36259 of 8-bit size in invalid integer register.
36261 2017-02-28 Martin Sebor <msebor@redhat.com>
36263 PR tree-optimization/79691
36264 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36266 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36269 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36270 gcc_unreachable with output_operand_lossage.
36272 2017-02-28 Richard Biener <rguenther@suse.de>
36274 PR tree-optimization/79740
36275 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36277 (visit_nary_op): Insert the nary into the hashtable if we
36278 pattern-matched sth.
36279 * tree-ssa-pre.c (eliminate_insert): Robustify.
36281 2017-02-28 Richard Biener <rguenther@suse.de>
36283 PR middle-end/79731
36284 * fold-const.c (decode_field_reference): Reject out-of-bound
36287 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36289 * config/i386/i386.c: Include intl.h.
36290 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36291 instead of just cond ? "..." : "...".
36292 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36293 * coverage.c (read_counts_file): Likewise.
36294 * omp-offload.c: Include intl.h.
36295 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36296 of just cond ? "..." : "...".
36297 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36298 of just cond ? "..." : "...".
36300 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36303 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36305 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36306 'tune for' CPU name.
36307 * config/arm/arm-cpu-data.h: Regenerated.
36309 2017-02-28 Richard Biener <rguenther@suse.de>
36311 PR tree-optimization/79732
36312 * tree-inline.c (expand_call_inline): Do not shadow var.
36314 2017-02-28 Richard Biener <rguenther@suse.de>
36316 PR tree-optimization/79723
36317 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36318 address-space properly.
36320 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36322 * doc/optinfo.texi (Optimization groups): Fix option used for
36324 * doc/invoke.texi (-fopt-info): Document "omp".
36325 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36326 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36327 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36328 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36329 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36331 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36333 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36334 OPTGROUP_OMP with "omp".
36336 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36339 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36340 for arithmetic shift of unsigned V2DI.
36342 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36344 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36345 arc/linux.h headers.
36346 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36347 (LINK_SPEC): Likewise.
36348 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36349 (EXTRA_SPECS): Likewise.
36350 (STARTFILE_SPEC): Likewise.
36351 (ENDFILE_SPEC): Likewise.
36352 (LIB_SPEC): Likewise.
36353 (TARGET_SDATA_DEFAULT): Likewise.
36354 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36355 (MULTILIB_DEFAULTS): Likewise.
36356 (DWARF2_UNWIND_INFO): Likewise.
36357 * config/arc/big.h: New file.
36358 * config/arc/elf.h: Likewise.
36359 * config/arc/linux.h: Likewise.
36360 * config/arc/t-uClibc: Remove.
36362 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36364 PR tree-optimization/77536
36365 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36366 (tree_transform_and_unroll_loop): Use above function to compute the
36367 estimated niter of unrolled loop and use it when scaling profile.
36368 Also use count info rather than frequency if it's non-zero.
36369 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36370 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36371 (vect_transform_loop): Call above function.
36373 2017-02-27 Richard Biener <rguenther@suse.de>
36375 PR tree-optimization/45397
36376 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36377 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36378 (visit_nary_op): Add pattern matching for CSEing sign-changed
36379 or truncated operations with wider ones.
36381 2017-02-27 Richard Biener <rguenther@suse.de>
36383 PR tree-optimization/79690
36384 * tree-vect-stmts.c (vectorizable_store): Use vector type
36385 built from the DR with address-space.
36387 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36389 * doc/invoke.texi (Optimize Options): Refine the description
36390 of asan-use-after-return.
36392 2017-02-25 Alan Modra <amodra@gmail.com>
36394 PR rtl-optimization/79584
36395 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36396 base, not ad->base_term, the reg within base. Remove assertion
36397 that ad->base == ad->base_term. Replace gen_int_mode using
36398 bogus mode with const0_rtx.
36400 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36402 PR middle-end/79396
36403 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36404 FMA_EXPR like tcc_binary or tcc_unary.
36406 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36409 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36411 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36412 (output_loc_operands): Handle DW_OP_call_ref and
36413 DW_OP_GNU_variable_value.
36414 (struct variable_value_struct): New type.
36415 (struct variable_value_hasher): Likewise.
36416 (variable_value_hash): New variable.
36417 (string_types): Remove.
36418 (copy_loc_descr): New function.
36419 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36420 (prepend_loc_descr_to_each): New function.
36421 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36422 instead of add_loc_descr_to_each if the first argument is single
36423 location list and the second has multiple.
36424 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36425 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36426 when looking for variable value which doesn't have other location info.
36427 (loc_list_from_tree): Formatting fix.
36428 (gen_array_type_die): Simplify DW_AT_string_length handling.
36429 (adjust_string_types): Remove.
36430 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36431 string_types. Call resolve_variable_values.
36432 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36433 (resolve_addr_in_expr): Likewise. Add A argument.
36434 (copy_deref_exprloc): Remove deref argument. Adjust for the
36435 original expression being DW_OP_GNU_variable_value with optionally
36436 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36437 optionally after it.
36438 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36439 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36440 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36441 (variable_value_hasher::hash, variable_value_hasher::equal): New
36443 (resolve_variable_value_in_expr, resolve_variable_value,
36444 resolve_variable_values, note_variable_value_in_expr,
36445 note_variable_value): New functions.
36446 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36449 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36452 * opts.h (handle_generated_option): Add GENERATED_P argument.
36453 * opts-common.c (handle_option): Adjust function comment.
36454 (handle_generated_option): Add GENERATED_P argument, pass it to
36456 (control_warning_option): Pass false to handle_generated_option
36458 * opts.c (maybe_default_option): Pass true to handle_generated_option
36460 * optc-gen.awk: Likewise.
36462 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36464 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36465 a REG, look at the REG it is a SUBREG of.
36466 (splitter for cmpeqsi_t): Ditto.
36468 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36470 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36471 the special USEs with the pattern of the insn, not the insn itself.
36473 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36476 * doc/invoke.texi: Document -mload-store-pairs.
36478 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36479 Sandra Loosemore <sandra@codesourcery.com>
36481 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36482 argument isn't a CONST_INT.
36483 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36484 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36485 (nios2_validate_compare): Bypass alternate compare logic if *op2
36486 is not a CONST_INT.
36487 (ldstwm_operation_p): Return false if first_base is not a REG or
36488 if first_offset is not a CONST_INT.
36490 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36492 * config/cris/cris.md: Use correct operand in a define_peephole2.
36494 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36496 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36498 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36500 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36501 this_insn if it is an INSN or JUMP_INSN.
36502 (force_offsettable): Look at base, not at addr.
36503 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36504 on things that aren't necessarily CONST_INTs.
36506 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36508 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36509 -mfpmath=sse is the default also for x86-32 targets with SSE2
36510 instruction set when @option{-ffast-math} is enabled
36512 2017-02-24 Jeff Law <law@redhat.com>
36514 PR rtl-optimizatoin/79286
36515 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36518 2017-02-24 Richard Biener <rguenther@suse.de>
36520 PR tree-optimization/79389
36521 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36524 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36526 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36527 function comment to reflect reality.
36528 (loop_exits_before_overflow): Fix typo in function description.
36530 2017-02-24 Richard Biener <rguenther@suse.de>
36532 PR tree-optimization/79389
36533 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36534 properly that a threading opportunity exists. Detect conditional
36535 copy/constant propagation opportunities.
36537 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36539 * config/visium/visium.md (type): Add trap.
36540 (b): New mode attribute.
36541 (*btst): Rename into...
36542 (*btst<mode>): ...this and adjust.
36543 (*cbranchsi4_btst_insn): Rename into...
36544 (*cbranch<mode>4_btst_insn): ...this and adjust.
36545 (trap): New define_insn.
36547 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36549 PR tree-optimization/79389
36550 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36551 (noce_reversed_cond_code): New function.
36552 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36553 reversed_comparison_code. Formatting fix.
36554 (noce_try_store_flag): Test rev_cond != NULL in addition to
36555 reversed_comparison_code.
36556 (noce_try_store_flag_constants): Likewise.
36557 (noce_try_store_flag_mask): Likewise.
36558 (noce_try_addcc): Use rev_cond if non-NULL instead of
36559 reversed_comparison_code.
36560 (noce_try_cmove_arith): Likewise. Formatting fixes.
36561 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36562 (noce_find_if_block): Initialize rev_cond.
36563 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36564 instead of false as last argument never attempt to reverse it
36567 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36569 PR tree-optimization/79663
36570 * tree-predcom.c (combine_chains): Process refs in reverse order
36571 only for ZERO length chains, and add explaining comment.
36573 2017-02-23 Jeff Law <law@redhat.com>
36575 PR tree-optimization/79578
36576 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36577 in call to operand_equal_p.
36579 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36582 * config/i386/cpuid.h: Fix another undefined behavior.
36584 2017-02-23 Richard Biener <rguenther@suse.de>
36586 PR tree-optimization/79683
36587 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36588 vector types for data-refs.
36590 2017-02-23 Martin Liska <mliska@suse.cz>
36592 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36594 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36596 PR middle-end/79665
36597 * internal-fn.c (get_range_pos_neg): Moved to ...
36598 * tree.c (get_range_pos_neg): ... here. No longer static.
36599 * tree.h (get_range_pos_neg): New prototype.
36600 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36601 are known to be in between 0 and signed maximum inclusive, try to
36602 expand both unsigned and signed divmod and use the cheaper one from
36605 2017-02-22 Jeff Law <law@redhat.com>
36607 PR tree-optimization/79578
36608 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36609 to compare base operands.
36611 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36614 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36615 gpc_reg_operand instead of fpr_reg_operand.
36617 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36619 * config/mips/mips.c (mips_return_in_memory): Force FP
36620 vector types to be returned in memory for o32 ABI.
36622 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36624 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36625 instead of DW_TAG_member for static data member declarations and don't
36626 set no_linkage_name for static inline data members.
36627 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36630 2017-02-22 Martin Liska <mliska@suse.cz>
36632 * doc/invoke.texi: Replace inequality signs with square brackets
36635 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36638 * lra-constraints.c (simplify_operand_subreg): Handle
36639 WORD_REGISTER_OPERATIONS targets.
36641 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36644 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36645 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36646 elimination by swapping fld*.
36648 2017-02-22 Richard Biener <rguenther@suse.de>
36650 PR tree-optimization/79673
36651 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36652 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36653 irrelevant address-space qualifiers and avoiding a
36654 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36656 2017-02-22 Richard Biener <rguenther@suse.de>
36658 PR tree-optimization/79666
36659 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36660 to not symbolically negate if that may introduce undefined
36663 2017-02-22 Martin Liska <mliska@suse.cz>
36666 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36667 * data-streamer-out.c (streamer_write_gcov_count_stream):
36669 * value-prof.c (stream_out_histogram_value): Make assert more
36670 precise based on type of counter.
36672 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36675 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36676 Use nonimmediate_operand instead of memory_operand for operand 1.
36677 (float-extend standard_x87sse_constant_load splitter): Ditto.
36679 2017-02-21 Jeff Law <law@redhat.com>
36681 PR tree-optimization/79621
36682 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36683 blocks with edges to themselves.
36685 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36688 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36689 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36690 Use gimple_call_builtin_p.
36693 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36694 on temporarily removed DEBUG_INSNs.
36696 PR tree-optimization/79649
36697 * tree-loop-distribution.c (classify_partition): Give up on
36698 non-generic address space loads/stores.
36700 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36702 * doc/loop.texi (Loop manipulation): Remove nonexistent
36703 tree_ssa_loop_version from the documentation.
36704 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36706 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36709 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36710 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36711 * config/rs6000/rs6000.c: Include except.h.
36712 (rs6000_expand_split_stack_prologue): Call
36713 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36715 2017-02-21 Martin Jambor <mjambor@suse.cz>
36718 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36719 have been analyzed.
36721 2017-02-21 Martin Jambor <mjambor@suse.cz>
36723 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36724 for backward compatibility only.
36725 * doc/invoke.texi (Option Summary): Remove all references to
36726 -fipa-cp-alignment.
36728 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36732 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36734 * lra-constraints.c (curr_insn_transform): Handle
36735 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36737 2017-02-21 Martin Liska <mliska@suse.cz>
36739 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36741 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36743 PR translation/79638
36744 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36746 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36749 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36750 (arm_function_ok_for_sibcall): Return false for an indirect call by
36751 descriptor if all the argument registers are used.
36752 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36753 alignment of the function.
36755 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36757 PR tree-optimization/61441
36758 * simplify-rtx.c (simplify_const_unary_operation): For
36759 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36760 the sNaN unmodified.
36762 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36764 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36765 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36766 instead of SYSTEM_HEADER_DIR.
36768 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36769 Martin Liška <mliska@suse.cz>
36771 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36772 Fix typos and grammar, use active voice, and clarify.
36774 2017-02-20 Marek Polacek <polacek@redhat.com>
36776 PR middle-end/79537
36777 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36780 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36782 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36785 * config/i386/i386.c (ix86_expand_builtin): Handle
36786 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36787 ix86_builtins_isa[fcode].isa as a requirement of those
36788 flags and any other flag in the bitmask.
36789 (ix86_init_mmx_sse_builtins): Use 0 instead of
36790 ~OPTION_MASK_ISA_64BIT as mask.
36791 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36792 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36793 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36794 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36796 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36799 * lra-constraints.c (split_reg): Check requested split mode
36800 is supported by the register.
36802 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36804 * lra-constraints.c (simplify_operand_subreg): Remove early
36807 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36810 * lra-constraints.c (curr_insn_transform): Tighten condition
36811 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36813 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36816 * lra-constraints.c (curr_insn_transform): Handle
36817 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36819 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36822 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36824 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36826 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36829 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36832 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36834 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36837 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36839 * final.c (last_columnnum, override_columnnum): New variables.
36840 (final_start_function): Set last_columnnum, pass it to begin_prologue
36841 hook and pass 0 to dwarf2out_begin_prologue.
36842 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36843 to source_line debug hook.
36844 (notice_source_line): Compute last_columnnum and for debug_column_info
36845 return true on column changes.
36846 * debug.h (struct gcc_debug_hooks): Add column argument to
36847 source_line and begin_prologue hooks.
36848 (debug_nothing_int_charstar_int_bool): Remove prototype.
36849 (debug_nothing_int_int_charstar,
36850 debug_nothing_int_int_charstar_int_bool): New prototypes.
36851 (dwarf2out_begin_prologue): Add column argument.
36852 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36853 begin_prologue hooks.
36854 (debug_nothing_int_charstar_int_bool): Remove.
36855 (debug_nothing_int_int_charstar,
36856 debug_nothing_int_int_charstar_int_bool): New functions.
36857 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
36858 through to dwarf2out_source_line.
36859 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
36860 (dwarf2out_source_line): Add column argument, emit it if requested.
36861 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
36863 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36864 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36865 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
36866 through to dwarf2out_begin_prologue.
36867 (vmsdbgout_source_line): Add column argument, pass it through to
36868 dwarf2out_source_line.
36869 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
36870 dbxout_source_line caller.
36871 (dbxout_source_line): Add column argument.
36873 * common.opt (gno-column-info, gcolumn-info): New options.
36874 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
36875 (check_die): Also test for multiple DW_AT_decl_column attributes.
36876 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
36877 DW_AT_decl_column if requested.
36878 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
36880 (gen_variable_die): Likewise.
36881 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
36882 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
36885 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
36886 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
36887 (ix86_handle_option): Handle OPT_m3dnowa.
36888 * doc/invoke.texi (-m3dnowa): Document.
36889 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
36890 -m3dnowa instead of -m3dnow -march=athlon.
36893 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
36894 instead of gcc_assert for K, r and R code checks. Formatting fixes.
36896 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36899 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
36900 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
36901 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
36902 generator for vsx_xxpermdi_<mode>_be.
36903 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
36904 force big-endian semantics.
36905 (vsx_xxpermdi_<mode>_be): New define_expand with same
36906 implementation as previous version of vsx_xxpermdi_<mode>.
36908 2017-02-17 Jakub Jelinek <jakub@redhat.com>
36910 PR tree-optimization/79327
36911 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
36912 variable, its initialization and use.
36914 2017-02-17 Julia Koval <julia.koval@intel.com>
36916 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
36917 (OPTION_MASK_ISA_PKU_UNSET): New.
36918 (ix86_handle_option): Handle -mrdpid.
36919 * config/i386/cpuid.h (bit_RDPID): New.
36920 * config/i386/driver-i386.c (host_detect_local_cpu):
36921 Detect RDPID feature.
36922 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
36923 * config/i386/i386-c.c (ix86_target_macros_internal):
36925 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
36926 (ix86_valid_target_attribute_inner_p): Add "rdpid".
36927 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
36928 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
36929 * config/i386/i386.md (define_insn "rdpid"): New.
36930 * config/i386/i386.opt Add -mrdpid.
36931 * config/i386/immintrin.h (_rdpid_u32): New.
36933 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
36935 PR rtl-optimization/79541
36936 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
36937 instead of transforming it into USE.
36939 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
36941 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
36942 If HONOR_SNANS (SFmode) force the input to a register.
36943 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
36944 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
36945 an frsp or similar insn.
36947 2017-02-17 Martin Liska <mliska@suse.cz>
36949 PR rtl-optimization/79577
36950 * params.def (selsched-max-sched-times): Increase minimum to 1.
36952 2017-02-17 Martin Liska <mliska@suse.cz>
36954 PR rtl-optimization/79574
36955 * gcse.c (want_to_gcse_p): Prevent integer overflow.
36957 2017-02-17 Martin Liska <mliska@suse.cz>
36959 PR tree-optimization/79529
36960 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
36961 ssa_defined_default_def_p to handle cases which are implicitly
36963 * tree-ssa.c (ssa_defined_default_def_p): New function.
36964 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
36965 which are implicitly defined.
36966 * tree-ssa.h (ssa_defined_default_def_p): Declare.
36968 2017-02-17 Richard Biener <rguenther@suse.de>
36970 PR middle-end/79576
36971 * params.def (max-ssa-name-query-depth): Limit to 10.
36973 2017-02-17 Richard Biener <rguenther@suse.de>
36975 PR tree-optimization/79552
36976 * tree-ssa-structalias.c (visit_loadstore): Properly verify
36979 2017-02-17 Richard Biener <rguenther@suse.de>
36982 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
36984 2017-02-17 Marek Polacek <polacek@redhat.com>
36986 PR middle-end/79536
36987 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
36988 (fold_negate_expr): New wrapper.
36990 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
36992 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
36993 Correct terminology and de-emphasize pre-standard behavior.
36995 2017-02-16 Alan Modra <amodra@gmail.com>
36997 PR rtl-optimization/79286
36998 * ira.c (def_dominates_uses): New function.
36999 (update_equiv_regs): Don't create an equivalence for insns that
37000 may trap where the register def does not dominate the use.
37002 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37004 PR rtl-optimization/78127
37005 * lra.c (lra): Call lra_eliminate before finish the loop after
37008 2017-02-16 Richard Biener <rguenther@suse.de>
37010 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37012 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37013 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37014 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37015 (isl_val_int_from_wi): New function.
37016 (extract_affine_gmp): Rename to ...
37017 (extract_affine_wi): ... this, take a widest_int.
37018 (extract_affine_int): Just wrap extract_affine_wi.
37019 (add_param_constraints): Use isl_val_int_from_wi.
37020 (add_loop_constraints): Likewise, and extract_affine_wi.
37022 2017-02-15 Jeff Law <law@redhat.com>
37024 PR middle-end/79521
37025 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37026 ira_init_register_move_cost_if_necessary.
37028 2017-02-15 Martin Sebor <msebor@redhat.com>
37030 PR middle-end/32003
37031 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37032 removed in a prior commit.
37034 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37036 PR tree-optimization/79347
37037 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37038 counters during peeling.
37040 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37042 * Makefile.in (site.exp): Remove "set ISLVER".
37044 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37047 * real.c (real_from_integer): Call real_convert even for decimal.
37049 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37052 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37054 2017-02-14 Andrew Pinski <apinski@cavium.com>
37056 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37057 cores and change the partno/implementer to be correct.
37058 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37059 the 'B" as the implementer.
37060 * config/aarch64/aarch64-tune.md: Regenerate.
37062 2017-02-14 Carl Love <cel@us.ibm.com>
37064 * config/rs6000/rs6000.c: Add case statement entry to make the
37065 xvcvuxdsp built-in argument unsigned.
37066 * config/rs6000/vsx.md: Fix the source and return operand types so they
37067 match the instruction definitions from the ISA document. Fix typo
37068 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37071 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37074 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37075 member early_clobber_alts.
37076 * lra-lives.c (reg_early_clobber_p): New.
37077 (process_bb_lives): Use it.
37078 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37079 (debug_operand_data): Initialize early_clobber_alts.
37080 (setup_operand_alternative): Set up early_clobber_alts.
37081 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37082 alternatives to new_insn_reg.
37083 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37085 (lra_update_insn_regno_info): Pass the new arg.
37087 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37089 PR middle-end/79505
37090 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37091 (new_oacc_loop_raw): Don't clear already cleared fields.
37094 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37095 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37096 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37098 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37101 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37103 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37106 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37107 the extra instruction to the right place to store 128-bit constant
37110 2017-02-14 Martin Sebor <msebor@redhat.com>
37112 PR middle-end/79448
37113 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37114 warning for strings of unknown length.
37116 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37118 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37120 2017-02-14 Jeff Law <law@redhat.com>
37123 * ira-costs.c (scan_one_insn): Initialize register move costs
37124 for pseudos seen in USE/CLOBBER insns.
37126 PR tree-optimization/79095
37127 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37128 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37129 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37130 if the operands are known to be not equal, then the resulting range
37132 (intersect_ranges): If the new range is ~[0,0] and the old range is
37133 wide, then prefer ~[0,0].
37134 * tree-vrp.c (overflow_comparison_p_1): New function.
37135 (overflow_comparison_p): New function.
37136 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37137 if NAME is used in an overflow test.
37138 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37139 overflow check that can be expressed as an equality test, then adjust
37140 ops to be that equality test.
37142 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37144 * config/s390/s390-builtin-types.def: Remove flags argument.
37145 * config/s390/s390.c (s390_init_builtins): Likewise.
37147 2017-02-14 Martin Liska <mliska@suse.cz>
37149 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37150 vector. Fix trailing white spaces.
37152 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37154 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37157 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37159 PR rtl-optimization/68664
37160 * config/arm/arm.c (arm_sched_can_speculate_insn):
37161 New function. Declare prototype.
37162 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37164 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37166 PR rtl-optimization/68664
37167 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37169 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37171 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37173 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37174 max skip bytes for function, loop and jump.
37176 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37178 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37179 ABS_EXPR for gimple dump.
37181 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37184 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37186 PR tree-optimization/79408
37187 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37188 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37189 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37190 also if rhs1 is INTEGER_CST.
37192 2017-02-14 Richard Biener <rguenther@suse.de>
37194 PR middle-end/79432
37195 * tree-into-ssa.c (insert_phi_nodes): When the function can
37196 have abnormal edges rewrite SSA names with broken use-def
37197 dominance out of SSA and register them for PHI insertion.
37199 2017-02-13 Martin Sebor <msebor@redhat.com>
37201 PR middle-end/79496
37202 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37203 clearing info.nowrite flag when snprintf size argument is a range.
37205 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37207 * cprop.c (cprop_jump): Add missing space in string literal.
37208 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37209 (get_constraint_for_component_ref): Likewise.
37210 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37211 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37212 * lra-constraints.c (process_alt_operands): Likewise.
37213 * ipa-inline.c (inline_small_functions): Likewise.
37214 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37215 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37216 * trans-mem.c (diagnose_tm_1_op): Likewise.
37217 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37218 (grid_parallel_clauses_gridifiable): Likewise.
37220 * config/nvptx/mkoffload.c (process): Add space in between
37223 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37224 "MOD4_SSE_REGS" and "ALL_REGS".
37226 * spellcheck.c (test_data): Add , in between "foo" and "food".
37228 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37231 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37232 boundary crossing check and subsequent code generation agree.
37234 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37236 * config/aarch64/aarch64.c (has_memory_op): Delete.
37237 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37240 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37242 PR rtl-optimization/79388
37243 PR rtl-optimization/79450
37244 * combine.c (distribute_notes): When removing TEM_INSN for which
37245 corresponding dest has last value recorded, invalidate that last
37248 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37250 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37251 of explicit '@'. Add missing assembly comment marker on branch costs
37254 2017-02-13 Nathan Sidwell <nathan@acm.org>
37256 * gengtype-lex.l (<in_struct>): Add '/'.
37258 2017-02-13 Martin Liska <mliska@suse.cz>
37261 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37263 2017-02-13 Richard Biener <rguenther@suse.de>
37265 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37267 * configure: Re-generate.
37268 * config.in: Likewise.
37269 * graphite-dependences.c: Simplify as if
37270 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37271 * graphite-isl-ast-to-gimple.c: Likewise.
37272 * graphite-optimize-isl.c: Likewise.
37273 * graphite-poly.c: Likewise.
37274 * graphite-sese-to-poly.c: Likewise.
37275 * graphite.h: Likewise.
37276 * toplev.c: Include isl/version.h and use isl_version () for
37277 printing the ISL version.
37278 * doc/install.texi: Update ISL requirement.
37280 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37282 * doc/standards.texi (Standards): Update reference to
37285 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37287 * doc/extend.texi (Named Address Spaces): sourceware.org now
37289 * doc/install.texi (Binaries): Ditto.
37292 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37294 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37295 standard terminology "stringize"/"stringizing" throughout.
37296 * doc/cppinternals.texi: Likewise.
37298 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37300 * doc/extend.texi: Fix some spelling mistakes and typos.
37301 * doc/invoke.texi: Likewise.
37303 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37306 * params.def (inline-min-speedup) Change from 10 to 8.
37308 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37310 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37313 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37316 * ipa-inline-analysis.c (get_minimal_bb): New function.
37317 (record_modified): Use it.
37318 (remap_edge_change_prob): Handle also ancestor functions.
37320 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37322 * doc/contrib.texi (Contributors): Remove broken link into
37323 the Mauve CVS repository.
37325 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37327 PR middle-end/79454
37328 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37329 result computation whenever lhs doesn't have vector mode, not
37330 just when it has BLKmode.
37332 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37334 * doc/makefile.texi (profiledbootstrap): Refer to the
37335 installation instructions only in textual form.
37337 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37340 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37342 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37344 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37345 (Specific): Update mingw-w64 reference.
37347 (Specific): Remove broken link to Renesas RX processor.
37349 2017-02-10 Richard Biener <rguenther@suse.de>
37351 * toplev.c (process_options): Do not mention obsolete graphite
37352 options when printing sorry message about missing graphite support.
37353 Mention -floop-nest-optimize.
37355 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37357 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37358 (vtst_p16): Likewise.
37359 (vtstq_p8): Likewise.
37360 (vtstq_p16): Likewise.
37362 (vtstq_p64): Likewise.
37363 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37364 (vset_lane_p64): New.
37365 (vsetq_lane_p64): New.
37367 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37369 PR tree-optimization/79411
37370 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37371 stmt operands are SSA_NAMEs used in abnormal phis.
37372 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37375 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37378 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37381 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37383 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37385 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37387 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37391 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37392 not arbitrary TREE_CONSTANT.
37395 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37396 "omp declare target link" attribute unless is_global_var.
37397 * omp-offload.c (find_link_var_op): Likewise.
37399 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37400 Chung-Lin Tang <cltang@codesourcery.com>
37402 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37404 (gimplify_adjust_omp_clauses): Don't delete TILE.
37405 (gimplify_omp_for): Deal with TILE.
37406 * internal-fn.c (expand_GOACC_TILE): New function.
37407 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37409 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37410 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37412 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37413 avoid DIV for outermost collapse var.
37414 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37415 Remove out of date comments, fix whitespace.
37416 * omp-general.c (omp_extract_for_data): Deal with tiling.
37417 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37418 adjust OLF_DIM_BASE value.
37419 (struct omp_for_data): Add tiling field.
37420 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37421 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37422 for auto loops. Remove default auto determining, moved to
37423 oacc_loop_fixed_partitions.
37424 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37425 stmts, add e_mask field.
37426 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37427 (oacc_thread_numbers): Use oacc_dim_call.
37428 (oacc_xform_tile): New.
37429 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37430 (finish_oacc_loop): Adjust for ifns vector.
37431 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37432 add case for GOACC_TILE fns.
37433 (oacc_loop_xform_loop): Delete.
37434 (oacc_loop_process): Iterate over call list directly, and add
37435 handling for GOACC_TILE fns.
37436 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37438 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37439 vector partitioning to outer loops. Assign 2 partitions to loops
37440 when available. Add TILE handling.
37441 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37442 (execite_oacc_device_lower): Process GOACC_TILE fns,
37443 ignore unknown specs.
37444 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37445 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37446 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37448 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37450 * configure.ac (ACX_BUGURL): Update.
37451 * configure: Regenerate.
37453 2017-02-09 Richard Biener <rguenther@suse.de>
37455 PR tree-optimization/69823
37456 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37457 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37459 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37461 * config/arc/arc-c.def: Add __NPS400__ definition.
37462 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37463 (TARGET_NPS400): Define.
37465 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37467 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37469 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37470 pointer, arch_info.
37471 (arc_cpu_types): Fill the arch_info field with a pointer into the
37472 arc_arch_types table.
37473 (arc_selected_cpu): Declare.
37474 * config/arc/arc.c (arc_selected_cpu): Make global.
37475 (arc_selected_arch): Delete.
37476 (arc_base_cpu): Delete.
37477 (arc_override_options): Remove references to deleted variables,
37478 update access to arch information.
37479 (ARC_OPT): Update access to arch information.
37480 (ARC_OPTX): Likewise.
37481 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37482 (TARGET_ARC600): Update access to arch information.
37483 (TARGET_ARC601): Likewise.
37484 (TARGET_ARC700): Likewise.
37485 (TARGET_EM): Likewise.
37486 (TARGET_HS): Likewise.
37487 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37490 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37493 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37494 condition/operands for integer GE/LE/GEU/LEU operations.
37496 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37498 PR translation/79397
37499 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37502 2017-02-08 Martin Jambor <mjambor@suse.cz>
37505 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37506 whether allocation happened.
37507 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37508 nothing was allocated.
37510 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37512 PR tree-optimization/79408
37513 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37514 constant, but SSA_NAME with a known integer range, use the minimum
37515 of that range instead of op1 to determine if modulo can be replaced
37516 with its first operand.
37518 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37520 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37522 2017-02-08 Richard Biener <rguenther@suse.de>
37524 PR tree-optimization/71824
37525 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37526 Check all loops contained in the merged region.
37528 2017-02-07 Andrew Pinski <apinski@cavium.com>
37530 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37532 2017-02-07 Andrew Pinski <apinski@cavium.com>
37534 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37535 (thunderxt88): Likewise.
37536 (thunderxt81): Disable LSE and change v8.1 to v8.
37537 (thunderxt83): Likewise.
37539 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37540 Richard Biener <rguenther@suse.de>
37542 PR middle-end/79399
37543 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37544 type from int to size_t.
37545 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37547 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37549 PR rtl-optimization/79386
37550 * cprop.c (bypass_conditional_jumps): Initialize
37551 bypass_last_basic_block already before splitting bbs after
37552 unconditional traps...
37553 (bypass_conditional_jumps): ... rather than here.
37556 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37557 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37558 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37559 fix -masm=intel patterns.
37561 2017-02-07 Richard Biener <rguenther@suse.de>
37563 PR tree-optimization/79256
37564 PR middle-end/79278
37565 * builtins.c (get_object_alignment_2): Use min_align_of_type
37566 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37567 and ADJUST_FIELD_ALIGN.
37569 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37571 * doc/tm.texi: Regenerate.
37572 * stor-layout.c (layout_decl): Adjust.
37573 (update_alignment_for_field): Likewise.
37574 (place_field): Likewise.
37575 (min_align_of_type): Likewise.
37576 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37577 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37578 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37579 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37580 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37581 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37582 * config/i386/i386.c (x86_field_alignment): Likewise.
37583 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37584 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37585 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37586 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37587 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37588 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37592 2017-01-30 Richard Biener <rguenther@suse.de>
37594 PR tree-optimization/79256
37595 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37596 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37599 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37601 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37602 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37603 builtins to SImode and emit a zero-extend, if necessary.
37605 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37607 * docs/invoke.texi (RISC-V Options): Alphabetize.
37609 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37611 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37614 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37616 * config/riscv/riscv.c: New file.
37617 * common/config/riscv/riscv-common.c: Likewise.
37618 * config.gcc: Likewise.
37619 * config/riscv/constraints.md: Likewise.
37620 * config/riscv/elf.h: Likewise.
37621 * config/riscv/generic.md: Likewise.
37622 * config/riscv/linux.h: Likewise.
37623 * config/riscv/multilib-generator: Likewise.
37624 * config/riscv/peephole.md: Likewise.
37625 * config/riscv/pic.md: Likewise.
37626 * config/riscv/predicates.md: Likewise.
37627 * config/riscv/riscv-builtins.c: Likewise.
37628 * config/riscv/riscv-c.c: Likewise.
37629 * config/riscv/riscv-ftypes.def: Likewise.
37630 * config/riscv/riscv-modes.def: Likewise.
37631 * config/riscv/riscv-opts.h: Likewise.
37632 * config/riscv/riscv-protos.h: Likewise.
37633 * config/riscv/riscv.h: Likewise.
37634 * config/riscv/riscv.md: Likewise.
37635 * config/riscv/riscv.opt: Likewise.
37636 * config/riscv/sync.md: Likewise.
37637 * config/riscv/t-elf-multilib: Likewise.
37638 * config/riscv/t-linux: Likewise.
37639 * config/riscv/t-linux-multilib: Likewise.
37640 * config/riscv/t-riscv: Likewise.
37641 * configure.ac: Likewise.
37642 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37643 Waterman as RISC-V maintainers.
37644 * doc/install.texi: Add RISC-V entries.
37645 * doc/invoke.texi: Add RISC-V options section.
37646 * doc/md.texi: Add RISC-V constraints section.
37647 * configure: Regenerated.
37649 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37652 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37653 false values to be constant vectors with all 0 or all 1 bits set.
37654 (vcondu<mode><mode>): Likewise.
37655 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37657 (fpmask_comparison_operator): Update comment.
37658 (vecint_comparison_operator): New predicate.
37659 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37660 vector conditionals when the true and false values are constant
37661 vectors with all 0 bits or all 1 bits set.
37663 2017-02-06 Martin Sebor <msebor@redhat.com>
37665 PR tree-optimization/79376
37666 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37668 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37670 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37671 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37672 to simplify split condition.
37674 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37676 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37677 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37680 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37682 PR rtl-optimization/68664
37683 * target.def (can_speculate_insn): New hook.
37684 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37685 * doc/tm.texi: Regenerate.
37686 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37687 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37688 (rs6000_sched_can_speculate_insn): New function.
37690 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37692 PR tree-optimization/79284
37693 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37694 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37695 vectorizable_mask_load_store, vectorizable_operation,
37696 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37697 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37698 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37699 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37700 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37701 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37702 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37703 is_gimple_assign (stmt). Replace another such test with
37704 is_gimple_assign (stmt).
37706 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37709 * config/avr/avr.c (rtl-iter.h): Include it.
37710 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37711 (avr_legitimate_combined_insn): ...and implementation.
37713 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37715 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37716 * config/s390/s390.c (s390_const_operand_ok)
37717 (s390_canonicalize_comparison, s390_extract_part)
37718 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37719 (s390_contiguous_bitmask_p, s390_rtx_costs)
37720 (legitimize_pic_address): Likewise.
37721 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37722 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37723 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37724 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37725 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37727 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37729 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37730 REGNO($0) == REGNO($1).
37732 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37734 * config/s390/linux.h(SIZE_TYPE): Add comment.
37736 2017-02-06 Julian Brown <julian@codesourcery.com>
37737 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37738 Virendra Pathak <virendra.pathak@broadcom.com>
37740 * config/aarch64/aarch64-cores.def: Change the scheduler
37742 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37743 * config/aarch64/thunderx2t99.md: New file.
37745 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37747 * doc/standards.texi (Go Language): Update link to language
37750 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37752 * tree-eh.c (lower_resx): Sanitize profile.
37753 (cleanup_empty_eh_move_lp): Likewise.
37755 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37758 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37760 * cfgloopmanip.h (loop_version): Update prototype.
37761 * modulo-sched.c (sms_schedule): Update call of loop_version.
37762 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37763 * tree-parloops.c (gen_parallel_loop): Likewise.
37764 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37765 * tree-ssa-loop-split.c (split_loop): Likewise.
37766 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37767 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37769 2017-02-05 Martin Liska <mliska@suse.cz>
37772 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37774 (print_operand_address): Initialize a struct to zero.
37776 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37778 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37779 garbage collector only in textual form.
37781 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37783 * doc/extend.texi (x86 specific memory model extensions for
37784 transactional memory): Simplify a phrase.
37786 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37789 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37790 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37791 (atomic_storedi_1): Likewise.
37793 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37795 PR tree-optimization/79338
37796 * tree-parloops.c (gather_scalar_reductions): Don't call
37797 vect_analyze_loop_form for loop->inner before destroying loop's
37800 2017-02-03 Martin Sebor <msebor@redhat.com>
37802 PR tree-optimization/79327
37803 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37804 when precision has resulted in leading zeros.
37805 (format_integer): Adjust the likely counter to assume an unknown
37806 argument that may be zero is non-zero.
37808 2017-02-03 Jason Merrill <jason@redhat.com>
37811 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37812 avoid copying non-taken branch.
37814 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37816 PR tree-optimization/79340
37817 * tree-vect-loop.c (vectorizable_reduction): Release
37818 vec_defs elements after safe_splicing them into other vectors.
37821 PR tree-optimization/79327
37822 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37823 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37825 (format_integer): Use wide_int_to_tree instead of build_int_cst
37826 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37827 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37828 of shortest and longest sequence.
37830 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37832 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37833 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37835 2017-02-03 Walter Lee <walt@tilera.com>
37838 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37839 after initial stackframe link reg save.
37840 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37842 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37845 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37846 wu for stxssp alternative.
37848 2017-02-03 Martin Sebor <msebor@redhat.com>
37850 PR tree-optimization/79352
37851 * gimple-fold.c (get_range_strlen): Add argument.
37852 (get_range_strlen): Change return type to bool.
37853 (get_maxval_strlen): Pass in a dummy argument.
37854 * gimple-fold.h (get_range_strlen): Change return type to bool.
37855 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
37856 * tree.h (array_at_struct_end_p): Add argument.
37857 * tree.c (array_at_struct_end_p): Handle it.
37859 2017-02-03 Martin Liska <mliska@suse.cz>
37862 * multiple_target.c (create_dispatcher_calls): Redirect edge
37863 from a caller of a dispatcher.
37864 (expand_target_clones): Make the clones local.
37865 (ipa_target_clone): Do both target clones and resolvers.
37866 (ipa_dispatcher_calls): Remove the pass.
37867 (pass_dispatcher_calls::gate): Likewise.
37868 (make_pass_dispatcher_calls): Likewise.
37869 * passes.def (pass_target_clone): Put as very first IPA early
37872 2017-02-03 Martin Liska <mliska@suse.cz>
37874 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
37875 in case of a function with ifunc attribute.
37877 2017-02-03 Martin Liska <mliska@suse.cz>
37879 * cgraph.c (cgraph_node::dump): Dump function version info.
37880 * symtab.c (symtab_node::dump_base): Add missing new line.
37882 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37884 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
37885 (ifcombine_ifandif): Use it.
37887 2017-02-03 Martin Liska <mliska@suse.cz>
37889 * doc/invoke.texi: Document default value for
37890 use-after-scope-direct-emission-threshold.
37892 2017-02-03 Martin Liska <mliska@suse.cz>
37894 PR tree-optimization/79339
37895 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
37896 (format_floating): Likewise.
37898 2017-02-03 Martin Liska <mliska@suse.cz>
37901 * ipa-prop.c (ipa_node_params_t::insert): Remove current
37903 (ipa_node_params_t::remove): Likewise.
37904 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
37905 initialization from removed ipa_node_params_t::insert.
37906 (ipa_node_params::~ipa_node_params): Move from removed
37907 ipa_node_params_t::release.
37908 * symbol-summary.h (symbol_summary::m_released): New member.
37909 Do not release a summary twice. Do not allow to call finalizer
37910 for types of a summary that live in GGC memory.
37912 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37914 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
37917 2017-02-02 Martin Sebor <msebor@redhat.com>
37919 PR middle-end/79275
37920 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
37921 (format_string): Tighten up the range of output for non-constant
37922 strings and correct the expected range for wide non-constant strings.
37924 2017-02-02 Martin Sebor <msebor@redhat.com>
37926 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
37928 PR middle-end/32003
37929 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
37931 (-fdump-tree-@var): Add to index and document how to come up
37932 with pass-specific option and dump file names.
37933 (-fdump-passes): Clarify where to look for output.
37935 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37937 PR middle-end/77445
37938 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
37939 statistics of the analyzed path; allow threading for speed when
37940 any of BBs along the path are optimized for speed.
37942 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
37944 PR middle-end/78468
37945 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
37946 settings of the virtual registers.
37949 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
37951 * explow.c (get_dynamic_stack_size): Take known alignment of stack
37952 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
37955 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37957 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
37958 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
37960 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37962 * config/s390/s390.md: Add missing comments with the expanded
37964 * config/s390/vector.md: Likewise.
37965 * config/s390/vx-builtins.md: Likewise.
37967 2017-02-02 Jakub Jelinek <jakub@redhat.com>
37970 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
37971 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
37972 conditions on a single line.
37974 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37976 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37977 __S390_VX__ to __VX__.
37979 2017-02-01 Andrew Pinski <apinski@cavium.com>
37981 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
37982 stmt_info to record_stmt_cost.
37983 (vect_get_known_peeling_cost): Pass stmt_info if known to
37985 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
37986 cpu_vector_cost field into
37987 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
37988 field into vec_int_stmt_cost and vec_fp_stmt_cost.
37989 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
37990 splitting of scalar_stmt_cost and vec_stmt_cost.
37991 (thunderx_vector_cost): Likewise.
37992 (cortexa57_vector_cost): LIkewise.
37993 (exynosm1_vector_cost): Likewise.
37994 (xgene1_vector_cost): Likewise.
37995 (thunderx2t99_vector_cost): Improve after the splitting of the two
37997 (aarch64_builtin_vectorization_cost): Update for the splitting of
37998 scalar_stmt_cost and vec_stmt_cost.
38000 2017-02-01 Torvald Riegel <triegel@redhat.com>
38001 Richard Henderson <rth@redhat.com>
38003 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38004 conditional on existance of a fast atomic load.
38005 * optabs-query.c (can_atomic_load_p): New function.
38006 * optabs-query.h (can_atomic_load_p): Declare it.
38007 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38008 no fast atomic load is available for the particular size of access.
38009 (expand_atomic_compare_and_swap): Likewise.
38010 (expand_atomic_load): Likewise.
38011 (expand_atomic_store): Likewise.
38012 (expand_atomic_fetch_op): Likewise.
38013 * testsuite/lib/target-supports.exp
38014 (check_effective_target_sync_int_128): Remove x86 because it provides
38015 no fast atomic load.
38016 (check_effective_target_sync_int_128_runtime): Likewise.
38018 2017-02-01 Richard Biener <rguenther@suse.de>
38020 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38021 (graphite_transform_loops): Provide opt-info for optimized nests.
38022 * tree-parloop.c (parallelize_loops): Provide opt-info for
38023 parallelized loops.
38025 2017-02-01 Richard Biener <rguenther@suse.de>
38027 PR middle-end/79315
38028 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38029 was not set before.
38031 2017-02-01 Richard Biener <rguenther@suse.de>
38033 PR tree-optimization/71824
38034 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38035 Verify the loops are valid in the merged SESE region.
38036 (scop_detection::can_represent_loop_1): Check analyzing the
38037 evolution of the number of iterations in the region succeeds.
38039 2017-01-31 Ian Lance Taylor <iant@golang.org>
38041 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38042 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38044 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38046 PR preprocessor/79210
38047 * input.c (get_substring_ranges_for_loc): Replace line_width
38048 assertion with error-handling.
38050 2017-01-31 Richard Biener <rguenther@suse.de>
38052 PR tree-optimization/77318
38053 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38054 (create_pw_aff_from_tree): Take loop parameter.
38055 (add_condition_to_pbb): Pass loop of the condition to
38056 create_pw_aff_from_tree.
38058 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38060 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38061 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38063 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38067 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38069 (convert_int_to_float128): Likewise.
38070 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38071 (convert_int_to_float128): Likewise.
38072 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38073 (UNSPEC_IEEE128_CONVERT): Likewise.
38074 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38075 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38076 Use local variables for IBM extended format.
38077 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38078 (fix_trunc<mode>si2_fprs): Likewise.
38079 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38080 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38081 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38082 to know that we can now have integers of all sizes in vector
38084 (fix<uns>_<mode>di2_hw): Likewise.
38085 (float<uns>_<mode>si2_hw): Likewise.
38086 (fix_<mode>si2_hw): Likewise.
38087 (fixuns_<mode>si2_hw): Likewise.
38088 (float<uns>_<mode>di2_hw): Likewise.
38089 (float_<mode>di2_hw): Likewise.
38090 (float_<mode>si2_hw): Likewise.
38091 (floatuns_<mode>di2_hw): Likewise.
38092 (floatuns_<mode>si2_hw): Likewise.
38093 (xscvqp<su>wz_<mode>): Delete, no longer used.
38094 (xscvqp<su>dz_<mode>): Likewise.
38095 (xscv<su>dqp_<mode>): Likewise.
38096 (ieee128_mfvsrd_64bit): Likewise.
38097 (ieee128_mfvsrd_32bit): Likewise.
38098 (ieee128_mfvsrwz): Likewise.
38099 (ieee128_mtvsrw): Likewise.
38100 (ieee128_mtvsrd_64bit): Likewise.
38101 (ieee128_mtvsrd_32bit): Likewise.
38103 2017-01-31 Martin Liska <mliska@suse.cz>
38106 * ipa-prop.c (ipa_free_all_node_params): Call release method
38107 instead of ~sumbol_summary to not to trigger double times
38110 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38112 PR tree-optimization/71691
38113 * bitmap.h (class auto_bitmap): New.
38114 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38115 is_maybe_undefined instead of ssa_undefined_value_p.
38117 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38119 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38120 __S390_ARCH_LEVEL__ to __ARCH__.
38122 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38124 PR tree-optimization/79267
38125 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38126 if should_remove_lhs_p is true.
38128 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38131 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38132 (add_alignment_attribute): New.
38133 (base_type_die): Add alignment attribute.
38134 (subrange_type_die): Likewise.
38135 (modified_type_die): Likewise.
38136 (gen_array_type_die): Likewise.
38137 (gen_descr_array_type_die: Likewise.
38138 (gen_enumeration_type_die): Likewise.
38139 (gen_subprogram_die): Likewise.
38140 (gen_variable_die): Likewise.
38141 (gen_field_die): Likewise.
38142 (gen_ptr_to_mbr_type_die): Likewise.
38143 (gen_struct_or_union_type_die): Likewise.
38144 (gen_subroutine_type_die): Likewise.
38145 (gen_typedef_die): Likewise.
38146 (base_type_cmp): Compare alignment attribute.
38148 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38151 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38152 (setb_unsigned) New pattern for setb with CCUNS.
38153 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38154 subfc./subfe sequence to avoid overflow problems. Generate a
38155 shorter sequence with cmpld/setb for power9.
38156 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38157 for generating subfc. instruction.
38158 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38159 now uses this instruction.
38161 2017-01-30 Ian Lance Taylor <iant@google.com>
38164 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38165 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38167 2017-01-30 Martin Sebor <msebor@redhat.com>
38169 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38170 Move constant to the right of a relational operator.
38171 (get_mpfr_format_length, format_character, format_string): Ditto.
38172 (should_warn_p, maybe_warn): Same.
38174 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38176 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38179 * asan.c (get_translation_unit_decl): Remove function.
38180 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38182 2017-01-30 Martin Liska <mliska@suse.cz>
38184 PR gcov-profile/79259
38185 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38186 -fprofile-generate.
38188 2017-01-30 Martin Liska <mliska@suse.cz>
38191 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38192 Initialize variables with NULL value.
38194 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38197 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38199 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38201 2017-01-30 Richard Biener <rguenther@suse.de>
38203 PR tree-optimization/79276
38204 * tree-vrp.c (process_assert_insertions): Properly adjust common
38205 when removing a duplicate.
38207 2017-01-30 Richard Biener <rguenther@suse.de>
38209 PR tree-optimization/79256
38210 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38211 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38213 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38215 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38218 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38219 ("*r<noxa>sbg_<mode>_sll_bitmask")
38220 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38221 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38222 Use contiguous_bitmask_nowrap_operand.
38224 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38227 * config/rs6000/altivec.h (vec_xl): Revise #define.
38228 (vec_xst): Likewise.
38230 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38232 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38234 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38236 PR rtl-optimization/79194
38237 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38238 traps before call to bypass_conditional_jumps.
38240 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38242 PR tree-optimization/71374
38243 * lra-constraints.c (check_conflict_input_operands): New.
38244 (match_reload): Use it.
38246 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38249 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38250 account to calculate conflict_set.
38252 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38254 PR rtl-optimization/78559
38255 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38256 other_insn in combine.
38258 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38260 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38261 uint16_type_node for BT_UINT16.
38263 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38265 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38266 "RTL Tests" to menu.
38267 (GIMPLE Tests): New node.
38268 (RTL Tests): New node.
38270 2017-01-27 Richard Biener <rguenther@suse.de>
38272 PR tree-optimization/79245
38273 * tree-loop-distribution.c (distribute_loop): Apply cost
38274 modeling also to detected patterns.
38276 2017-01-27 Richard Biener <rguenther@suse.de>
38278 PR tree-optimization/71433
38279 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38280 (compare_assert_loc): New function.
38281 (process_assert_insertions): Sort and optimize assert locations
38282 to remove duplicates and push down identical assertions on
38283 edges to their destination block.
38285 2017-01-27 Richard Biener <rguenther@suse.de>
38287 PR tree-optimization/79244
38288 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38289 out SSA names even if abnormal.
38291 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38293 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38294 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38295 instead of MPFR_RNDN.
38297 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38300 * arm.c (arm_option_override): Don't call build_target_option_node
38301 until after doing all option overrides.
38302 (arm_valid_target_attribute_tree): Likewise.
38304 2017-01-27 Martin Liska <mliska@suse.cz>
38306 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38307 for {cd}tors and C++ {cd}tors.
38309 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38311 * config/s390/s390.md ("*setmem_long_and")
38312 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38314 2017-01-26 Martin Sebor <msebor@redhat.com>
38316 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38319 2017-01-26 Martin Sebor <msebor@redhat.com>
38321 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38322 HAVE_DFmode before using XFmode or DFmode.
38323 (parse_directive): Avoid using the z length modifier to avoid
38324 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38326 PR middle-end/78703
38327 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38328 to accept adjustment as an array.
38329 (get_int_range): New function.
38330 (struct directive): Make width and prec arrays.
38331 (directive::set_width, directive::set_precision): Call get_int_range.
38332 (format_integer, format_floating): Handle width and precision ranges.
38333 (format_string, parse_directive): Same.
38335 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38338 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38339 comdat_type_p set, just clone them, but keep the children in the
38343 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38344 which have direct callers with -fvar-tracking-assignments enabled
38346 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38347 inside of type units.
38349 2017-01-26 Martin Sebor <msebor@redhat.com>
38351 PR middle-end/78703
38352 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38354 (struct format_result): Replace number_chars, number_chars_min,
38355 and number_chars_max with a single member of struct result_range.
38357 (format_result::operator+=): Adjust.
38358 (struct fmtresult): Remove bounded. Handle likely and unlikely
38360 (fmtresult::adjust_for_width_or_precision): New function.
38361 (fmtresult:type_max_digits): New function.
38362 (bytes_remaining): Handle likely and unlikely counters.
38363 (min_bytes_remaining): Remove.
38364 (format_percent): Simplify.
38365 (format_integer, format_floating): Set likely and unlikely counters.
38366 (get_string_length, format_character, format_string): Same.
38367 (format_plain, should_warn_p): New function.
38368 (maybe_warn): Call should_warn_p. Update diagnostic messages
38369 and handle those for all directives, including plain strings.
38370 (format_directive): Handle likely and unlikely counters.
38371 Remove unnecessary quoting from diagnostics. Add an informational
38373 (add_bytes): Remove.
38374 (pass_sprintf_length::compute_format_length): Simplify.
38375 (try_substitute_return_value): Handle likely and unlikely counters.
38377 2017-01-26 Carl Love <cel@us.ibm.com>
38379 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38380 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38382 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38385 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38386 endianess for subregs into account.
38387 * lra-constraints.c (lra_constraints): Do risky transformations
38388 always on the first iteration.
38389 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38390 last_call_used_reg_set.
38391 (process_bb_lives): Define and use last_call_used_reg_set.
38392 * lra.c (lra): Always continue after lra_constraints on the first
38395 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38397 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38399 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38401 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38403 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38404 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38405 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38406 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38407 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38408 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38409 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38410 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38411 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38413 2017-01-26 Marek Polacek <polacek@redhat.com>
38416 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38417 for the third operand.
38419 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38421 PR middle-end/79236
38422 * omp-low.c (struct omp_context): Add simt_stmt field.
38423 (scan_omp_for): Return omp_context *.
38424 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38425 context to the _simt_ SIMD stmt.
38426 (lower_omp_for): For combined SIMD with sibling _simt_
38427 SIMD, make sure to use the same decls in _looptemp_
38428 clauses as in the sibling.
38430 2017-01-26 David Sherwood <david.sherwood@arm.com>
38432 PR middle-end/79212
38433 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38436 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38439 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38440 emit fld b; fld a; if possible.
38442 * brig-builtins.def: Update copyright years.
38443 * config/arm/arm_acle_builtins.def: Update copyright years.
38445 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38448 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38449 constraint instead of o for the stxsd instruction.
38451 2017-01-25 Carl Love <cel@us.ibm.com>
38453 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38454 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38456 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38458 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38460 2017-01-25 Richard Biener <rguenther@suse.de>
38462 PR tree-optimization/69264
38463 * target.def (vector_alignment_reachable): Improve documentation.
38464 * doc/tm.texi: Regenerate.
38465 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38467 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38468 earlier changes with respect to TYPE_USER_ALIGN.
38469 (vector_alignment_reachable_p): Likewise. Improve dumping.
38471 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38474 * config/arm/arm.md (xordi3): Force constant operand into a register
38477 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38479 * doc/invoke.texi (-fstore-merging): Correct default optimization
38480 levels at which it is enabled.
38481 (-O): Move -fstore-merging from list to...
38484 2017-01-25 Richard Biener <rguenther@suse.de>
38487 * omp-expand.c: Include debug.h.
38488 (expand_omp_taskreg): Make sure to generate early debug before
38489 outlining anything from a function.
38490 (expand_omp_target): Likewise.
38491 (grid_expand_target_grid_body): Likewise.
38493 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38496 * asan.c (get_translation_unit_decl): New function.
38497 (asan_add_global): Extract modules file name from globals
38498 TRANSLATION_UNIT_DECL name.
38500 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38503 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38504 for long calls with APCS frame and VFP.
38506 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38508 * cfg.c (original_copy_tables_initialized_p): New function.
38509 * cfg.h (original_copy_tables_initialized_p): New decl.
38510 * cfgrtl.c (relink_block_chain): Guard the call to
38511 free_original_copy_tables with a call to
38512 original_copy_tables_initialized_p.
38513 * cgraph.h (symtab_node::native_rtl_p): New decl.
38514 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38515 (symtab_node::needed_p): Don't assert for early assembly output
38516 for __RTL functions.
38517 (cgraph_node::finalize_function): Set "force_output" for __RTL
38519 (cgraph_node::analyze): Bail out early for __RTL functions.
38520 (analyze_functions): Update assertion to support __RTL functions.
38521 (cgraph_node::expand): Bail out early for __RTL functions.
38522 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38524 * function.h (struct function): Update comment for field
38526 * gimple-expr.c: Include "tree-pass.h".
38527 (gimple_has_body_p): Return false for __RTL functions.
38528 * Makefile.in (OBJS): Add run-rtl-passes.o.
38529 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38531 (gcc::pass_manager::get_clean_slate): New accessor.
38532 * passes.c: Include "insn-addr.h".
38533 (should_skip_pass_p): Add logging. Update logic for running
38534 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38535 property-provider override so it is only done for gimple passes.
38537 (skip_pass): New function.
38538 (execute_one_pass): Call skip_pass when skipping passes.
38539 * read-md.c (md_reader::read_char): Support filtering
38540 the input to a subset of line numbers.
38541 (md_reader::md_reader): Initialize fields
38542 m_first_line and m_last_line.
38543 (md_reader::read_file_fragment): New function.
38544 * read-md.h (md_reader::read_file_fragment): New decl.
38545 (md_reader::m_first_line): New field.
38546 (md_reader::m_last_line): New field.
38547 * read-rtl-function.c (function_reader::create_function): Only
38548 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38549 curr_properties. Set DECL_INITIAL to a dummy block.
38550 (read_rtl_function_body_from_file_range): New function.
38551 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38553 * run-rtl-passes.c: New file.
38554 * run-rtl-passes.h: New file.
38556 2017-01-24 Jeff Law <law@redhat.com>
38558 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38561 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38563 PR tree-optimization/79159
38564 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38565 (record_nonwrapping_iv): Improve boundary using above function if no
38566 value range information.
38568 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38569 Martin Jambor <mjambor@suse.cz>
38571 * brig-builtins.def: New file.
38572 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38573 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38574 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38575 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38576 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38577 * builtin-types.def (BT_INT8): New.
38578 (BT_INT16): Likewise.
38579 (BT_UINT8): Likewise.
38580 (BT_UINT16): Likewise.
38581 (BT_FN_ULONG): Likewise.
38582 (BT_FN_UINT_INT): Likewise.
38583 (BT_FN_UINT_ULONG): Likewise.
38584 (BT_FN_UINT_LONG): Likewise.
38585 (BT_FN_UINT_PTR): Likewise.
38586 (BT_FN_ULONG_PTR): Likewise.
38587 (BT_FN_INT8_FLOAT): Likewise.
38588 (BT_FN_INT16_FLOAT): Likewise.
38589 (BT_FN_UINT32_FLOAT): Likewise.
38590 (BT_FN_UINT16_FLOAT): Likewise.
38591 (BT_FN_UINT8_FLOAT): Likewise.
38592 (BT_FN_UINT64_FLOAT): Likewise.
38593 (BT_FN_UINT16_UINT32): Likewise.
38594 (BT_FN_UINT32_UINT16): Likewise.
38595 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38596 (BT_FN_INT_PTR_INT): Likewise.
38597 (BT_FN_UINT_PTR_UINT): Likewise.
38598 (BT_FN_LONG_PTR_LONG): Likewise.
38599 (BT_FN_ULONG_PTR_ULONG): Likewise.
38600 (BT_FN_VOID_UINT64_UINT64): Likewise.
38601 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38602 (BT_FN_INT8_INT8_INT8): Likewise.
38603 (BT_FN_INT16_INT16_INT16): Likewise.
38604 (BT_FN_INT_INT_INT): Likewise.
38605 (BT_FN_UINT_FLOAT_UINT): Likewise.
38606 (BT_FN_FLOAT_UINT_UINT): Likewise.
38607 (BT_FN_ULONG_UINT_UINT): Likewise.
38608 (BT_FN_ULONG_UINT_PTR): Likewise.
38609 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38610 (BT_FN_UINT_UINT_UINT): Likewise.
38611 (BT_FN_VOID_UINT_PTR): Likewise.
38612 (BT_FN_UINT_UINT_PTR: Likewise.
38613 (BT_FN_UINT32_UINT64_PTR): Likewise.
38614 (BT_FN_INT_INT_UINT_UINT): Likewise.
38615 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38616 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38617 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38618 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38619 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38620 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38621 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38622 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38623 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38624 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38625 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38626 * doc/frontends.texi: List BRIG FE.
38627 * doc/install.texi (Testing): Add BRIG tesring requirements.
38628 * doc/invoke.texi (Overall Options): Mention BRIG.
38629 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38631 2017-01-24 Richard Biener <rguenther@suse.de>
38633 PR translation/79208
38634 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38636 2017-01-24 Martin Jambor <mjambor@suse.cz>
38639 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38640 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38641 and known_contexts.
38643 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38645 PR middle-end/79123
38646 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38647 casts from signed to unsigned really don't have a range.
38649 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38651 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38652 GMP_RNDx for compatiblity.
38654 2017-01-24 Martin Liska <mliska@suse.cz>
38657 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38658 that would prevent us to call alloca with -1 as argument.
38660 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38662 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38663 Avoid -Wformat-security warning.
38665 2017-01-23 Andrew Pinski <apinski@cavium.com>
38667 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38670 2017-01-23 Martin Sebor <msebor@redhat.com>
38672 PR middle-end/78703
38673 * gimple-ssa-sprintf.c (warn_level): New global.
38674 (format_integer): Use it here and throughout the rest of the file.
38675 Use the same switch to compute sign as base.
38676 (maybe_warn): New function.
38677 (format_directive): Factor out warnings into maybe_warn.
38678 Add debugging output. Use warn_level.
38679 (add_bytes): Use warn_level.
38680 (pass_sprintf_length::compute_format_length): Add debugging output.
38681 (try_substitute_return_value): Same.
38682 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38684 PR middle-end/78703
38685 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38686 (struct fmtresult, format_integer, format_floating): Adjust.
38687 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38688 (get_string_length, format_string,format_directive): Same.
38689 (pass_sprintf_length::compute_format_length): Same.
38690 (try_substitute_return_value): Simplify slightly.
38692 PR middle-end/78703
38693 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38694 (fmtresult::operator+=): Outlined.
38695 (struct fmtresult): Add ctors.
38696 (struct conversion_spec): Rename...
38697 (struct directive): ...to this. Add and remove data members.
38698 (directive::set_width, directive::set_precision): New functions.
38699 (format_percent): Use fmtresult ctor.
38700 (get_width_and_precision): Remove.
38701 (format_integer): Make naming changes. Avoid computing width and
38703 (format_floating): Same. Adjust indentation.
38704 (format_character, format_none): New functions.
38705 (format_string): Moved character handling to format_character.
38706 (format_directive): Remove arguments, change return type.
38707 (parse_directive): New function.
38708 (pass_sprintf_length::compute_format_length): Move directive
38709 parsing to parse_directive.
38711 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38713 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38714 (assign_assembler_name_if_needed): ... this.
38715 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38716 (assign_assembler_name_if_needed): ... this.
38717 (free_lang_data_in_cgraph): Adjust callers.
38718 * cgraphunit.c (cgraph_node::analyze): Likewise.
38719 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38722 2017-01-23 Richard Biener <rguenther@suse.de>
38724 PR tree-optimization/79088
38725 PR tree-optimization/79188
38726 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38727 resetting loop bounds after last path deletion. Reset loop
38728 bounds of the target loop, make code match the comments.
38729 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38730 Make sure loops need no fixups.
38732 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38734 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38735 exponent support with double type for first argument.
38736 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38737 type returned by __builtin_vec_extract_sig,
38738 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38739 functions from "vector int" to "vector unsigned int" or from
38740 "vector long long int" to "vector unsigned long long int".
38741 Changed type returned by __builtin_vec_extract_exp,
38742 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38743 functions from "vector int" to "vector unsigned int" or from
38744 "vector long long int" to "vector unsigned long long int".
38745 Changed return type of __builtin_vec_test_data_class,
38746 __builtin_vec_test_data_class_sp, and
38747 __builtin_vec_test_data_class_dp from "vector int" to
38748 "vector bool int" or from "vector long long int" to "vector bool
38749 long long int" and changed second argument type from "unsigned
38750 int" to "int". Added new overloaded function forms "vector float
38751 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38752 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38753 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38754 double, vector unsigned long long int)" and "vector double
38755 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38756 long int)". Changed return type of
38757 __builtin_scalar_test_data_class and
38758 __builtin_scalar_test_data_class_sp and
38759 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38760 int" and changed second argument from "unsigned int" to "int".
38761 Changed type returned by __builtin_scalar_test_neg,
38762 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38763 from "int" to "bool int". Added new overloaded function form
38764 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38765 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38766 exponent double-precision with floating point first argument.
38767 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38768 documentation of scalar_test_data_class, scalar_test_neg,
38769 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38770 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38771 vec_test_data_class built-in functions to reflect refinements in
38772 their type signatures.
38774 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38776 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38778 (aarch64_elf_asm_destructor): Likewise.
38780 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38782 PR rtl-optimization/78634
38783 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38784 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38785 * ifcvt.c (noce_try_cmove): Add missing cost check.
38787 PR rtl-optimization/71724
38788 * combine.c (if_then_else_cond): Look for situations where it is
38789 beneficial to undo the work of one of the recursive calls.
38791 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38793 PR tree-optimization/70754
38794 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38795 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38796 combined stmt before it if not NULL.
38797 (combine_chains): Process refs reversely and compute dominance point
38800 2017-01-23 Martin Liska <mliska@suse.cz>
38802 PR tree-optimization/79196
38803 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38804 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38806 (strlen_optimize_stmt): Call the renamed function.
38808 2017-01-23 Michael Matz <matz@suse.de>
38810 PR tree-optimization/78384
38811 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38813 2017-01-23 Richard Biener <rguenther@suse.de>
38815 PR tree-optimization/79186
38816 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38817 both incoming edges before moving an assert.
38819 2017-01-23 Martin Jambor <mjambor@suse.cz>
38821 * ipa-prop.c (load_from_param_1): Removed.
38822 (load_from_unmodified_param): Bits from load_from_param_1 put back
38824 (load_from_param): Removed.
38825 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38826 with stmt. Reverted back to use of load_from_unmodified_param.
38828 2017-01-23 Martin Jambor <mjambor@suse.cz>
38831 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38832 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38833 field a pointer to garbage collected vector, mark lattices and
38834 ipcp_orig_node with GTY((skip)).
38835 (ipa_get_param_count): Adjust to descriptors being a pointer.
38836 (ipa_get_param): Likewise.
38837 (ipa_get_type): Likewise.
38838 (ipa_get_param_move_cost): Likewise.
38839 (ipa_set_param_used): Likewise.
38840 (ipa_get_controlled_uses): Likewise.
38841 (ipa_set_controlled_uses): Likewise.
38842 (ipa_is_param_used): Likewise.
38843 (ipa_node_params_t): Move into garbage collector. New methods insert
38845 (ipa_node_params_sum): Annotate wth GTY(()).
38846 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38848 (ipa_load_from_parm_agg): Adjust declaration.
38849 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38850 * ipa-profile.c (ipa_profile): Likewise.
38851 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
38852 (ipa_populate_param_decls): Make descriptors parameter garbage
38854 (ipa_dump_param): Adjust to descriptors being a pointer.
38855 (ipa_alloc_node_params): Likewise.
38856 (ipa_initialize_node_params): Likewise.
38857 (load_from_param_1): Make descriptors parameter garbage collected.
38858 (load_from_unmodified_param): Likewise.
38859 (load_from_param): Likewise.
38860 (ipa_load_from_parm_agg): Likewise.
38861 (ipa_node_params::~ipa_node_params): Removed.
38862 (ipa_free_all_node_params): Remove call to delete operator.
38863 (ipa_node_params_t::insert): New.
38864 (ipa_node_params_t::remove): Likewise.
38865 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
38866 copy known_csts and known_contexts vectors.
38867 (ipa_read_node_info): Adjust to descriptors being a pointer.
38868 (ipcp_modif_dom_walker): Make m_descriptors field garbage
38870 (ipcp_transform_function): Make descriptors variable garbage
38873 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
38875 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
38876 * config/i386/avx512dqintrin.h: Ditto.
38877 * config/i386/avx512fintrin.h: Ditto.
38878 * config/i386/i386.c: Handle new builtins.
38879 * config/i386/i386-builtin.def: Add new builtins.
38880 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
38881 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
38883 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38884 Martin Liska <mliska@suse.cz>
38886 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
38887 * asan.c (asan_expand_poison_ifn): Support stores and use
38888 appropriate ASAN report function.
38889 * internal-fn.c (expand_ASAN_POISON_USE): New function.
38890 * internal-fn.def (ASAN_POISON_USE): Declare.
38891 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
38892 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
38893 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
38894 ASAN_POISON calls w/o LHS.
38895 * tree-ssa.c (execute_update_addresses_taken): Create clobber
38896 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
38897 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
38898 * gimplify.c (asan_poison_variables): Add attribute
38899 use_after_scope_memory to variables that really needs to live
38901 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
38902 having the attribute.
38904 2017-01-23 Martin Liska <mliska@suse.cz>
38906 * asan.c (create_asan_shadow_var): New function.
38907 (asan_expand_poison_ifn): Likewise.
38908 * asan.h (asan_expand_poison_ifn): New declaration.
38909 * internal-fn.c (expand_ASAN_POISON): Likewise.
38910 * internal-fn.def (ASAN_POISON): New builtin.
38911 * sanopt.c (pass_sanopt::execute): Expand
38912 asan_expand_poison_ifn.
38913 * tree-inline.c (copy_decl_for_dup_finish): Make function
38915 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
38916 * tree-ssa.c (is_asan_mark_p): New function.
38917 (execute_update_addresses_taken): Rewrite local variables
38918 (identified just by use-after-scope as addressable) into SSA.
38920 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38922 * doc/install.texi (Specific): opensource.apple.com uses https
38923 now. Remove trailing slash.
38925 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38927 * README.Portability: Remove note on an Irix compatibility issue.
38929 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
38931 * gcov.c (INCLUDE_ALGORITHM): Define.
38932 (INCLUDE_VECTOR): Define.
38933 No longer include <vector> and <algorithm> directly.
38935 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
38937 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
38939 * doc/invoke.texi (Code Gen Options): Ditto.
38941 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
38944 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
38946 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
38948 rtl-optimization/79125
38949 * cprop.c (local_cprop_pass): Handle cases where we make an
38950 unconditional trap.
38952 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
38956 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
38957 read from, for big endian.
38959 2017-01-20 Jiong Wang <jiong.wang@arm.com>
38961 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
38962 register pauth builtins for LP64 only.
38964 2017-01-20 Marek Polacek <polacek@redhat.com>
38967 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
38970 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38972 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
38974 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
38975 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
38976 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
38978 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38981 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
38982 in big-endian mode when they are not a single duplicated value.
38984 2017-01-20 Richard Biener <rguenther@suse.de>
38986 * BASE-VER: Bump to 7.0.1.
38988 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38990 * omp-low.c (omplow_simd_context): New struct. Use it...
38991 (lower_rec_simd_input_clauses): ...here and...
38992 (lower_rec_input_clauses): ...here to hold common data. Adjust all
38993 references to idx, lane, max_vf, is_simt.
38995 2017-01-20 Graham Markall <graham.markall@embecosm.com>
38997 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39000 2017-01-20 Martin Jambor <mjambor@suse.cz>
39002 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39003 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39005 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39006 (GTFILES): Rename hsa.c to hsa-common.c.
39007 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39008 * hsa-dump.c: Likewise.
39009 * hsa-gen.c: Likewise.
39010 * hsa-regalloc.c: Likewise.
39011 * ipa-hsa.c: Likewise.
39012 * omp-expand.c: Likewise.
39013 * omp-low.c: Likewise.
39014 * toplev.c: Likewise.
39016 2017-01-20 Marek Polacek <polacek@redhat.com>
39019 * doc/invoke.texi: Document -Wduplicated-branches.
39020 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39021 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39022 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39023 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39024 return 0 only when not OEP_LEXICOGRAPHIC.
39025 (fold_build_cleanup_point_expr): Use the expression
39026 location when building CLEANUP_POINT_EXPR.
39027 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39028 * tree.c (add_expr): Handle error_mark_node.
39030 2017-01-20 Martin Liska <mliska@suse.cz>
39033 * tree-profile.c (init_ic_make_global_vars): Do not call
39035 (gimple_init_gcov_profiler): Likewise.
39037 2017-01-20 Martin Liska <mliska@suse.cz>
39040 * cgraph.h (maybe_create_reference): Remove argument and
39042 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39044 * ipa-cp.c (create_specialized_node): Likewise.
39045 * symtab.c (symtab_node::maybe_create_reference): Handle
39046 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39048 2017-01-20 Martin Liska <mliska@suse.cz>
39050 * read-rtl-function.c (function_reader::create_function): Use
39051 build_decl instread of build_decl_stat.
39053 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39055 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39056 * config/i386/avx512dqintrin.h: Ditto.
39057 * config/i386/avx512fintrin.h: Ditto.
39058 * config/i386/i386-builtin-types.def: Add new types.
39059 * config/i386/i386.c: Handle new types.
39060 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39061 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39062 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39063 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39064 (__builtin_ia32_kshiftridi): New.
39065 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39067 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39071 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39072 define to rs6000_init_stack_protect_guard.
39073 (rs6000_init_stack_protect_guard): New function.
39075 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39076 Yunqiang Su <yunqiang.su@imgtec.com>
39078 * config.gcc (supported_defaults): Add madd4.
39079 (with_madd4): Add validation.
39080 (all_defaults): Add madd4.
39081 * config/mips/mips.opt (mmadd4): New option.
39082 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39084 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39086 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39087 (ISA_HAS_FUSED_MADD4): Likewise.
39088 * doc/invoke.texi (-mmadd4): Document the new option.
39089 * doc/install.texi (--with-madd4): Document the new option.
39091 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39093 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39094 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39095 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39096 (aarch64_init_pauth_hint_builtins): New.
39097 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39098 (aarch64_expand_builtin): Expand new builtins.
39100 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39102 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39103 * combine-stack-adj.c (no_unhandled_cfa): Handle
39104 REG_CFA_TOGGLE_RA_MANGLE.
39105 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39106 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39107 info for return address signing.
39108 (aarch64_expand_epilogue): Likewise.
39110 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39112 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39113 * config/aarch64/aarch64-protos.h
39114 (aarch64_return_address_signing_enabled): New declaration.
39115 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39117 (aarch64_expand_prologue): Sign return address before it's pushed onto
39119 (aarch64_expand_epilogue): Authenticate return address fetched from
39121 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39122 (aarch64_attributes): New function attributes for "sign-return-address".
39123 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39124 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39125 ("*do_return"): Generate combined instructions according to key index.
39126 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39127 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39129 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39130 * config/aarch64/aarch64.opt (msign-return-address=): New.
39131 * doc/extend.texi (AArch64 Function Attributes): Documents
39132 "sign-return-address=".
39133 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39135 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39137 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39138 overall option summary.
39140 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39142 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39143 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39144 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39145 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39147 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39149 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39150 -mpower9-minmax by default for -mcpu=power9.
39151 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39152 128-bit floating point.
39154 2017-01-20 Alan Modra <amodra@gmail.com>
39156 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39157 optimizing for size.
39159 2017-01-20 Alan Modra <amodra@gmail.com>
39162 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39163 for strcmp and strncmp from corresponding builtin decl.
39165 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39167 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39168 instead of i386/rtems-64.h.
39169 * config/i386/rtems-64.h: Remove.
39171 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39175 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39177 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39179 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39181 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39182 Change int to HOST_WIDE_INT.
39183 * config/aarch64/aarch64-protos.h
39184 (aarch64_simd_gen_const_vector_dup): Likewise.
39185 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39187 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39189 * langhooks-def.h (lhd_type_for_size): New decl.
39190 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39191 * langhooks.c (lhd_type_for_size): New function, taken from
39194 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39196 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39197 define_bypass for CR latency.
39198 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39199 (power9-alu2): Add define_bypass for CR latency.
39201 (power9-mul): Update insn latency.
39202 (power9-mul-compare): Update insn latency, bypass latency and remove
39205 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39207 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39209 * config/aarch64/aarch64.md
39210 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39211 aarch64_nopcrelative_literal_loads.
39212 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39214 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39216 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39217 TARGET_LOONGSON_3A.
39218 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39220 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39223 * config.gcc (supported_defaults): Add lxc1-sxc1.
39224 (with_lxc1_sxc1): Add validation.
39225 (all_defaults): Add lxc1-sxc1.
39226 * config/mips/mips.opt (mlxc1-sxc1): New option.
39227 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39229 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39230 __mips_no_lxc1_sxc1.
39231 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39232 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39233 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39235 2017-01-19 Richard Biener <rguenther@suse.de>
39237 PR tree-optimization/72488
39238 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39239 sure to restore SSA info.
39240 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39242 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39244 PR rtl-optimization/79121
39245 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39246 of the inner type when shifting an extended value.
39248 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39251 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39252 interposable aliases.
39254 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39257 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39258 Use the evmergelohi instruction.
39259 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39260 (mov_sitf_e500_subreg8_2_be): Likewise.
39261 (mov_sitf_e500_subreg12_2_le): Likewise.
39262 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39263 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39264 (mov_sitf_e500_subreg8_2_le): Likewise.
39265 (mov_sitf_e500_subreg12_2_be): Likewise.
39267 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39269 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39270 attribute from vecsimple to vecperm.
39271 (altivec_vbpermq2): Likewise.
39273 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39276 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39278 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39279 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39280 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39281 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39282 case where N arg is SIZE_MAX.
39283 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39284 (cmpstrsi): Add pattern.
39286 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39288 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39289 __builtin_vec_revb builtins.
39290 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39291 built-in functions to support generation of the ISA 3.0 XXBR<x>
39292 vector byte reverse instructions.
39293 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39294 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39295 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39296 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39297 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39298 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39299 (P9V_BUILTIN_VEC_REVB): Likewise.
39300 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39301 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39302 (p9_xxbrq_v16qi): Likewise.
39303 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39304 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39305 (p9_xxbrh_v8hi): Likewise.
39306 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39307 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39308 vec_revb built-in functions.
39310 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39312 PR rtl-optimization/78952
39313 * config/i386/i386.md (any_extract): New code iterator.
39314 (*insvqi_2): Use any_extract for source operand.
39315 (*insvqi_3): Use any_shiftrt for source operand.
39317 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39319 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39321 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39323 2017-01-18 Matthias Klose <doko@ubuntu.com>
39325 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39327 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39329 * config/rs6000/altivec.h (vec_bperm): Change #define.
39330 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39331 (altivec_vbpermq2): New define_insn.
39332 (altivec_vbpermd): Likewise.
39333 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39334 function interface.
39335 (VBPERMD): Likewise.
39336 (VBPERM): New polymorphic function interface.
39337 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39338 Add entries for P9V_BUILTIN_VEC_VBPERM.
39339 * doc/extend.texi: Add interfaces for vec_bperm.
39341 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39343 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39344 first letter of error messages.
39345 (s390_resolve_overloaded_builtin): Likewise.
39346 * config/s390/s390.c (s390_expand_builtin): Likewise.
39347 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39348 (s390_valid_target_attribute_inner_p): Likewise.
39349 * config/s390/s390.md ("tabort"): Likewise.
39351 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39353 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39354 (ISA_AVOID_DIV_HILO): New macro.
39355 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39356 (ISA_HAS_DDIV): Likewise.
39358 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39360 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39362 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39364 * doc/invoke.texi (fabi-version): Spelling fix.
39366 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39369 * doc/invoke.texi (fabi-version): Mention mangling fix for
39372 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39375 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39377 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39380 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39381 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39383 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39385 (stack_protect_test): Ditto.
39386 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39387 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39389 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39390 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39391 -mstack-protector-guard-offset=.
39392 (RS/6000 and PowerPC Options): Ditto.
39394 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39396 * config/i386/i386.h (MASK_CLASS_P): New define.
39397 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39398 there are no registers from different register sets also when
39399 mask registers are used. Update function comment.
39400 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39401 to (*k/*r) and (*k/*km) alternatives.
39403 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39405 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39406 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39407 (EH_RETURN_HANDLER_RTX): New define.
39408 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39409 Force frame pointer in EH return functions.
39410 (aarch64_expand_epilogue): Add barrier for eh_return.
39411 (aarch64_final_eh_return_addr): Remove.
39412 (aarch64_eh_return_handler_rtx): New function.
39413 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39415 (aarch64_eh_return_handler_rtx): New prototype.
39417 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39419 * config/rs6000/altivec.h (vec_rlmi): New #define.
39420 (vec_vrlnm): Likewise.
39421 (vec_rlnm): Likewise.
39422 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39423 (UNSPEC_VRLNM): Likewise.
39424 (VIlong): New mode iterator.
39425 (altivec_vrl<VI_char>mi): New define_insn.
39426 (altivec_vrl<VI_char>nm): Likewise.
39427 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39429 (VRLDNM): Likewise.
39430 (RLNM): New polymorphic function entry.
39431 (VRLWMI): New monomorphic function entry.
39432 (VRLDMI): Likewise.
39433 (RLMI): New polymorphic function entry.
39434 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39435 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39436 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39439 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39442 * dwarf2out.c (field_byte_offset): Restore the
39443 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39444 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39445 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39448 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39451 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39453 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39456 * dwarf2out.c (add_data_member_location_attribute): For constant
39457 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39458 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39459 DW_AT_byte_size attributes.
39461 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39463 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39464 after forcing to constant memory when the code model is medium.
39466 2017-01-17 Julia Koval <julia.koval@intel.com>
39469 * config/i386/avx512fintrin.h
39470 (_mm512_i32gather_ps): Change __addr type to void const*.
39471 (_mm512_mask_i32gather_ps): Ditto.
39472 (_mm512_i32gather_pd): Ditto.
39473 (_mm512_mask_i32gather_pd): Ditto.
39474 (_mm512_i64gather_ps): Ditto.
39475 (_mm512_mask_i64gather_ps): Ditto.
39476 (_mm512_i64gather_pd): Ditto.
39477 (_mm512_mask_i64gather_pd): Ditto.
39478 (_mm512_i32gather_epi32): Ditto.
39479 (_mm512_mask_i32gather_epi32): Ditto.
39480 (_mm512_i32gather_epi64): Ditto.
39481 (_mm512_mask_i32gather_epi64): Ditto.
39482 (_mm512_i64gather_epi32): Ditto.
39483 (_mm512_mask_i64gather_epi32): Ditto.
39484 (_mm512_i64gather_epi64): Ditto.
39485 (_mm512_mask_i64gather_epi64): Ditto.
39486 (_mm512_i32scatter_ps): Change __addr type to void*.
39487 (_mm512_mask_i32scatter_ps): Ditto.
39488 (_mm512_i32scatter_pd): Ditto.
39489 (_mm512_mask_i32scatter_pd): Ditto.
39490 (_mm512_i64scatter_ps): Ditto.
39491 (_mm512_mask_i64scatter_ps): Ditto.
39492 (_mm512_i64scatter_pd): Ditto.
39493 (_mm512_mask_i64scatter_pd): Ditto.
39494 (_mm512_i32scatter_epi32): Ditto.
39495 (_mm512_mask_i32scatter_epi32): Ditto.
39496 (_mm512_i32scatter_epi64): Ditto.
39497 (_mm512_mask_i32scatter_epi64): Ditto.
39498 (_mm512_i64scatter_epi32): Ditto.
39499 (_mm512_mask_i64scatter_epi32): Ditto.
39500 (_mm512_i64scatter_epi64): Ditto.
39501 (_mm512_mask_i64scatter_epi64): Ditto.
39502 * config/i386/avx512pfintrin.h
39503 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39504 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39505 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39506 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39507 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39508 (_mm512_prefetch_i32scatter_ps): Ditto.
39509 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39510 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39511 (_mm512_prefetch_i64scatter_pd): Ditto.
39512 (_mm512_prefetch_i64scatter_ps): Ditto.
39513 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39514 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39515 * config/i386/avx512vlintrin.h
39516 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39517 (_mm_mmask_i32gather_ps): Ditto.
39518 (_mm256_mmask_i32gather_pd): Ditto.
39519 (_mm_mmask_i32gather_pd): Ditto.
39520 (_mm256_mmask_i64gather_ps): Ditto.
39521 (_mm_mmask_i64gather_ps): Ditto.
39522 (_mm256_mmask_i64gather_pd): Ditto.
39523 (_mm_mmask_i64gather_pd): Ditto.
39524 (_mm256_mmask_i32gather_epi32): Ditto.
39525 (_mm_mmask_i32gather_epi32): Ditto.
39526 (_mm256_mmask_i32gather_epi64): Ditto.
39527 (_mm_mmask_i32gather_epi64): Ditto.
39528 (_mm256_mmask_i64gather_epi32): Ditto.
39529 (_mm_mmask_i64gather_epi32): Ditto.
39530 (_mm256_mmask_i64gather_epi64): Ditto.
39531 (_mm_mmask_i64gather_epi64): Ditto.
39532 (_mm256_i32scatter_ps): Change __addr type to void*.
39533 (_mm256_mask_i32scatter_ps): Ditto.
39534 (_mm_i32scatter_ps): Ditto.
39535 (_mm_mask_i32scatter_ps): Ditto.
39536 (_mm256_i32scatter_pd): Ditto.
39537 (_mm256_mask_i32scatter_pd): Ditto.
39538 (_mm_i32scatter_pd): Ditto.
39539 (_mm_mask_i32scatter_pd): Ditto.
39540 (_mm256_i64scatter_ps): Ditto.
39541 (_mm256_mask_i64scatter_ps): Ditto.
39542 (_mm_i64scatter_ps): Ditto.
39543 (_mm_mask_i64scatter_ps): Ditto.
39544 (_mm256_i64scatter_pd): Ditto.
39545 (_mm256_mask_i64scatter_pd): Ditto.
39546 (_mm_i64scatter_pd): Ditto.
39547 (_mm_mask_i64scatter_pd): Ditto.
39548 (_mm256_i32scatter_epi32): Ditto.
39549 (_mm256_mask_i32scatter_epi32): Ditto.
39550 (_mm_i32scatter_epi32): Ditto.
39551 (_mm_mask_i32scatter_epi32): Ditto.
39552 (_mm256_i32scatter_epi64): Ditto.
39553 (_mm256_mask_i32scatter_epi64): Ditto.
39554 (_mm_i32scatter_epi64): Ditto.
39555 (_mm_mask_i32scatter_epi64): Ditto.
39556 (_mm256_i64scatter_epi32): Ditto.
39557 (_mm256_mask_i64scatter_epi32): Ditto.
39558 (_mm_i64scatter_epi32): Ditto.
39559 (_mm_mask_i64scatter_epi32): Ditto.
39560 (_mm256_i64scatter_epi64): Ditto.
39561 (_mm256_mask_i64scatter_epi64): Ditto.
39562 (_mm_i64scatter_epi64): Ditto.
39563 (_mm_mask_i64scatter_epi64): Ditto.
39564 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39565 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39566 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39567 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39568 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39569 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39570 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39571 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39572 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39573 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39574 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39575 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39576 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39577 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39578 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39579 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39580 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39581 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39582 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39583 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39584 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39585 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39586 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39587 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39588 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39589 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39590 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39591 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39592 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39593 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39594 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39595 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39596 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39597 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39598 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39599 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39600 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39601 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39602 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39603 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39604 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39605 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39606 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39607 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39608 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39609 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39610 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39611 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39612 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39613 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39614 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39615 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39616 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39617 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39618 definitions accordingly.
39620 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39621 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39624 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39627 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39630 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39631 hard regs for inner regno.
39633 2017-01-17 Martin Liska <mliska@suse.cz>
39636 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39637 assumption and add comment.
39639 2017-01-17 Nathan Sidwell <nathan@acm.org>
39641 * ipa-visibility.c (localize_node): New function, broken out of ...
39642 (function_and_variable_visibility): ... here. Call it.
39644 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39646 PR middle-end/77445
39647 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39648 correctly set frequency of oudgoing edge.
39649 (duplicate_thread_path): Fix profile updating.
39651 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39654 * configure.ac: Add GCC_BASE_VER.
39655 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39656 version from BASE-VER file.
39657 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39658 (gcc.o): Depend on $(BASEVER).
39659 * common.opt (dumpfullversion): New option.
39660 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39661 * doc/invoke.texi: Document -dumpfullversion.
39662 * doc/install.texi: Document --with-gcc-major-version-only.
39663 * configure: Regenerated.
39665 2017-01-17 Richard Biener <rguenther@suse.de>
39667 PR tree-optimization/71433
39668 * tree-vrp.c (register_new_assert_for): Merge same asserts
39669 on all incoming edges.
39670 (process_assert_insertions_for): Handle insertions at the
39673 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39675 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39676 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39678 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39681 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39684 2017-01-17 Alan Modra <amodra@gmail.com>
39687 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39688 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39689 symbolic stack limit when pic.
39691 2017-01-16 Martin Sebor <msebor@redhat.com>
39693 PR tree-optimization/78608
39694 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39696 2017-01-16 Jeff Law <law@redhat.com>
39699 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39700 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39701 for several include directories that may be relative to sysroot.
39702 * config/i386/x-mingw32 (gplus_includedir): Define.
39703 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39704 (native_system_includedir): Likewise.
39705 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39706 override if TARGET_SYSTEM_ROOT is defined.
39707 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39709 PR tree-optimization/79090
39710 PR tree-optimization/33562
39711 PR tree-optimization/61912
39712 PR tree-optimization/77485
39713 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39714 and computed trims into the dump file.
39716 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39718 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39720 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39723 * gimplify.c (gimplify_init_constructor): If want_value and
39724 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39728 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39729 sequence. Formatting fixes.
39730 (doloop_optimize): Formatting fixes.
39733 * gcc.c (debug_level_greater_than_spec_func): New function.
39734 (static_spec_functions): Add debug-level-gt spec function.
39735 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39737 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39738 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39739 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39740 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39741 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39742 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39744 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39746 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39747 QImode fixups to general and mask registers only.
39749 2017-01-16 Carl Love <cel@us.ibm.com>
39751 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39752 for built-in functions
39753 vector signed char vec_nabs (vector signed char)
39754 vector signed short vec_nabs (vector signed short)
39755 vector signed int vec_nabs (vector signed int)
39756 vector signed long long vec_nabs (vector signed long long)
39757 vector float vec_nabs (vector float)
39758 vector double vec_nabs (vector double)
39759 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39761 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39762 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39763 * doc/extend.texi: Update the documentation file for the new built-in
39766 2017-01-16 Martin Sebor <msebor@redhat.com>
39768 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39771 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39773 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39774 UNSPEC_VSX__XXSPLTD to require special splat handling.
39776 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39779 * system.h: Poison strndup.
39781 2017-01-16 Alan Modra <amodra@gmail.com>
39784 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39787 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39789 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39791 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39793 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39794 call recog here. Assert that INSN_CODE (insn) is non-negative.
39796 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39799 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39801 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39802 in the currently scheduled RTL fragment.
39804 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39806 PR rtl-optimization/78751
39807 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39810 2017-01-14 Jeff Law <law@redhat.com>
39812 PR tree-optimization/79090
39813 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39814 variable length stores.
39815 (compute_trims): Delete dead assignment to *trim_tail.
39816 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39819 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39821 PR rtl-optimization/78626
39822 PR rtl-optimization/78727
39823 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39824 of a block, and split such blocks after everything else is finished.
39826 2017-01-14 Alan Modra <amodra@gmail.com>
39829 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39830 target legitimate_combined_insn.
39831 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39832 (rs6000_legitimate_combined_insn): New function.
39833 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39835 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39836 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39837 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39839 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39841 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39843 2017-01-13 Jeff Law <law@redhat.com>
39845 PR tree-optimization/33562
39846 PR tree-optimization/61912
39847 PR tree-optimization/77485
39848 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39850 (delete_dead_assignment): Likewise.
39851 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
39852 statement to delete_dead_call and delete_dead_assignment.
39854 2017-01-13 David Malcolm <dmalcolm@redhat.com>
39857 * substring-locations.c (format_warning_va): Strengthen case 1 so
39858 that both endpoints of the substring must be within the format
39859 range for just the substring to be printed.
39861 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
39863 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
39864 * config/i386/i386.c (ix86_target_string): Add missing options
39865 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
39866 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
39867 flags_other and ix86_target_other to flags2_other. Display unknown
39869 (ix86_valid_target_attribute_inner_p): Add missing options and
39870 reorder options by implied ISAs, as in ix86_target_string.
39872 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39874 * hash-table.h (hash_table::too_empty_p): New function.
39875 (hash_table::expand): Use it.
39876 (hash_table::traverse): Likewise.
39877 (hash_table::empty_slot): Use sizeof (value_type) instead of
39878 sizeof (PTR) to convert bytes to elements. Shrink the table
39879 if the current size is excessive for the current number of
39882 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39884 * ira-costs.c (record_reg_classes): Break from the inner loop
39885 early once alt_fail is known to be true. Update outer loop
39886 handling accordingly.
39888 2017-01-13 Jeff Law <law@redhat.com>
39890 * tree-ssa-dse.c (decrement_count): New function.
39891 (increment_start_addr, maybe_trim_memstar_call): Likewise.
39892 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
39893 when we know the partially dead statement is a mem* function.
39895 PR tree-optimization/61912
39896 PR tree-optimization/77485
39897 * tree-ssa-dse.c: Include expr.h.
39898 (maybe_trim_constructor_store): New function.
39899 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
39901 PR tree-optimization/33562
39902 PR tree-optimization/61912
39903 PR tree-optimization/77485
39904 * doc/invoke.texi: Document new dse-max-object-size param.
39905 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
39906 * tree-ssa-dse.c: Include params.h.
39907 (dse_store_status): New enum.
39908 (initialize_ao_ref_for_dse): New, partially extracted from
39910 (valid_ao_ref_for_dse, normalize_ref): New.
39911 (setup_live_bytes_from_ref, compute_trims): Likewise.
39912 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
39913 (maybe_trim_partially_dead_store): Likewise.
39914 (maybe_trim_complex_store): Likewise.
39915 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
39916 Track what bytes live from the original store. Return tri-state
39917 for dead, partially dead or live.
39918 (dse_dom_walker): Add constructor, destructor and new private members.
39919 (delete_dead_call, delete_dead_assignment): New extracted from
39921 (dse_optimize_stmt): Make a member of dse_dom_walker.
39922 Use initialize_ao_ref_for_dse.
39924 PR tree-optimization/33562
39925 PR tree-optimization/61912
39926 PR tree-optimization/77485
39927 * sbitmap.h (bitmap_count_bits): Prototype.
39928 (bitmap_clear_range, bitmap_set_range): Likewise.
39929 * sbitmap.c (bitmap_clear_range): New function.
39930 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
39932 2017-01-13 Martin Liska <mliska@suse.cz>
39935 * function.c (set_cfun): Add new argument force.
39936 * function.h (set_cfun): Likewise.
39937 * ipa-inline-transform.c (inline_call): Use the function when
39938 strict alising from is dropped for function we inline to.
39940 2017-01-13 Richard Biener <rguenther@suse.de>
39942 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
39943 for dumping GIMPLE INTEGER_CSTs.
39945 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
39947 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
39948 to 201112L since C++17.
39950 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
39953 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
39954 if -fsanitize=kernel-address is present.
39956 2017-01-13 Richard Biener <rguenther@suse.de>
39958 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
39959 as _Literal ( type ) number in case usual suffixes do not
39960 preserve all information.
39962 2017-01-13 Richard Biener <rguenther@suse.de>
39964 PR tree-optimization/77283
39965 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
39966 and ssa-iterators.h.
39967 (is_feasible_trace): Implement a cost model based on joiner
39970 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
39973 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
39974 char or short to __float128/_Float128 directly.
39976 2017-01-12 Martin Sebor <msebor@redhat.com>
39978 to -Wformat-overflow.
39979 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
39980 (min_bytes_remaining): Same.
39981 (get_string_length): Same.
39982 (format_string): Same.
39983 (format_directive): Same.
39985 (pass_sprintf_length::handle_gimple_call): Same.
39987 2017-01-12 Jakub Jelinek <jakub@redhat.com>
39989 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
39990 info.nowrite calls with no lhs that can't throw. Return bool
39991 whether gsi_remove has been called or not.
39992 (pass_sprintf_length::handle_gimple_call): Return bool whether
39993 try_substitute_return_value called gsi_remove. Formatting fix.
39994 (pass_sprintf_length::execute): Don't use gsi_remove if
39995 handle_gimple_call returned true.
39998 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
39999 be removed due to side-effects, don't remove following barrier nor
40000 turn the successor edge into fallthru edge.
40002 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40005 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40006 element-reversing loads and stores as not swappable.
40008 2017-01-12 Nathan Sidwell <nathan@acm.org>
40009 Nicolai Stange <nicstange@gmail.com>
40011 * combine.c (try_combine): Don't ignore result of overlap checking
40012 loop. Combine overlap & asm check into single loop.
40014 2017-01-12 Richard Biener <rguenther@suse.de>
40016 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40017 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40019 2017-01-12 Richard Biener <rguenther@suse.de>
40021 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40022 and TS_TARGET_OPTION directly derive from TS_BASE.
40023 * tree-core.h (tree_optimization_option): Derive from tree_base.
40024 (tree_target_option): Likewise.
40026 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40028 * config/i386/i386.c (memory_address_length): Increase len
40029 only when rip_relative_addr_p returns false.
40031 2017-01-11 Julia Koval <julia.koval@intel.com>
40033 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40034 (OPTION_MASK_ISA_SGX_SET): New.
40035 (ix86_handle_option): Handle OPT_msgx.
40036 * config.gcc: Added sgxintrin.h.
40037 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40038 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40039 * config/i386/i386.c (ix86_target_string): Add -msgx.
40041 (ix86_option_override_internal): Handle new options.
40042 (ix86_valid_target_attribute_inner_p): Add sgx.
40043 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40044 * config/i386/i386.opt: Add msgx.
40045 * config/i386/sgxintrin.h: New file.
40046 * config/i386/x86intrin.h: Add sgxintrin.h.
40048 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40051 * fold-const.c (maybe_nonzero_address): Return 1 for function
40053 (tree_single_nonzero_warnv_p): Don't handle function local objects
40057 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40060 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40063 * opts.c: Include "spellcheck.h"
40064 (struct string_fragment): New struct.
40065 (struct edit_distance_traits<const string_fragment &>): New
40067 (get_closest_sanitizer_option): New function.
40068 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40070 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40072 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40074 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40075 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40076 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40077 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40078 for initial die_offset if dwarf_split_debug_info.
40079 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40080 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40082 (output_skeleton_debug_sections): Formatting fix. Use
40083 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40084 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40086 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40088 * config/arm/cortex-a53.md: Add bypasses for
40089 cortex_a53_r2f_cvt.
40090 (cortex_a53_r2f): Only use for transfers.
40091 (cortex_a53_f2r): Likewise.
40092 (cortex_a53_r2f_cvt): Add reservation for conversions.
40093 (cortex_a53_f2r_cvt): Likewise.
40095 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40097 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40098 to all inlined functions, change static to extern.
40100 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40103 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40105 (arm_assemble_integer): Likewise.
40107 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40109 * config.gcc: Use new awk script to check CPU, FPU and architecture
40110 parameters for --with-... options.
40111 * config/arm/parsecpu.awk: New file
40112 * config/arm/arm-cpus.in: New file.
40113 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40115 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40117 * config/arm/t-arm: Update dependency rules.
40118 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40119 of processing .def files.
40120 * config/arm/genopt.sh: Deleted.
40121 * config/arm/gentune.sh: Deleted.
40122 * config/arm/arm-cores.def: Deleted.
40123 * config/arm/arm-arches.def: Deleted.
40124 * config/arm/arm-fpus.def: Deleted.
40125 * config/arm/arm-tune.md: Regenerated.
40126 * config/arm/arm-tables.opt: Regenerated.
40127 * config/arm/arm-cpu.h: New generated file.
40128 * config/arm/arm-cpu-data.h: New generated file.
40129 * config/arm/arm-cpu-cdata.h: New generated file.
40131 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40134 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40136 (input_varpool_node): Unpack dynamically_initialized bit.
40138 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40140 PR rtl-optimization/79032
40141 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40142 the alignment of the adjusted memory reference against that of MODE,
40143 instead of the alignment of the original memory reference.
40145 2017-01-11 Martin Jambor <mjambor@suse.cz>
40147 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40149 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40150 decorated functions.
40152 2017-01-11 Richard Biener <rguenther@suse.de>
40154 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40155 set range/nonnull info for PHI results. Do not set it on
40156 stmts marked for removal.
40158 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40160 * expr.c (store_field): In the bitfield case, fetch the return value
40161 from the registers before applying a single big-endian adjustment.
40162 Always do a final load for a BLKmode value not larger than a word.
40164 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40167 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40168 that we correctly handle column numbers greater than
40169 LINE_MAP_MAX_COLUMN_NUMBER.
40171 2017-01-10 Martin Sebor <msebor@redhat.com>
40173 PR middle-end/78245
40174 * gimple-ssa-sprintf.c (get_destination_size): Call
40175 {init,fini}object_sizes.
40176 * tree-object-size.c (addr_object_size): Adjust.
40177 (pass_through_call): Adjust.
40178 (pass_object_sizes::execute): Adjust.
40179 * tree-object-size.h (fini_object_sizes): Declare.
40181 2017-01-10 Martin Sebor <msebor@redhat.com>
40183 PR tree-optimization/78775
40184 * builtins.c (get_size_range): Move...
40185 * calls.c: ...to here.
40186 (alloc_max_size): Accept zero argument.
40187 (operand_signed_p): Remove.
40188 (maybe_warn_alloc_args_overflow): Call get_size_range.
40189 * calls.h (get_size_range): Declare.
40191 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40193 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40194 from TI's devices.csv file as of September 2016.
40195 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40197 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40199 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40200 * doc/invoke.texi: Likewise.
40201 * doc/md.texi: Likewise.
40202 * doc/objc.texi: Likewise.
40204 2017-01-10 Joshua Conner <joshconner@google.com>
40206 * config/arm/fuchsia-elf.h: New file.
40207 * config/fuchsia.h: New file.
40208 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40209 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40211 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40213 2016-01-10 Richard Biener <rguenther@suse.de>
40215 PR tree-optimization/79034
40216 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40217 Propagate out degenerate PHIs in the joiner.
40219 2017-01-10 Martin Liska <mliska@suse.cz>
40221 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40222 (sort_congruence_classes_by_decl_uid): Likewise.
40223 (sort_congruence_class_groups_by_decl_uid): Likewise.
40224 (sem_item_optimizer::merge_classes): Sort class, groups in these
40225 classes and members in the groups by DECL_UID of declarations.
40226 This would make merge operations stable.
40228 2017-01-10 Martin Liska <mliska@suse.cz>
40230 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40231 usage of m_classes_vec.
40232 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40233 (sem_item_optimizer::get_group_by_hash): Likewise.
40234 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40235 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40236 (sem_item_optimizer::verify_classes): Likewise.
40237 (sem_item_optimizer::process_cong_reduction): Likewise.
40238 (sem_item_optimizer::dump_cong_classes): Likewise.
40239 (sem_item_optimizer::merge_classes): Likewise.
40240 * ipa-icf.h (congruence_class_hash): Rename from
40241 congruence_class_group_hash. Remove declaration of m_classes_vec.
40243 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40245 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40246 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40247 * config.gcc: Add avx512vpopcntdqintrin.h.
40248 * config/i386/avx512vpopcntdqintrin.h: New.
40249 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40250 * config/i386/i386-builtin-types.def: Add new types.
40251 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40252 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40253 __builtin_ia32_vpopcountq_v8di_mask): New.
40254 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40255 __AVX512VPOPCNTDQ__.
40256 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40257 (PTA_AVX512VPOPCNTDQ): Define.
40258 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40259 TARGET_AVX512VPOPCNTDQ_P): Define.
40260 * config/i386/i386.opt: Add mavx512vpopcntdq.
40261 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40262 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40264 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40266 PR middle-end/77484
40267 * predict.def (PRED_CALL): Set to 67.
40269 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40271 * expr.c (store_field): In the bitfield case, if the value comes from
40272 a function call and is of an aggregate type returned in registers, do
40273 not modify the field mode; extract the value in all cases if the mode
40274 is BLKmode and the size is not larger than a word.
40276 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40279 * config/i386/cpuid.h: Fix undefined behavior.
40281 2017-01-04 Jeff Law <law@redhat.com>
40283 PR tree-optimization/79007
40284 PR tree-optimization/67955
40285 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40286 conservative for pt.null when flag_non_call_exceptions is on.
40288 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40290 PR translation/79019
40291 PR translation/79020
40292 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40293 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40294 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40296 * config/avr/avr.opt (maccumulate-args): Likewise.
40297 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40298 * common.opt (freport-bug): Likewise.
40299 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40300 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40301 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40302 translatable string.
40303 * config/i386/i386.c (function_value_32): Likewise.
40304 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40305 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40307 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40308 * common/config/msp430/msp430-common.c (msp430_handle_option):
40310 * symtab.c (symtab_node::verify_base): Likewise.
40311 * opts.c (set_debug_level): Likewise.
40312 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40313 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40314 missing whitespace to translatable strings.
40315 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40316 * config/sh/superh.h: Likewise.
40317 * config/i386/xopintrin.h: Likewise.
40318 * config/i386/znver1.md: Likewise.
40319 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40320 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40321 * double-int.h (struct double_int): Likewise.
40322 * double-int.c (div_and_round_double): Likewise.
40323 * wide-int.cc: Likewise.
40324 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40325 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40326 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40327 (crossjumps_occurred): ... this.
40328 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40331 PR tree-optimization/78899
40332 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40333 returning bool return struct loop *, NULL for failure and the new
40335 (versionable_outer_loop_p): Don't version outer loop if it has
40336 dont_vectorized bit set.
40337 (tree_if_conversion): When versioning outer loop, ensure
40338 tree_if_conversion is performed also on the inner loop of the
40339 non-vectorizable outer loop copy.
40340 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40341 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40342 prevent vectorization of it.
40343 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40344 the outer loop vectorization of the non-scalar version is attempted
40345 before vectorization of the inner loop in scalar version. If
40346 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40347 vectorization of its inner loop.
40348 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40349 has 2 inner loops, rename also on edges from bb whose single pred
40350 is outer_loop->header. Fix typo in function comment.
40352 2017-01-09 Martin Sebor <msebor@redhat.com>
40355 * asan.c (asan_emit_stack_protection): Increase local buffer size
40356 to avoid snprintf truncation warning.
40358 2017-01-09 Andrew Pinski <apinski@cavium.com>
40360 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40361 to reference thunderx2t99 for the tuning structure
40362 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40364 (thunderx2t99_extra_costs): This.
40365 * config/aarch64/aarch64-tune.md: Regenerate.
40366 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40367 (vulcan_addrcost_table): This.
40368 (vulcan_regmove_cost): Rename to ...
40369 (thunderx2t99_regmove_cost): This.
40370 (vulcan_vector_cost): Rename to ...
40371 (thunderx2t99_vector_cost): this.
40372 (vulcan_branch_cost): Rename to ...
40373 (thunderx2t99_branch_cost): This.
40374 (vulcan_tunings): Rename to ...
40375 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40376 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40378 2017-01-09 Martin Jambor <mjambor@suse.cz>
40382 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40383 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40384 (propagate_vr_accross_jump_function): Use the above function for all
40385 value range computations for pass-through jump functions and type
40386 converasion from explicit value range values.
40387 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40388 parameters from TYPE_ARG_TYPES.
40389 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40390 (ipa_write_node_info): Stream type of the actual argument.
40391 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40393 2017-01-09 Martin Liska <mliska@suse.cz>
40396 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40397 (lookup_compiler): Do not show error message with have_E.
40399 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40401 PR tree-optimization/78938
40402 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40403 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40404 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40405 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40408 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40410 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40413 2017-01-09 Richard Biener <rguenther@suse.de>
40415 PR tree-optimization/78997
40416 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40417 name condition properly.
40419 2017-01-09 Richard Biener <rguenther@suse.de>
40422 * dwarf2out.c (is_cxx): New overload with context.
40423 (is_naming_typedef_decl): Use it.
40425 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40427 * invoke.texi (Option Summary): Correct spacing in option lists
40428 and add line breaks to fix over-long lines.
40430 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40432 PR middle-end/17660
40434 * extend.texi (Common Variable Attributes): Add xref to GCC
40435 Internals manual to explain mode attribute keywords.
40437 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40440 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40441 and Preprocessor Options.
40442 (Options for Linking): Document -pthread here....
40443 (RS/6000 and PowerPC Options): ...not here.
40444 (Solaris 2 Options): ...or here.
40445 * doc/cppopts.texi: Document -pthread.
40447 2017-01-08 Martin Sebor <msebor@redhat.com>
40449 PR middle-end/77708
40450 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40451 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40452 New member functions.
40453 (format_directive): Used them.
40455 (pass_sprintf_length::handle_gimple_call): Same.
40456 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40457 to avoid truncation for any argument.
40458 (extract_affine_mul): Same.
40459 * tree.c (get_file_function_name): Same.
40461 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40463 PR middle-end/77484
40464 * predict.def (PRED_INDIR_CALL): Set to 86.
40466 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40468 PR preprocessor/54124
40469 * doc/cppopts.texi: Reformat -d subtable to list the full name
40470 of the options. Add cross-reference to the docs for the general
40471 compiler -d options.
40472 * doc/invoke.texi (Developer Options): Add cross-reference to the
40473 preprocessor-specific -d option documentation.
40475 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40477 PR preprocessor/13498
40478 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40479 redudant material, and reflect new command-line options.
40480 (System Headers): Likewise.
40482 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40484 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40485 -isystem, and -idirafter. Copy-edit.
40486 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40487 default for -ftrack-macro-expansion. Delete obsolete and
40488 badly-formatted implementation details about -fdebug-cpp output.
40489 * doc/cppwarnopts.texi: Copy-edit.
40491 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40494 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40495 that the transition from a max line width >= 1<<10 to narrower
40496 lines works correctly.
40498 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40500 * doc/options.texi (PerFunction): New.
40501 * opt-functions.awk (switch_flags): Map both Optimization and
40502 PerFunction to CL_OPTIMIZATION.
40503 * opth-gen.awk: Test for PerFunction flag along with
40505 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40506 it only when the latter is present. Skip those that don't in
40507 the hash function generator.
40508 * common.opt (fvar-tracking): Mark as PerFunction instead of
40510 (fvar-tracking-assignments): Likewise.
40511 (fvar-tracking-assignments-toggle): Likewise.
40512 (fvar-tracking-uninit): Likewise.
40514 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40516 PR translation/79018
40517 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40520 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40523 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40524 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40525 TARGET_LONG_JUMP_TABLE_OFFSETS.
40526 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40527 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40528 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40529 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40530 * config/m68k/m68k.md (tablejump expander): Likewise.
40531 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40532 TARGET_LONG_JUMP_TABLE_OFFSETS.
40533 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40534 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40536 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40537 David Holsgrove <david.holsgrove@xilinx.com>
40539 * common/config/microblaze/microblaze-common.c
40540 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40541 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40543 * config/microblaze/microblaze.c (microblaze_must_save_register)
40544 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40546 (microblaze_eh_return): New function.
40547 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40548 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40549 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40550 * config/microblaze/microblaze.md (eh_return): New pattern.
40552 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40554 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40555 GCC_DIAGNOSTIC_STRINGIFY): Define.
40557 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40559 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40561 * config/arm/arm.md (<mcrr>): New.
40563 * config/arm/arm.c (arm_arch5te): New.
40564 (arm_option_override): Set arm_arch5te.
40565 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40567 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40568 (arm_mcrr_qualifiers): ... this. New.
40569 (MRRC_QUALIFIERS): Define to...
40570 (arm_mrrc_qualifiers): ... this. New.
40571 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40573 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40574 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40575 (MRRCI, mrrc, MRRC): New.
40576 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40577 VUNSPEC_MRRC2): New.
40579 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40581 * config/arm/arm.md (<mcr>): New.
40583 * config/arm/arm.c (arm_coproc_builtin_available): Add
40584 support for mcr, mrc, mcr2 and mrc2.
40585 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40586 (arm_mcr_qualifiers): ... this. New.
40587 (MRC_QUALIFIERS): Define to ...
40588 (arm_mrc_qualifiers): ... this. New.
40589 (MCR_QUALIFIERS): Define to ...
40590 (arm_mcr_qualifiers): ... this. New.
40591 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40593 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40594 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40595 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40596 VUNSPEC_MRC2): New.
40598 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40600 * config/arm/arm.md (*ldc): New.
40604 * config/arm/arm.c (arm_coproc_builtin_available): Add
40605 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40606 (arm_coproc_ldc_stc_legitimate_address): New.
40607 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40608 'qualifier_const_pointer'.
40609 (LDC_QUALIFIERS): Define to...
40610 (arm_ldc_qualifiers): ... this. New.
40611 (STC_QUALIFIERS): Define to...
40612 (arm_stc_qualifiers): ... this. New.
40613 * config/arm/arm-protos.h
40614 (arm_coproc_ldc_stc_legitimate_address): New.
40615 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40616 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40617 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40618 stc2, stcl, stc2l): New.
40619 * config/arm/constraints.md (Uz): New.
40620 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40621 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40622 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40623 VUNSPEC_STC2L): New.
40625 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40627 * config/arm/arm.md (<cdp>): New.
40628 * config/arm/arm.c (neon_const_bounds): Rename this ...
40629 (arm_const_bounds): ... this.
40630 (arm_coproc_builtin_available): New.
40631 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40632 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40633 (CDP_QUALIFIERS): Define to...
40634 (arm_cdp_qualifiers): ... this. New.
40636 (arm_expand_builtin_args): Add case for 6 arguments.
40637 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40638 (arm_const_bounds): ... this.
40639 (arm_coproc_builtin_available): New.
40640 * config/arm/arm_acle.h (__arm_cdp): New.
40642 * config/arm/arm_acle_builtins.def (cdp): New.
40644 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40645 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40646 'arm_const_bounds'.
40647 * config/arm/types.md (coproc): New.
40648 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40649 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40650 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40651 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40653 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40655 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40656 (UBINOP_QUALIFIERS): New.
40658 (acle_builtin_data): New. Change comment.
40659 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40660 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40661 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40662 arm_acle_builtins.def.
40663 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40664 (arm_init_acle_builtins): New.
40665 (CRC32_BUILTIN): Remove.
40666 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40667 crc32cb, crc32ch and crc32cw.
40668 (arm_init_crc32_builtins): Remove.
40669 (arm_init_builtins): Use arm_init_acle_builtins rather
40670 than arm_init_crc32_builtins.
40671 (arm_expand_acle_builtin): New.
40672 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40673 * config/arm/arm_acle_builtins.def: New.
40675 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40677 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40678 (arm_builtin_datum): ... this.
40679 (arm_init_neon_builtin): Rename to ...
40680 (arm_init_builtin): ... this. Add a new parameters PREFIX
40681 and USE_SIG_IN_NAME.
40682 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40683 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40684 'arm_builtin_datum'.
40685 (arm_init_vfp_builtins): Likewise.
40686 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40687 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40688 (arm_expand_neon_args): Rename to ...
40689 (arm_expand_builtin_args): ... this. Rename builtin_arg
40690 enum values and differentiate between ARG_BUILTIN_MEMORY
40691 and ARG_BUILTIN_NEON_MEMORY.
40692 (arm_expand_neon_builtin_1): Rename to ...
40693 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40694 values, arm_expand_builtin_args and add bool parameter NEON.
40695 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40696 (arm_expand_vfp_builtin): Likewise.
40697 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40699 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40701 PR middle-end/77484
40702 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40703 * predict.c (tree_estimate_probability_bb): Reverse direction of
40704 polymorphic call predictor.
40706 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40708 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40709 (determine_pass_name_match): ...this new function and...
40710 (should_skip_pass_p): ...this new function.
40712 2017-01-06 Nathan Sidwell <nathan@acm.org>
40714 * ipa-visibility.c (function_and_variable_visibility): Reformat
40715 comments and long lines. Remove extrneous if.
40716 * symtab.c (symtab_node::make_decl_local): Fix code format.
40717 (symtab_node::set_section_for_node): Fix comment typo.
40719 2017-01-06 Martin Liska <mliska@suse.cz>
40722 * lra-constraints.c: Rename invariant to lra_invariant.
40723 * predict.c (set_even_probabilities): Initialize e to NULL.
40725 2017-01-05 Martin Sebor <msebor@redhat.com>
40727 PR tree-optimization/78910
40728 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40729 (format_integer): Correct off-by-one error in the handling
40730 of precision with negative numbers in signed conversions..
40732 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40734 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40736 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40738 PR tree-optimization/71016
40739 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40740 factor_out_conditional_conversion. Formatting fix.
40741 (factor_out_conditional_conversion): Add cond_stmt argument.
40742 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40743 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40746 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40748 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40749 read-rtl-function.o, and selftest-rtl.o.
40750 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40751 (selftest::aarch64_test_loading_full_dump): New function.
40752 (selftest::aarch64_run_selftests): New function.
40753 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40754 selftest::aarch64_run_selftests.
40755 * config/i386/i386.c
40756 (selftest::ix86_test_loading_dump_fragment_1): New function.
40757 (selftest::ix86_test_loading_call_insn): New function.
40758 (selftest::ix86_test_loading_full_dump): New function.
40759 (selftest::ix86_test_loading_unspec): New function.
40760 (selftest::ix86_run_selftests): Call the new functions.
40761 * emit-rtl.c (maybe_set_max_label_num): New function.
40762 * emit-rtl.h (maybe_set_max_label_num): New decl.
40763 * function.c (instantiate_decls): Guard call to
40764 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40765 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40767 * gensupport.c (gen_reader::gen_reader): Pass "false"
40768 for new "compact" param of rtx_reader.
40769 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40770 rather than an empty string for NULL strings.
40771 * read-md.c: Potentially include config.h rather than bconfig.h.
40772 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40773 (have_error): New global, copied from errors.c.
40774 (md_reader::read_name): Rename to...
40775 (md_reader::read_name_1): ...this, adding "out_loc" param,
40776 and converting "missing name or number" to returning false, rather
40778 (md_reader::read_name): Reimplement in terms of read_name_1.
40779 (md_reader::read_name_or_nil): New function.
40780 (md_reader::read_string): Handle "(nil)" by returning NULL.
40781 (md_reader::md_reader): Add new param "compact".
40782 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40783 (md_reader::read_file): New method.
40784 * read-md.h (md_reader::md_reader): Add new param "compact".
40785 (md_reader::read_file): New method.
40786 (md_reader::is_compact): New accessor.
40787 (md_reader::read_name): Convert return type from void to file_location.
40788 (md_reader::read_name_or_nil): New decl.
40789 (md_reader::read_name_1): New decl.
40790 (md_reader::m_compact): New field.
40791 (noop_reader::noop_reader): Pass "false" for new "compact" param
40793 (rtx_reader::rtx_reader): Add new "compact" param.
40794 (rtx_reader::read_rtx_operand): Make virtual and convert return
40795 type from void to rtx.
40796 (rtx_reader::read_until): New decl.
40797 (rtx_reader::handle_any_trailing_information): New virtual function.
40798 (rtx_reader::postprocess): New virtual function.
40799 (rtx_reader::finalize_string): New virtual function.
40800 (rtx_reader::m_in_call_function_usage): New field.
40801 (rtx_reader::m_reuse_rtx_by_id): New field.
40802 * read-rtl-function.c: New file.
40803 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40804 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40805 (selftest::verify_three_block_rtl_cfg): New decl.
40806 * read-rtl-function.h: New file.
40807 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40808 For host, include function.h, memmodel.h, and emit-rtl.h.
40809 (one_time_initialization): New function.
40810 (struct compact_insn_name): New struct.
40811 (compact_insn_names): New array.
40812 (find_code): Handle insn codes in compact dumps.
40813 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40814 (bind_subst_iter_and_attr): Likewise.
40815 (add_condition_to_string): Likewise.
40816 (add_condition_to_rtx): Likewise.
40817 (apply_attribute_uses): Likewise.
40818 (add_current_iterators): Likewise.
40819 (apply_iterators): Likewise.
40820 (initialize_iterators): Guard usage of apply_subst_iterator with
40821 #ifdef GENERATOR_FILE.
40822 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40823 (md_reader::read_mapping): Likewise.
40824 (add_define_attr_for_define_subst): Likewise.
40825 (add_define_subst_attr): Likewise.
40826 (read_subst_mapping): Likewise.
40827 (check_code_iterator): Likewise.
40828 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40830 (one_time_initialization): New function.
40831 (rtx_reader::read_until): New method.
40832 (read_flags): New function.
40833 (parse_reg_note_name): New function.
40834 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40835 Handle reuse_rtx ids.
40836 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40837 Add parsing support for RTL dumps, mirroring the special-cases in
40838 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40839 values, and calling handle_any_trailing_information.
40840 (rtx_reader::read_rtx_operand): Convert return type from void
40841 to rtx, returning return_rtx. Handle case 'e'. Call
40842 finalize_string on XSTR and XTMPL fields.
40843 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40844 "(nil)" values were omitted. Call the postprocess vfunc on the
40846 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40847 class ctor. Initialize m_in_call_function_usage. Call
40848 one_time_initialization.
40849 * rtl-tests.c (selftest::test_uncond_jump): Call
40850 set_new_first_and_last_insn.
40851 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
40852 * selftest-rtl.c: New file.
40853 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
40854 (selftest::get_insn_by_uid): New decl.
40855 * selftest-run-tests.c (selftest::run_tests): Call
40856 read_rtl_function_c_tests.
40857 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
40858 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
40861 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
40863 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
40864 operands in a special way. Assert that pos+len <= mode precision.
40866 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40868 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
40869 3 argument Alias with unlimited for the negative form.
40870 (fno-vect-cost-model): Removed.
40872 2017-01-05 Martin Liska <mliska@suse.cz>
40874 * hsa-gen.c (gen_hsa_divmod): New function.
40875 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
40877 2017-01-05 Martin Liska <mliska@suse.cz>
40880 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
40883 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40885 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
40886 small constant length operands.
40888 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40890 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
40891 between loop iterations.
40893 2017-01-05 Martin Liska <mliska@suse.cz>
40896 * gimplify.c (gimplify_decl_expr): Compare to
40897 asan_poisoned_variables instread of checking flags.
40898 (gimplify_target_expr): Likewise.
40899 (gimplify_expr): Likewise.
40900 (gimplify_function_tree): Conditionally initialize
40901 asan_poisoned_variables.
40903 2017-01-04 Jeff Law <law@redhat.com>
40905 PR tree-optimizatin/78812
40906 * rtl.h (contains_mem_rtx_p): Prototype.
40907 * ifcvt.c (containts_mem_rtx_p): Move from here to...
40908 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
40909 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
40910 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
40911 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
40913 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40915 * input.c (assert_char_at_range): Default-initialize actual_range.
40917 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40919 * df-scan.c (df_ref_create_structure): Make regno unsigned,
40920 to match the caller.
40922 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40924 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
40925 insns after final jump in test to emit dummy move.
40927 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40929 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
40930 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
40932 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40934 * multiple_target.c (create_dispatcher_calls): Init e_next.
40935 * tree-ssa-loop-split.c (split_loop): Init border.
40936 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
40939 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
40944 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
40945 (altivec_register_operand): Do not return true if the operand
40946 contains a SUBREG mixing SImode and SFmode.
40947 (vsx_register_operand): Likewise.
40948 (vsx_reg_sfsubreg_ok): New predicate.
40949 (vfloat_operand): Do not return true if the operand contains a
40950 SUBREG mixing SImode and SFmode.
40951 (vint_operand): Likewise.
40952 (vlogical_operand): Likewise.
40953 (gpc_reg_operand): Likewise.
40954 (int_reg_operand): Likewise.
40955 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
40956 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
40957 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
40959 (rs6000_emit_move_si_sf_subreg): New helper function.
40960 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
40961 fixup SUBREGs involving SImode and SFmode.
40962 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
40963 numbers for the new peephole2 optimization.
40964 (peephole2 for SFmode unions): New peephole2 to optimize cases in
40965 the GLIBC math library that do AND/IOR/XOR operations on single
40966 precision floating point.
40967 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
40968 target macros to say whether we need to avoid SUBREGs mixing
40970 (TARGET_ALLOW_SF_SUBREG): Likewise.
40971 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
40972 (UNSPEC_SI_FROM_SF): Likewise.
40973 (iorxor): Change spacing.
40974 (and_ior_xor): New iterator for AND, IOR, and XOR.
40975 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
40976 (movdi_from_sf_zero_ext): Likewise.
40977 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
40978 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
40979 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
40980 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
40981 (fms<mode>4): Likewise.
40982 (fnma<mode>4): Likewise.
40983 (fnms<mode>4): Likewise.
40984 (nfma<mode>4): Likewise.
40985 (nfms<mode>4): Likewise.
40987 2017-01-04 Marek Polacek <polacek@redhat.com>
40990 * doc/invoke.texi: Document -Wpointer-compare.
40992 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40994 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
40997 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
40998 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41001 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41004 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41005 documentation of the powerpc_popcntb_ok attribute.
41006 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41007 code to issue warning messages if a requested CPU configuration is
41008 not supported by the binary (assembler and loader) toolchain.
41009 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41010 made to define a built-in function that has been disabled.
41011 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41012 made to define a built-in function that has been disabled.
41013 (altivec_init_builtins): Add comment explaining why definition
41014 of the DST built-in functions is not preceded by an assertion
41015 check. Add assertions to prevent ICE if attempts are made to
41016 define an altivec predicate or an abs* built-in function that has
41018 (htm_init_builtins): Add comment explaining why definition of the
41019 htm built-in functions is not preceded by an assertion check.
41021 2017-01-04 Jeff Law <law@redhat.com>
41023 PR tree-optimizatin/67955
41024 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41025 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41026 the points-to solution does not include pt_null. Use DECL_PT_UID
41029 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41031 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41032 Use gen_int_mode instead of gen_lopwart for const_int operands.
41034 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41036 PR tree-optimization/71563
41037 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41038 out of range value - has low bits known to be zero.
41040 2017-01-04 Alan Modra <amodra@gmail.com>
41042 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41043 * configure: Regenerate.
41044 * config.in: Regenerate.
41046 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41049 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41050 a substring of the message, but strcmp with the whole message. Ifdef
41051 ENABLE_NLS, translate the message first using dgettext.
41053 2017-01-03 Jeff Law <law@redhat.com>
41055 PR tree-optimizatin/78856
41056 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41057 (mark_threaded_blocks): Remove code to truncate thread paths that
41058 cross multiple loop headers. Instead invalidate the cached loop
41059 iteration information and handle case of a thread path walking
41060 into an irreducible region.
41062 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41065 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41066 assertions. Add support for doing the signbit if the IEEE 128-bit
41067 floating point value is in a GPR.
41068 * config/rs6000/rs6000.md (Fsignbit): Delete.
41069 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41070 Update the length attribute if the value is in a GPR.
41071 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41072 the sign or zero extension instruction, since the value is always 0/1.
41073 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41076 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41077 extracting SImode to a GPR register so that we can generate a
41078 store, limit the vector to be in a traditional Altivec register
41079 for the vextuwrx instruction.
41081 2017-01-03 Ian Lance Taylor <iant@google.com>
41083 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41085 2017-01-03 Martin Sebor <msebor@redhat.com>
41087 PR tree-optimization/78696
41088 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41089 precision. Use MPFR for %f for greater fidelity. Correct handling
41091 (pass_sprintf_length::compute_format_length): Set width and precision
41092 specified by asrerisk to void_node for vararg functions.
41093 (try_substitute_return_value): Adjust dump output.
41095 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41097 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41099 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41101 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41102 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41103 -mlra/-mno-lra was passed to the compiler.
41105 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41107 PR rtl-optimization/65618
41108 * emit-rtl.c (try_split): Move initialization of "before" and
41109 "after" to just before the call to emit_insn_after_setloc.
41111 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41113 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41115 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41117 * dwarf2out.c (gen_enumeration_type_die): When
41118 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41120 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41122 PR tree-optimization/78965
41123 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41124 Change first argument from const call_info & to call_info &. For %n
41125 set info.nowrite to false.
41127 PR middle-end/78901
41128 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41129 possibly throwing calls.
41131 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41132 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41133 and fns handling, rather than in a separate case SSA_NAME.
41135 2017-01-02 Jeff Law <law@redhat.com>
41137 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41138 fixes for first_period and second_period variables.
41140 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41143 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41144 (*insvqi_1): New insn pattern.
41145 (*insvqi_1_mem_rex64): Ditto.
41146 (*insvqi_2): Ditto.
41147 (*insvqi_3): Rename from *insvqi.
41149 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41151 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41153 * doc/cfg.texi (Edges): Remove reference to Java.
41154 (Maintaining the CFG): Ditto.
41156 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41158 PR middle-end/77674
41159 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41160 transparent aliases.
41162 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41164 PR middle-end/77484
41165 * predict.def (PRED_CALL): Update hitrate.
41166 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41167 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41168 into direct/indirect/polymorphic variants.
41170 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41172 Update copyright years.
41174 * gcc.c (process_command): Update copyright notice dates.
41175 * gcov-dump.c (print_version): Ditto.
41176 * gcov.c (print_version): Ditto.
41177 * gcov-tool.c (print_version): Ditto.
41178 * gengtype.c (create_file): Ditto.
41179 * doc/cpp.texi: Bump @copying's copyright year.
41180 * doc/cppinternals.texi: Ditto.
41181 * doc/gcc.texi: Ditto.
41182 * doc/gccint.texi: Ditto.
41183 * doc/gcov.texi: Ditto.
41184 * doc/install.texi: Ditto.
41185 * doc/invoke.texi: Ditto.
41187 Copyright (C) 2017 Free Software Foundation, Inc.
41189 Copying and distribution of this file, with or without modification,
41190 are permitted in any medium without royalty provided the copyright
41191 notice and this notice are preserved.