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 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
6 rather than a HOST_WIDE_INT.
7 * tree.c (get_binfo_at_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 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
14 rather than a HOST_WIDE_INT.
15 * tree-sra.c (build_ref_for_offset): Likewise.
17 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
18 Alan Hayward <alan.hayward@arm.com>
19 David Sherwood <david.sherwood@arm.com>
21 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
23 * tree.c (mem_ref_offset): Likewise.
24 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
25 * builtins.c (get_object_alignment_2): Likewise.
26 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
27 * gimple-fold.c (get_base_constructor): Likewise.
28 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
29 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
31 * ipa-polymorphic-call.c
32 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
33 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
34 (get_ancestor_addr_info): Likewise.
35 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
37 * tree-data-ref.c (dr_analyze_innermost): Likewise.
38 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
39 * tree-eh.c (tree_could_trap_p): Likewise.
40 * tree-object-size.c (addr_object_size): Likewise.
41 * tree-ssa-address.c (copy_ref_info): Likewise.
42 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
43 (indirect_refs_may_alias_p): Likewise.
44 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
45 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
46 (non_rewritable_mem_ref_base): Likewise.
47 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
48 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
49 * varasm.c (decode_addr_const): Likewise.
51 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
52 Alan Hayward <alan.hayward@arm.com>
53 David Sherwood <david.sherwood@arm.com>
55 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
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 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
63 as a poly_uint64 rather than an unsigned int.
64 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
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 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
72 (decode_addr_const): Update accordingly.
74 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
75 Alan Hayward <alan.hayward@arm.com>
76 David Sherwood <david.sherwood@arm.com>
78 * tree.h (bit_field_size, bit_field_offset): New functions.
79 * hsa-gen.c (gen_hsa_addr): Use them.
80 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
81 (simplify_vector_constructor): Likewise.
82 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
83 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
84 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
85 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
88 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
89 Alan Hayward <alan.hayward@arm.com>
90 David Sherwood <david.sherwood@arm.com>
92 * expr.h (emit_group_load, emit_group_load_into_temps)
93 (emit_group_store): Take the size as a poly_int64 rather than an int.
94 * expr.c (emit_group_load_1, emit_group_load): Likewise.
95 (emit_group_load_into_temp, emit_group_store): Likewise.
97 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
98 Alan Hayward <alan.hayward@arm.com>
99 David Sherwood <david.sherwood@arm.com>
101 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
102 an unsigned int to a poly_uint64.
103 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
104 sizes as poly_uint64s rather than unsigned ints.
105 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
108 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
109 Alan Hayward <alan.hayward@arm.com>
110 David Sherwood <david.sherwood@arm.com>
112 * lra-constraints.c (emit_inc): Change inc_amount from an int
115 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
116 Alan Hayward <alan.hayward@arm.com>
117 David Sherwood <david.sherwood@arm.com>
119 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
121 (add_stack_var, stack_var_cmp, partition_stack_vars)
122 (dump_stack_var_partition): Update accordingly.
123 (alloc_stack_frame_space): Take the size as a poly_int64 rather
124 than a HOST_WIDE_INT.
125 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
126 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
127 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
128 rather than a HOST_WIDE_INT.
130 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
131 Alan Hayward <alan.hayward@arm.com>
132 David Sherwood <david.sherwood@arm.com>
134 * target.def (return_pops_args): Treat both the input and output
135 sizes as poly_int64s rather than HOST_WIDE_INTS.
136 * targhooks.h (default_return_pops_args): Update accordingly.
137 * targhooks.c (default_return_pops_args): Likewise.
138 * doc/tm.texi: Regenerate.
139 * emit-rtl.h (incoming_args): Change pops_args, size and
140 outgoing_args_size from int to poly_int64_pod.
141 * function.h (expr_status): Change x_pending_stack_adjust and
142 x_stack_pointer_delta from int to poly_int64.
143 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
144 (ARGS_SIZE_RTX): Update accordingly.
145 * calls.c (highest_outgoing_arg_in_use): Change from int to
147 (stack_usage_watermark, stored_args_watermark): New variables.
148 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
149 (emit_call_1): Change the stack_size and rounded_stack_size
150 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
152 (save_fixed_argument_area): Check stack_usage_watermark.
153 (initialize_argument_information): Change old_pending_adj from
154 a HOST_WIDE_INT * to a poly_int64_pod *.
155 (compute_argument_block_size): Return the size as a poly_int64
157 (finalize_must_preallocate): Track polynomial argument sizes.
158 (compute_argument_addresses): Likewise.
159 (internal_arg_pointer_based_exp): Track polynomial offsets.
160 (mem_overlaps_already_clobbered_arg_p): Rename to...
161 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
162 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
163 Check stored_args_used_watermark.
164 (load_register_parameters): Update accordingly.
165 (check_sibcall_argument_overlap_1): Likewise.
166 (combine_pending_stack_adjustment_and_call): Take the unadjusted
167 args size as a poly_int64 rather than an int. Return a bool
168 indicating whether the optimization was possible and return
169 the new adjustment by reference.
170 (check_sibcall_argument_overlap): Track polynomail argument sizes.
171 Update stored_args_watermark.
172 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
173 (expand_call): Likewise. Maintain stack_usage_watermark and
174 stored_args_watermark. Update calls to
175 combine_pending_stack_adjustment_and_call.
176 (emit_library_call_value_1): Handle polynomial argument sizes.
177 Call stack_region_maybe_used_p and mark_stack_region_used.
178 Maintain stack_usage_watermark.
179 (store_one_arg): Likewise. Update call to
180 mem_overlaps_already_clobbered_arg_p.
181 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
183 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
184 * config/microblaze/microblaze.c (microblaze_function_prologue):
186 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
187 TARGET_RETURN_POPS_ARGS interface.
188 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
190 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
191 * config/i386/i386.c (ix86_return_pops_args): Update for new
192 TARGET_RETURN_POPS_ARGS interface.
193 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
194 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
195 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
196 TARGET_RETURN_POPS_ARGS interface.
197 * config/vax/vax.c (vax_return_pops_args): Likewise.
198 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
199 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
200 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
201 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
202 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
203 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
204 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
205 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
206 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
207 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
208 and x_stack_pointer_delta from int to poly_int64.
209 * dojump.c (do_pending_stack_adjust): Update accordingly.
210 * explow.c (allocate_dynamic_stack_space): Handle polynomial
211 stack_pointer_deltas.
212 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
213 (pad_to_arg_alignment): Track polynomial offsets.
214 (assign_parm_find_stack_rtl): Likewise.
215 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
216 * toplev.c (output_stack_usage): Update reference to
217 current_function_pushed_stack_size.
219 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
220 Alan Hayward <alan.hayward@arm.com>
221 David Sherwood <david.sherwood@arm.com>
223 * function.c (in_arg_offset, var_offset, dynamic_offset)
224 (out_arg_offset, cfa_offset): Change from int to poly_int64.
225 (instantiate_new_reg): Return the new offset as a poly_int64_pod
226 rather than a HOST_WIDE_INT.
227 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
228 (instantiate_virtual_regs_in_insn): Likewise.
230 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
231 Alan Hayward <alan.hayward@arm.com>
232 David Sherwood <david.sherwood@arm.com>
234 * rtl.h (get_args_size, add_args_size_note): New functions.
235 (find_args_size_adjust): Return a poly_int64 rather than a
237 (fixup_args_size_notes): Likewise. Make the same change to the
238 end_args_size parameter.
239 * rtlanal.c (get_args_size, add_args_size_note): New functions.
240 * builtins.c (expand_builtin_trap): Use add_args_size_note.
241 * calls.c (emit_call_1): Likewise.
242 * explow.c (adjust_stack_1): Likewise.
243 * cfgcleanup.c (old_insns_match_p): Update use of
244 find_args_size_adjust.
245 * combine.c (distribute_notes): Track polynomial arg sizes.
246 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
247 end_true_args_size, beg_delay_args_size and end_delay_args_size
248 from HOST_WIDE_INT to poly_int64.
249 (add_cfi_args_size): Take the args_size as a poly_int64 rather
250 than a HOST_WIDE_INT.
251 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
252 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
253 polynomial arg sizes.
254 * emit-rtl.c (try_split): Use get_args_size.
255 * recog.c (peep2_attempt): Likewise.
256 * reload1.c (reload_as_needed): Likewise.
257 * expr.c (find_args_size_adjust): Return the adjustment as a
258 poly_int64 rather than a HOST_WIDE_INT.
259 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
260 to a poly_int64 and change the return type in the same way.
261 (emit_single_push_insn): Track polynomial arg sizes.
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 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
268 from HOST_WIDE_INT to poly_int64.
269 * expr.c (push_block, emit_push_insn): Likewise.
271 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
272 Alan Hayward <alan.hayward@arm.com>
273 David Sherwood <david.sherwood@arm.com>
275 * function.h (frame_space): Change start and length from HOST_WIDE_INT
277 (get_frame_size): Return the size as a poly_int64 rather than a
279 (frame_offset_overflow): Take the offset as a poly_int64 rather
280 than a HOST_WIDE_INT.
281 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
282 (assign_stack_temp): Likewise for the size.
283 * function.c (get_frame_size): Return a poly_int64 rather than
285 (frame_offset_overflow): Take the offset as a poly_int64 rather
286 than a HOST_WIDE_INT.
287 (try_fit_stack_local): Take the start, length and size as poly_int64s
288 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
289 rather than a HOST_WIDE_INT.
290 (add_frame_space): Take the start and end as poly_int64s rather than
292 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
293 (assign_stack_temp): Likewise for the size.
294 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
296 (find_temp_slot_from_address): Handle polynomial offsets.
297 (combine_temp_slots): Likewise.
298 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
300 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
301 poly_int64 rather than a HOST_WIDE_INT.
302 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
303 than a HOST_WIDE_INT.
304 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
305 polynomial frame offsets.
306 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
307 as a poly_int64 rather than an int.
308 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
309 * config/v850/v850-protos.h (compute_frame_size): Likewise.
310 * config/v850/v850.c (compute_frame_size): Likewise.
311 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
312 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
313 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
314 * config/pa/pa.c (pa_compute_frame_size): Likewise.
315 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
316 rather than a HOST_WIDE_INT.
317 * explow.c (get_dynamic_stack_base): Likewise.
318 * final.c (final_start_function): Use the constant lower bound
319 of the frame size for -Wframe-larger-than.
320 * ira.c (do_reload): Adjust for new get_frame_size return type.
321 * lra.c (lra): Likewise.
322 * reload1.c (reload): Likewise.
323 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
324 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
325 * rtlanal.c (get_initial_register_offset): Return the offset as
326 a poly_int64 rather than a HOST_WIDE_INT.
328 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
329 Alan Hayward <alan.hayward@arm.com>
330 David Sherwood <david.sherwood@arm.com>
332 * reload1.c (elim_table): Change initial_offset, offset and
333 previous_offset from HOST_WIDE_INT to poly_int64_pod.
334 (offsets_at): Change the target array's element type from
335 HOST_WIDE_INT to poly_int64_pod.
336 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
337 (elimination_costs_in_insn, update_eliminable_offsets)
338 (verify_initial_elim_offsets, set_offsets_for_label)
339 (init_eliminable_invariants): Update after above changes.
341 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
342 Alan Hayward <alan.hayward@arm.com>
343 David Sherwood <david.sherwood@arm.com>
345 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
346 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
347 (decomposition): Change start and end from HOST_WIDE_INT
349 (decompose, immune_p): Update accordingly.
350 (find_inc_amount): Return a poly_int64 rather than an int.
351 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
354 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
355 Alan Hayward <alan.hayward@arm.com>
356 David Sherwood <david.sherwood@arm.com>
358 * tree.h (get_inner_reference): Return the bitsize and bitpos
359 as poly_int64_pods rather than HOST_WIDE_INT.
360 * fold-const.h (ptr_difference_const): Return the pointer difference
361 as a poly_int64_pod rather than a HOST_WIDE_INT.
362 * expr.c (get_inner_reference): Return the bitsize and bitpos
363 as poly_int64_pods rather than HOST_WIDE_INT.
364 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
366 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
367 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
368 (optimize_bit_field_compare): Update call to get_inner_reference.
369 (decode_field_reference): Likewise.
370 (fold_unary_loc): Track polynomial offsets and sizes.
371 (split_address_to_core_and_offset): Return the bitpos as a
372 poly_int64_pod rather than a HOST_WIDE_INT.
373 (ptr_difference_const): Likewise for the pointer difference.
374 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
375 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
376 * dbxout.c (dbxout_expand_expr): Likewise.
377 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
378 (loc_list_from_tree_1, fortran_common): Likewise.
379 * gimple-laddress.c (pass_laddress::execute): Likewise.
380 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
381 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
382 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
383 * tree-affine.c (tree_to_aff_combination): Likewise.
384 (get_inner_reference_aff): Likewise.
385 * tree-data-ref.c (split_constant_offset_1): Likewise.
386 (dr_analyze_innermost): Likewise.
387 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
388 * tree-sra.c (ipa_sra_check_caller): Likewise.
389 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
390 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
391 (instrument_bool_enum_load, instrument_object_size): Likewise.
392 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
393 to get_inner_reference.
394 * hsa-gen.c (gen_hsa_addr): Likewise.
395 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
396 * tsan.c (instrument_expr): Likewise.
397 * match.pd: Update call to ptr_difference_const.
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 * fold-const.c (fold_comparison): Track sizes and offsets as
404 poly_int64s rather than HOST_WIDE_INTs when folding address
407 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
408 Alan Hayward <alan.hayward@arm.com>
409 David Sherwood <david.sherwood@arm.com>
411 * expr.h (get_bit_range): Return the bitstart and bitend as
412 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
413 as a poly_int64 rather than a HOST_WIDE_INT.
414 * expr.c (get_bit_range): Likewise.
415 (expand_assignment): Update call accordingly.
416 * fold-const.c (optimize_bit_field_compare): Likewise.
418 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
419 Alan Hayward <alan.hayward@arm.com>
420 David Sherwood <david.sherwood@arm.com>
422 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
423 (aarch64_legitimate_address_p): Use it instead of an rtx code,
424 as an optional final parameter.
425 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
426 (aarch64_legitimate_address_p): Likewise.
427 (aarch64_print_address_internal): Take an aarch64_addr_query_type
428 instead of an rtx code.
429 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
430 (aarch64_legitimate_address_hook_p): Likewise.
431 (aarch64_print_ldpstp_address): Likewise.
432 (aarch64_print_operand_address): Likewise.
433 (aarch64_address_cost): Likewise.
434 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
435 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
436 (aarch64_mem_pair_lanes_operand): Likewise.
438 2017-12-20 Richard Biener <rguenther@suse.de>
440 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
441 update_stmt_if_modified.
443 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
445 PR tree-optimization/83491
446 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
447 before walking uses. Improve coding style and comments.
449 2017-12-20 Tom de Vries <tom@codesourcery.com>
451 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
453 2017-12-20 Jakub Jelinek <jakub@redhat.com>
456 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
458 (pass_ipa_free_fn_summary): Add small_p private data member,
459 initialize to false in the ctor.
460 (pass_ipa_free_fn_summary::clone,
461 pass_ipa_free_fn_summary::set_pass_param,
462 pass_ipa_free_fn_summary::gate): New methods.
463 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
464 | TODO_dump_symtab if small_p.
465 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
466 entry and add another instance of the pass with false parm after
468 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
469 ipa_free_fn_summary here.
471 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
473 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
475 2017-12-20 Martin Sebor <msebor@redhat.com>
478 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
479 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
480 (expand_builtin_strncmp): Same.
482 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
485 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
486 allowed debug stmts before labels.
487 (expand_gimple_basic_block): Likewise.
488 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
489 * gimple-iterator.h (gsi_after_labels): Likewise.
490 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
491 rename reused variable, and simplify using gsi_move_before.
492 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
493 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
494 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
495 (gimple_verify_flow_info, gimple_block_label): Likewise.
496 (make_blocks): Move debug markers after adjacent labels.
497 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
498 allowed debug insns outside blocks.
499 * df-scan.c (df_insn_delete): Likewise.
500 * lra-constraints.c (update_ebb_live_info): Likewise.
501 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
502 (vt_initialize, delete_vta_debug_insns): Likewise.
503 (reemit_marker_as_note): Drop BB parm. Adjust callers.
505 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
506 Alan Hayward <alan.hayward@arm.com>
507 David Sherwood <david.sherwood@arm.com>
509 * poly-int-types.h (round_down_to_byte_boundary): New macro.
510 (round_up_to_byte_boundary): Likewise.
511 * expr.h (get_bit_range): Add temporary shim.
512 * gimple-ssa-store-merging.c (store_operand_info): Change the
513 bitsize, bitpos, bitregion_start and bitregion_end fields from
514 unsigned HOST_WIDE_INT to poly_uint64.
515 (merged_store_group): Likewise load_align_base.
516 (compatible_load_p, compatible_load_p): Update accordingly.
517 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
518 (split_group, imm_store_chain_info::output_merged_store): Likewise.
519 (mem_valid_for_store_merging): Return the bitsize, bitpos,
520 bitregion_start and bitregion_end as poly_uint64s rather than
521 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
522 (handled_load): Take the bitsize, bitpos,
523 bitregion_start and bitregion_end as poly_uint64s rather than
524 unsigned HOST_WIDE_INTs.
525 (pass_store_merging::process_store): Update call to
526 mem_valid_for_store_merging.
528 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
529 Alan Hayward <alan.hayward@arm.com>
530 David Sherwood <david.sherwood@arm.com>
532 * builtins.c (get_object_alignment_2): Track polynomial offsets
533 and sizes. Update the alignment handling.
535 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
536 Alan Hayward <alan.hayward@arm.com>
537 David Sherwood <david.sherwood@arm.com>
539 * tree.h (get_inner_reference): Add a version that returns the
540 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
541 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
542 the case in which bitpos is not associated with the first byte.
544 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
545 Alan Hayward <alan.hayward@arm.com>
546 David Sherwood <david.sherwood@arm.com>
548 * tree-affine.h (get_inner_reference_aff): Return the size as a
550 * tree-affine.c (get_inner_reference_aff): Likewise.
551 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
552 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
554 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
555 Alan Hayward <alan.hayward@arm.com>
556 David Sherwood <david.sherwood@arm.com>
558 * fold-const.c (pointer_may_wrap_p): Take the offset as a
559 HOST_WIDE_INT rather than a poly_int64.
561 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
562 Alan Hayward <alan.hayward@arm.com>
563 David Sherwood <david.sherwood@arm.com>
565 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
566 HOST_WIDE_INT to poly_int64_pod.
567 (perform_symbolic_merge): Update accordingly.
568 (bswap_replace): Likewise.
570 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
571 Alan Hayward <alan.hayward@arm.com>
572 David Sherwood <david.sherwood@arm.com>
574 * tree-affine.h (aff_tree::offset): Change from widest_int
576 (wide_int_ext_for_comb): Delete.
577 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
578 constants as poly_widest_int rather than widest_int.
579 (aff_combination_constant_multiple_p): Return the multiplier
580 as a poly_widest_int.
581 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
583 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
584 version static and add an overload for poly_widest_int.
585 (aff_combination_const, aff_combination_add_cst)
586 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
587 the constants as poly_widest_int rather than widest_int.
588 (tree_to_aff_combination): Generalize INTEGER_CST case to
590 (aff_combination_to_tree): Track offsets as poly_widest_ints.
591 (aff_combination_add_product, aff_combination_mult): Handle
593 (aff_combination_constant_multiple_p): Return the multiplier
594 as a poly_widest_int.
595 * tree-predcom.c (determine_offset): Return the offset as a
597 (split_data_refs_to_components, suitable_component_p): Update
599 (valid_initializer_p): Update call to
600 aff_combination_constant_multiple_p.
601 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
602 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
603 as a poly_int64 rather than a HOST_WIDE_INT.
604 (get_address_cost): Handle polynomial offsets.
605 (iv_elimination_compare_lt): Likewise.
606 (rewrite_use_nonlinear_expr): Likewise.
608 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
609 Alan Hayward <alan.hayward@arm.com>
610 David Sherwood <david.sherwood@arm.com>
612 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
613 as a poly_int64_pod rather than a HOST_WIDE_INT.
614 (get_addr_base_and_unit_offset): Likewise.
615 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
616 (get_addr_base_and_unit_offset): Likewise.
617 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
618 to poly_int64 in example.
619 * fold-const.c (fold_binary_loc): Update call to
620 get_addr_base_and_unit_offset.
621 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
622 (maybe_canonicalize_mem_ref_addr): Likewise.
623 (gimple_fold_stmt_to_constant_1): Likewise.
624 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
626 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
627 * match.pd: Likewise.
628 * omp-low.c (lower_omp_target): Likewise.
629 * tree-sra.c (build_ref_for_offset): Likewise.
630 (build_debug_ref_for_model): Likewise.
631 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
632 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
633 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
634 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
635 (constant_pointer_difference): Likewise.
636 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
637 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
638 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
639 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
640 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
641 (set_ssa_val_to): Likewise.
642 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
643 (maybe_diag_stxncpy_trunc): Likewise.
644 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
645 * tree.c (build_simple_mem_ref_loc): Likewise.
646 (array_at_struct_end_p): Likewise.
648 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
649 Alan Hayward <alan.hayward@arm.com>
650 David Sherwood <david.sherwood@arm.com>
652 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
653 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
654 (get_ref_base_and_extent_hwi): Declare.
655 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
656 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
657 (get_ref_base_and_extent_hwi): New function.
658 * cfgexpand.c (expand_debug_expr): Update call to
659 get_ref_base_and_extent.
660 * dwarf2out.c (add_var_loc_to_decl): Likewise.
661 * gimple-fold.c (get_base_constructor): Return the offset as a
662 poly_int64_pod rather than a HOST_WIDE_INT.
663 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
664 * ipa-polymorphic-call.c
665 (ipa_polymorphic_call_context::set_by_invariant)
666 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
667 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
668 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
669 rather than get_ref_base_and_extent.
670 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
671 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
672 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
674 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
675 call to get_ref_base_and_extent.
676 * tree-sra.c (create_access, get_access_for_expr): Likewise.
677 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
678 (stmt_kills_ref_p): Likewise.
679 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
680 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
682 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
683 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
684 when calling native_encode_expr.
685 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
686 call to get_ref_base_and_extent.
687 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
688 get_ref_base_and_extent.
689 * var-tracking.c (track_expr_p): Likewise.
691 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
692 Alan Hayward <alan.hayward@arm.com>
693 David Sherwood <david.sherwood@arm.com>
695 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
696 HOST_WIDE_INT to poly_int64_pod.
697 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
698 polynomail parameter offsets.
700 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
701 Alan Hayward <alan.hayward@arm.com>
702 David Sherwood <david.sherwood@arm.com>
704 * gengtype.c (main): Handle poly_int64_pod.
705 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
706 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
707 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
708 from HOST_WIDE_INT to poly_int64_pod.
709 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
710 (copy_cfa): New function.
711 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
712 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
713 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
714 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
715 (initial_return_save): Treat offsets as poly_ints.
716 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
717 offset is nonconstant.
718 (reg_save): Take the offset as a poly_int64. Fall back to
719 DW_CFA_expression for nonconstant offsets.
720 (queue_reg_save): Take the offset as a poly_int64.
721 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
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 * rtl.h (operand_subword, operand_subword_force): Take the offset
728 as a poly_uint64 an unsigned int.
729 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
731 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
732 Alan Hayward <alan.hayward@arm.com>
733 David Sherwood <david.sherwood@arm.com>
735 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
736 'p' format code. Use INT_LIST rather than SUBREG as the example of
737 a code with an XINT and an XEXP. Remove the implication that
738 accessing an rtx field using XINT is expected to work.
739 * rtl.def (SUBREG): Change format from "ei" to "ep".
740 * rtl.h (rtunion::rt_subreg): New field.
741 (XCSUBREG): New macro.
742 (SUBREG_BYTE): Use it.
743 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
744 Update constructor accordingly.
745 (subreg_shape::operator ==): Update accordingly.
746 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
747 than an unsigned int.
748 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
749 a poly_uint64 rather than an unsigned int.
750 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
751 than an unsigned int.
752 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
753 (subreg_size_highpart_offset): Return a poly_uint64 rather than
754 an unsigned int. Take the sizes as poly_uint64s.
755 (subreg_offset_from_lsb): Return a poly_uint64 rather than
756 an unsigned int. Take the shift as a poly_uint64 rather than
758 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
759 as a poly_uint64 rather than an unsigned int.
760 (simplify_subreg_regno): Likewise.
761 (byte_lowpart_offset): Return the memory offset as a poly_int64
763 (subreg_memory_offset): Likewise. Take the subreg offset as a
764 poly_uint64 rather than an unsigned int.
765 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
766 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
767 poly_uint64 rather than an unsigned int.
768 * rtl.c (rtx_format): Describe 'p' in comment.
769 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
770 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
771 offset as a poly_uint64 rather than an unsigned int.
772 (byte_lowpart_offset): Return the memory offset as a poly_int64
774 (subreg_memory_offset): Likewise. Take the subreg offset as a
775 poly_uint64 rather than an unsigned int.
776 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
777 mode sizes as poly_uint64s rather than unsigned ints. Return a
778 poly_uint64 rather than an unsigned int.
779 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
780 (copy_insn_1): Handle 'p'.
781 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
782 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
783 an unsigned int. Return the shift in the same way.
784 (subreg_lsb): Return the shift as a poly_uint64 rather than an
786 (subreg_size_offset_from_lsb): Take the sizes and shift as
787 poly_uint64s rather than unsigned ints. Return the offset as
789 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
790 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
792 * rtlhash.c (add_rtx): Handle 'p'.
793 * genemit.c (gen_exp): Likewise.
794 * gengenrtl.c (type_from_format, gendef): Likewise.
795 * gensupport.c (subst_pattern_match, get_alternatives_number)
796 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
797 (subst_dup): Likewise.
798 * gengtype.c (adjust_field_rtx_def): Likewise.
799 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
800 (match_pattern_2): Likewise.
801 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
802 (rtx_test::subreg_field): New function.
803 (operator ==, safe_to_hoist_p, transition_parameter_type)
804 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
805 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
806 * genpeep.c (match_rtx): Likewise.
807 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
808 (rtx_writer::print_rtx_operand): Handle 'p'.
809 (print_value): Handle SUBREG.
810 * read-rtl.c (apply_int_iterator): Likewise.
811 (rtx_reader::read_rtx_operand): Handle 'p'.
812 * alias.c (rtx_equal_for_memref_p): Likewise.
813 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
814 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
816 * calls.c (expand_call): Likewise.
817 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
818 (make_extraction, gen_lowpart_for_combine): Likewise.
819 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
821 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
822 rather than an unsigned int. Treat subreg offsets as poly_ints.
823 (exp_equiv_p): Handle 'p'.
824 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
825 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
826 * dse.c (find_shift_sequence): Likewise.
827 * dwarf2out.c (rtl_for_decl_location): Likewise.
828 * expmed.c (extract_low_bits): Likewise.
829 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
830 (expand_expr_real_2): Likewise.
831 * final.c (alter_subreg): Likewise.
832 (leaf_renumber_regs_insn): Handle 'p'.
833 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
834 Treat subreg offsets as poly_ints.
835 * fwprop.c (forward_propagate_and_simplify): Likewise.
836 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
837 * ira.c (get_subreg_tracking_sizes): Likewise.
838 * ira-conflicts.c (go_through_subreg): Likewise.
839 * ira-lives.c (process_single_reg_class_operands): Likewise.
840 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
841 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
842 as a poly_uint64 rather than an unsigned int.
843 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
844 subreg offsets as poly_ints.
845 * lra-constraints.c (operands_match_p): Handle 'p'.
846 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
847 * lra-spills.c (assign_mem_slot): Likewise.
848 * postreload.c (move2add_valid_value_p): Likewise.
849 * recog.c (general_operand, indirect_operand): Likewise.
850 * regcprop.c (copy_value, maybe_mode_change): Likewise.
851 (copyprop_hardreg_forward_1): Likewise.
852 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
853 (record_subregs_of_mode): Likewise.
854 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
855 * reload.c (operands_match_p): Handle 'p'.
856 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
857 * reload1.c (alter_reg, choose_reload_regs): Likewise.
858 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
859 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
860 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
861 (simplify_const_poly_int_tests<N>::run): Likewise.
862 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
863 a poly_uint64 rather than an unsigned int.
864 * valtrack.c (debug_lowpart_subreg): Likewise.
865 * var-tracking.c (var_lowpart): Likewise.
866 (loc_cmp): Handle 'p'.
868 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
869 Alan Hayward <alan.hayward@arm.com>
870 David Sherwood <david.sherwood@arm.com>
872 * ira.c (get_subreg_tracking_sizes): New function.
873 (init_live_subregs): Take an integer size rather than a register.
874 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
875 to init_live_subregs.
877 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
878 Alan Hayward <alan.hayward@arm.com>
879 David Sherwood <david.sherwood@arm.com>
881 * expr.c (store_constructor_field): Change bitsize from a
882 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
883 HOST_WIDE_INT to a poly_int64.
884 (store_constructor): Change size from a HOST_WIDE_INT to
886 (store_field): Likewise bitsize and bitpos.
888 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
889 Alan Hayward <alan.hayward@arm.com>
890 David Sherwood <david.sherwood@arm.com>
892 * expmed.h (store_bit_field): Change bitregion_start and
893 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
894 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
895 (store_bit_field_1, store_integral_bit_field, store_bit_field)
896 (store_fixed_bit_field, store_split_bit_field): Likewise.
897 * expr.c (store_constructor_field, store_field): Likewise.
898 (optimize_bitfield_assignment_op): Likewise. Make the same change
899 to bitsize and bitpos.
900 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
901 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
902 same change in the constructor arguments.
903 (get_best_mode): Change bitregion_start and bitregion_end from
904 unsigned HOST_WIDE_INT to poly_uint64.
905 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
906 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
908 (bit_field_mode_iterator::next_mode): Update for new types
909 of m_bitregion_start and m_bitregion_end.
910 (get_best_mode): Change bitregion_start and bitregion_end from
911 unsigned HOST_WIDE_INT to poly_uint64.
913 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
914 Alan Hayward <alan.hayward@arm.com>
915 David Sherwood <david.sherwood@arm.com>
917 * rtl.h (simplify_gen_subreg): Add a temporary overload that
918 accepts poly_uint64 offsets.
919 * expmed.h (extract_bit_field): Take bitsize and bitnum as
920 poly_uint64s rather than unsigned HOST_WIDE_INTs.
921 * expmed.c (lowpart_bit_field_p): Likewise.
922 (extract_bit_field_as_subreg): New function, split out from...
923 (extract_bit_field_1): ...here. Take bitsize and bitnum as
924 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
925 extractions, check that BITSIZE matches the size of the extracted
926 value and that BITNUM is an exact multiple of that size.
927 If all else fails, try forcing the value into memory if
928 BITNUM is variable, and adjusting the address so that the
929 offset is constant. Split the part that can only handle constant
930 bitsize and bitnum out into...
931 (extract_integral_bit_field): ...this new function.
932 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
933 rather than unsigned HOST_WIDE_INTs.
935 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
936 Alan Hayward <alan.hayward@arm.com>
937 David Sherwood <david.sherwood@arm.com>
939 * expmed.h (store_bit_field): Take bitsize and bitnum as
940 poly_uint64s rather than unsigned HOST_WIDE_INTs.
941 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
942 that returns the byte size.
943 (store_bit_field_1): Take bitsize and bitnum as
944 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
945 to simple_mem_bitfield_p. Split the part that can only handle
946 constant bitsize and bitnum out into...
947 (store_integral_bit_field): ...this new function.
948 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
949 than unsigned HOST_WIDE_INTs.
950 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
952 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
953 Alan Hayward <alan.hayward@arm.com>
954 David Sherwood <david.sherwood@arm.com>
956 * lra-int.h (lra_reg): Change offset from int to poly_int64.
957 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
959 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
960 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
961 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
962 offset as a poly_int64 rather than an int.
963 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
964 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
965 * lra-constraints.c (equiv_address_substitution): Track offsets
967 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
968 (curr_insn_transform): Handle the new form of sp_offset.
969 * lra-eliminations.c (lra_elim_table): Change previous_offset
970 and offset from HOST_WIDE_INT to poly_int64.
971 (print_elim_table, update_reg_eliminate): Update accordingly.
972 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
973 (get_elimination): Update accordingly.
974 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
975 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
976 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
977 poly_int64 offsets generally.
978 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
979 (mark_not_eliminable, init_elimination): Update accordingly.
980 (remove_reg_equal_offset_note): Return a bool and pass the new
981 offset back by pointer as a poly_int64.
982 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
983 rather than a HOST_WIDE_INT.
984 (do_remat): Track offsets poly_int64s.
985 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
987 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
988 Alan Hayward <alan.hayward@arm.com>
989 David Sherwood <david.sherwood@arm.com>
991 * rtl.h (mem_attrs): Add a default constructor. Change size and
992 offset from HOST_WIDE_INT to poly_int64.
993 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
994 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
995 (widen_memory_access): Take the sizes and offsets as poly_int64s
996 rather than HOST_WIDE_INTs.
997 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
998 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1000 (adjust_offset_for_component_ref): Change the offset from a
1001 HOST_WIDE_INT to a poly_int64.
1002 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1003 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1004 * dce.c (find_call_stack_args): Likewise.
1005 * dse.c (record_store): Likewise.
1006 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1007 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1008 * read-rtl-function.c (test_loading_mem): Likewise.
1009 * rtlanal.c (may_trap_p_1): Likewise.
1010 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1011 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1012 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1013 (mem_attrs::mem_attrs): New function.
1014 (set_mem_attributes_minus_bitpos): Change bitpos from a
1015 HOST_WIDE_INT to poly_int64.
1016 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1017 (clear_mem_offset, clear_mem_size, change_address)
1018 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1019 initialize mem_attrs.
1020 (set_mem_offset, set_mem_size, adjust_address_1)
1021 (adjust_automodify_address_1, offset_address, widen_memory_access):
1022 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1024 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1025 Alan Hayward <alan.hayward@arm.com>
1026 David Sherwood <david.sherwood@arm.com>
1028 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1029 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1030 rather than 0 to represent an unknown size. Assert that the size
1031 is known when the mode isn't BLKmode.
1032 (may_trap_p_1): Use -1 for unknown sizes.
1033 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1035 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1036 Alan Hayward <alan.hayward@arm.com>
1037 David Sherwood <david.sherwood@arm.com>
1039 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1040 to poly_int64. Update commentary for positions_needed.large.
1041 (read_info_type): Change offset and width from HOST_WIDE_INT
1043 (set_usage_bits): Likewise.
1044 (canon_address): Return the offset as a poly_int64 rather than
1045 a HOST_WIDE_INT. Use strip_offset_and_add.
1046 (set_all_positions_unneeded, any_positions_needed_p): Use
1047 positions_needed.large to track stores with non-constant widths.
1048 (all_positions_needed_p): Likewise. Take the offset and width
1049 as poly_int64s rather than ints. Assert that rhs is nonnull.
1050 (record_store): Cope with non-constant offsets and widths.
1051 Nullify the rhs of an earlier store if we can't tell which bytes
1053 (find_shift_sequence): Take the access_size and shift as poly_int64s
1055 (get_stored_val): Take the read_offset and read_width as poly_int64s
1056 rather than HOST_WIDE_INTs.
1057 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1058 non-constant offsets and widths.
1060 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1061 Alan Hayward <alan.hayward@arm.com>
1062 David Sherwood <david.sherwood@arm.com>
1064 * inchash.h (inchash::hash::add_poly_int): New function.
1065 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1066 Use poly_int64 rather than HOST_WIDE_INT.
1067 (ao_ref::max_size_known_p): New function.
1068 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1069 rather than HOST_WIDE_INT.
1070 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1071 to temporaries until its interface is adjusted to match.
1072 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1073 (aliasing_component_refs_p, decl_refs_may_alias_p)
1074 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1075 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1076 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1078 * alias.c (ao_ref_from_mem): Likewise.
1079 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1080 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1081 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1082 (maybe_trim_complex_store, maybe_trim_constructor_store)
1083 (live_bytes_read, dse_classify_store): Likewise.
1084 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1085 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1086 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1087 (vn_reference_lookup_3): Likewise.
1088 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
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 (indirect_ref_may_alias_decl_p)
1095 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1096 instead of ranges_overlap_p.
1098 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1099 Alan Hayward <alan.hayward@arm.com>
1100 David Sherwood <david.sherwood@arm.com>
1102 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1103 sizes as poly_int64s rather than HOST_WIDE_INTs.
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 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1110 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1111 * gimple-fold.c (fold_ctor_reference): Likewise.
1113 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1114 Alan Hayward <alan.hayward@arm.com>
1115 David Sherwood <david.sherwood@arm.com>
1117 * target.def (dwarf_poly_indeterminate_value): New hook.
1118 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1119 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1120 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1121 * doc/tm.texi: Regenerate.
1122 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1123 offset as a poly_int64.
1124 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1125 offset as a poly_int64.
1126 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1127 Take the offset as a poly_int64.
1128 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1129 (frame_pointer_fb_offset): Change to a poly_int64.
1130 (int_loc_descriptor): Take the offset as a poly_int64. Use
1131 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1132 (based_loc_descr): Take the offset as a poly_int64.
1133 Use strip_offset_and_add to handle (plus X (const)).
1134 Use new_reg_loc_descr instead of an open-coded version of the
1135 previous implementation.
1136 (mem_loc_descriptor): Handle CONST_POLY_INT.
1137 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1138 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1140 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1141 Alan Hayward <alan.hayward@arm.com>
1142 David Sherwood <david.sherwood@arm.com>
1144 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1145 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1146 * inchash.h (inchash::hash::add_poly_hwi): New function.
1147 * gengtype.c (main): Register poly_int64.
1148 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1149 offset as a poly_int.
1150 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1151 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1152 offset as a poly_int64.
1153 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1154 * print-rtl.c (print_poly_int): New function.
1155 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1157 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1159 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1160 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1161 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1162 Enforce track_offset_p for parts of a PARALLEL.
1163 (vt_add_function_parameter): Use const_offset for the final
1164 offset to track. Use get_tracked_reg_offset for the parts
1167 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1168 Alan Hayward <alan.hayward@arm.com>
1169 David Sherwood <david.sherwood@arm.com>
1171 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1172 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1173 * doc/tm.texi: Regenerate.
1174 * hooks.h (hook_bool_uint_uint_true): Delete.
1175 (hook_bool_puint64_puint64_true): Declare.
1176 * hooks.c (hook_bool_uint_uint_true): Delete.
1177 (hook_bool_puint64_puint64_true): New function.
1178 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1179 instead of unsigned ints.
1180 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1181 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1183 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1184 Alan Hayward <alan.hayward@arm.com>
1185 David Sherwood <david.sherwood@arm.com>
1187 * optabs.h (expand_operand): Add an int_value field.
1188 (create_expand_operand): Add an int_value parameter and use it
1189 to initialize the new expand_operand field.
1190 (create_integer_operand): Replace with a declaration of a function
1191 that accepts poly_int64s. Move the implementation to...
1192 * optabs.c (create_integer_operand): ...here.
1193 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1194 the mode preserves the value of int_value, instead of calling
1195 const_int_operand on the rtx. Use gen_int_mode to generate
1198 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1199 Alan Hayward <alan.hayward@arm.com>
1200 David Sherwood <david.sherwood@arm.com>
1202 * dumpfile.h (dump_dec): Declare.
1203 * dumpfile.c (dump_dec): New function.
1204 * pretty-print.h (pp_wide_integer): Turn into a function and
1205 declare a poly_int version.
1206 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1208 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1209 Alan Hayward <alan.hayward@arm.com>
1210 David Sherwood <david.sherwood@arm.com>
1212 * doc/generic.texi (POLY_INT_CST): Document.
1213 * tree.def (POLY_INT_CST): New tree code.
1214 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1215 * tree-core.h (tree_poly_int_cst): New struct.
1216 (tree_node): Add a poly_int_cst field.
1217 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1218 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1219 instead of a wide_int_ref.
1220 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1222 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1223 instead of an unsigned HOST_WIDE_INT.
1224 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1225 (ptrdiff_tree_p): Declare.
1226 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1227 extern inline implementations if the target doesn't use POLY_INT_CST.
1228 (poly_int_tree_p): New function.
1229 (wi::unextended_tree): New class.
1230 (wi::int_traits <unextended_tree>): New override.
1231 (wi::extended_tree): Add a default constructor.
1232 (wi::extended_tree::get_tree): New function.
1233 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1234 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1235 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1236 (wi::tree_to_poly_wide_ref): New typedefs.
1237 (wi::ints_for): Provide overloads for extended_tree and
1239 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1240 (wi::to_wide): New functions.
1241 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1242 * tree.c (poly_int_cst_hasher): New struct.
1243 (poly_int_cst_hash_table): New variable.
1244 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1245 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1247 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1248 (init_ttree): Initialize poly_int_cst_hash_table.
1249 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1250 a poly_int64 instead of a HOST_WIDE_INT.
1251 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1252 instead of an unsigned HOST_WIDE_INT.
1253 (wide_int_to_tree): Rename to...
1254 (wide_int_to_tree_1): ...this.
1255 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1256 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1257 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1258 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1259 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1260 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1262 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1263 (streamer_read_tree_body): Likewise.
1264 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1265 (streamer_write_tree_body): Likewise.
1266 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1267 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1268 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1269 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1270 * gimple-expr.h (is_gimple_constant): Likewise.
1271 * gimplify.c (maybe_with_size_expr): Likewise.
1272 * print-tree.c (print_node): Likewise.
1273 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1274 * tree-pretty-print.c (dump_generic_node): Likewise.
1275 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1276 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1277 * tree-vrp.c (compare_values_warnv): Likewise.
1278 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1279 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1280 (force_expr_to_var_cost): Likewise.
1281 * tree-ssa-loop.c (for_each_index): Likewise.
1282 * fold-const.h (build_invariant_address, size_int_kind): Take a
1283 poly_int64 instead of a HOST_WIDE_INT.
1284 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1285 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1287 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1288 (int_const_binop_2): New function, split out from...
1289 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1290 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1291 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1292 * gimple-ssa-strength-reduction.c (slsr_process_add)
1293 (slsr_process_mul): Check for INTEGER_CSTs before using them
1295 * stor-layout.c (bits_from_bytes): New function.
1296 (bit_from_pos): Use it.
1297 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1298 by BITS_PER_UNIT to get the TYPE_SIZE.
1299 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1300 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1302 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1303 Alan Hayward <alan.hayward@arm.com>
1304 David Sherwood <david.sherwood@arm.com>
1306 * doc/rtl.texi (const_poly_int): Document. Also document the
1307 rtl sharing behavior.
1308 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1309 * rtl.h (const_poly_int_def): New struct.
1310 (rtx_def::u): Add a cpi field.
1311 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1312 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1313 (wi::rtx_to_poly_wide_ref): New typedef
1314 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1315 (poly_int_rtx_p): New functions.
1316 (trunc_int_for_mode): Declare a poly_int64 version.
1317 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1318 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1320 (strip_offset): Declare.
1321 (strip_offset_and_add): New function.
1322 * rtl.def (CONST_POLY_INT): New rtx code.
1323 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1324 (shared_const_p): Use poly_int_rtx_p.
1325 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1327 (gen_int_shift_amount): Likewise.
1328 * emit-rtl.c (const_poly_int_hasher): New class.
1329 (const_poly_int_htab): New variable.
1330 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1331 (const_poly_int_hasher::hash): New function.
1332 (const_poly_int_hasher::equal): Likewise.
1333 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1334 (immed_wide_int_const): Rename to...
1335 (immed_wide_int_const_1): ...this and make static.
1336 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1337 instead of a wide_int_ref.
1338 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1339 (gen_lowpart_common): Handle CONST_POLY_INT.
1340 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1341 * cselib.c (cselib_hash_rtx): Likewise.
1342 * dwarf2out.c (const_ok_for_output_1): Likewise.
1343 * expr.c (convert_modes): Likewise.
1344 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1345 * rtlhash.c (add_rtx): Likewise.
1346 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1347 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1348 Handle existing CONST_POLY_INT rtxes.
1349 * expmed.h (expand_shift): Take a poly_int64 instead of a
1351 * expmed.c (expand_shift): Likewise.
1352 * rtlanal.c (strip_offset): New function.
1353 (commutative_operand_precedence): Give CONST_POLY_INT the same
1354 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1356 * rtl-tests.c (const_poly_int_tests): New struct.
1357 (rtl_tests_c_tests): Use it.
1358 * simplify-rtx.c (simplify_const_unary_operation): Handle
1360 (simplify_const_binary_operation): Likewise.
1361 (simplify_binary_operation_1): Fold additions of symbolic constants
1362 and CONST_POLY_INTs.
1363 (simplify_subreg): Handle extensions and truncations of
1365 (simplify_const_poly_int_tests): New struct.
1366 (simplify_rtx_c_tests): Use it.
1367 * wide-int.h (storage_ref): Add default constructor.
1368 (wide_int_ref_storage): Likewise.
1369 (trailing_wide_ints): Use GTY((user)).
1370 (trailing_wide_ints::operator[]): Add a const version.
1371 (trailing_wide_ints::get_precision): New function.
1372 (trailing_wide_ints::extra_size): Likewise.
1374 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1375 Alan Hayward <alan.hayward@arm.com>
1376 David Sherwood <david.sherwood@arm.com>
1378 * emit-rtl.h (gen_int_shift_amount): Declare.
1379 * emit-rtl.c (gen_int_shift_amount): New function.
1380 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1382 * calls.c (shift_return_value): Likewise.
1383 * cse.c (fold_rtx): Likewise.
1384 * dse.c (find_shift_sequence): Likewise.
1385 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1386 (expand_shift, expand_smod_pow2): Likewise.
1387 * lower-subreg.c (shift_cost): Likewise.
1388 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1389 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1390 (expand_vec_perm_var): Likewise.
1391 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1392 (simplify_binary_operation_1): Likewise.
1393 * combine.c (try_combine, find_split_point, force_int_to_mode)
1394 (simplify_shift_const_1, simplify_shift_const): Likewise.
1395 (change_zero_ext): Likewise. Use simplify_gen_binary.
1397 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1399 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1401 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1403 * doc/invoke.texi (ARM Options): Document accepted extension options
1404 for -march=armv8.3-a.
1406 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1409 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1410 or --with-float=softfp, set the default CPU to arm10e.
1412 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1414 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1415 * config/visium/predicates.md (const_shift_operand): Likewise.
1416 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1417 (visium_legitimize_reload_address): Likewise.
1419 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1422 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1424 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1427 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1429 2017-12-20 Julia Koval <julia.koval@intel.com>
1431 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1432 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1433 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1434 * config.gcc: Include vpclmulqdqintrin.h.
1435 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1436 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1438 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1439 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1440 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1441 * config/i386/i386.c (isa2_opts): Add -mcx16.
1442 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1443 (ix86_option_override_internal): Move mcx16 to flags2.
1444 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1445 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1446 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1447 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1448 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1449 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1450 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1451 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1452 * doc/invoke.texi: Add -mvpclmulqdq.
1454 2017-12-20 Tom de Vries <tom@codesourcery.com>
1457 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
1458 * calls.c (rtx_for_static_chain): ... here. New function.
1459 * calls.h (rtx_for_static_chain): Declare.
1460 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1461 instead of targetm.calls.static_chain.
1462 * df-scan.c (df_get_entry_block_def_set): Same.
1464 2017-12-19 Tom de Vries <tom@codesourcery.com>
1466 PR tree-optimization/83493
1467 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1470 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1472 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1473 inform with hardcoded english plural handling.
1475 2017-12-18 Jeff Law <law@redhat.com>
1477 PR tree-optimization/83477
1478 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1479 a non-virtual PHI, always push a new range.
1481 2017-12-19 Martin Sebor <msebor@redhat.com>
1484 * builtins.c (compute_objsize): Handle non-constant offsets.
1486 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1488 PR tree-optimization/83444
1489 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1490 character load case, if get_stridx on MEM_REF's operand doesn't
1491 look usable, retry with get_addr_stridx.
1493 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1496 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1497 (variable_tracking_main_1): Keep markers even when VTA fails.
1500 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1501 even if there are markers before it.
1502 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1504 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1506 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1507 typical order conditions.
1508 * sel-sched.c (extract_new_fences_from): Likewise.
1509 * config/visium/constraints.md (J, K, L): Likewise.
1510 * config/visium/predicates.md (const_shift_operand): Likewise.
1511 * config/visium/visium.c (visium_legitimize_address,
1512 visium_legitimize_reload_address): Likewise.
1513 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1514 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1515 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1516 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1517 SET_DUMP_DETAIL): Likewise.
1518 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1519 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1520 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1521 avr_set_core_architecture, avr_set_current_function,
1522 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1523 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1524 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1525 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1526 avr_map_decompose, avr_fold_builtin): Likewise.
1527 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1528 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1529 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1530 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1531 * config/m32c/m32c.c (m32c_conditional_register_usage,
1532 m32c_address_cost): Likewise.
1533 * config/m32c/predicates.md (shiftcount_operand,
1534 longshiftcount_operand): Likewise.
1535 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1536 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1537 can_use_cdx_ldstw): Likewise.
1538 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1539 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1541 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1542 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1543 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1544 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1545 * config/vax/vax.c (adjacent_operands_p): Likewise.
1546 * config/ft32/constraints.md (L, b, KA): Likewise.
1547 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1549 * cfgexpand.c (expand_stack_alignment): Likewise.
1550 * gcse.c (insert_expr_in_table): Likewise.
1551 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1552 * cgraphunit.c (cgraph_node::expand): Likewise.
1553 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1554 * emit-rtl.c (add_insn): Likewise.
1555 * input.c (dump_location_info): Likewise.
1556 * passes.c (NEXT_PASS): Likewise.
1557 * read-rtl-function.c (parse_note_insn_name,
1558 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1560 * sched-rgn.c (sched_rgn_init): Likewise.
1561 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1562 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1563 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
1564 * explow.c (eliminate_constant_term): Likewise.
1565 * final.c (leaf_renumber_regs_insn): Likewise.
1566 * cfgrtl.c (print_rtl_with_bb): Likewise.
1567 * genhooks.c (emit_init_macros): Likewise.
1568 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1569 * tree-data-ref.c (conflict_fn): Likewise.
1570 * selftest.c (assert_streq): Likewise.
1571 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1572 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1573 fold_binary_loc, multiple_of_p): Likewise.
1574 * reload.c (push_reload, find_equiv_reg): Likewise.
1575 * et-forest.c (et_nca, et_below): Likewise.
1576 * dbxout.c (dbxout_symbol_location): Likewise.
1577 * reorg.c (relax_delay_slots): Likewise.
1578 * dojump.c (do_compare_rtx_and_jump): Likewise.
1579 * gengtype-parse.c (type): Likewise.
1580 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1581 simplify_const_relational_operation): Likewise.
1582 * reload1.c (do_output_reload): Likewise.
1583 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1584 * gengtype.c (type_for_name): Likewise.
1585 * gimple-ssa-sprintf.c (format_directive): Likewise.
1587 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1590 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1591 accessing it. Adjust comment.
1593 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1596 * predict.c (zero_one_minusone): New function.
1597 (apply_return_prediction): Avoid return prediction for functions
1598 returning only -1, 0 and 1 values, unless they only return -1 and 0
1601 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1603 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1604 fall-through warning.
1605 (arc_compute_frame_size): Remove unused variables.
1606 (arc_print_operand): Fix fprintif format.
1607 (arc_can_follow_jump): Clean fall-through warning.
1609 2017-12-19 Marek Polacek <polacek@redhat.com>
1612 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1615 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1617 * config/arc/arc.c (overriderregs): New variable.
1618 (arc_override_options): Track fixed/call saved/call options.
1619 (arc_conditional_register_usage): Check against overrideregs
1620 variable whenever we change register properties.
1622 2017-12-19 Nathan Sidwell <nathan@acm.org>
1624 * opts.c (finish_options): Don't prefix dump_base_name if it
1625 already contains directories.
1627 2017-12-19 Martin Liska <mliska@suse.cz>
1629 PR rtl-optimization/82675
1630 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1631 more element in sbitmap.
1633 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1635 * gimplify.c (gimplify_expr): Use error_operand_p.
1637 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1640 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1643 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1645 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1646 the jump is to a label.
1648 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1650 PR tree-optimization/83444
1651 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1656 * ipa-inline.c (flatten_remove_node_hook): New function.
1657 (ipa_inline): Keep only nodes with flatten attribute at the end of
1658 the array in the order from ipa_reverse_postorder, only walk that
1659 portion of array for flattening, if there is more than one such
1660 node, temporarily register a removal hook and ignore removed nodes.
1662 PR tree-optimization/80631
1663 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1664 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1665 IFN_REDUC_MAX or IFN_REDUC_MIN.
1667 2017-12-18 Jeff Law <law@redhat.com>
1669 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1670 record symbolic equivalences from backedges in the CFG.
1673 2017-11-19 Jeff Law <law@redhat.com>
1675 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1676 of degenerates resulting from ignoring an edge.
1678 2017-12-18 Martin Sebor <msebor@redhat.com>
1681 PR tree-optimization/78450
1682 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1683 (handle_builtin_strlen): Call it.
1685 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1687 PR rtl-optimization/83424
1688 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1690 2017-12-18 Marek Polacek <polacek@redhat.com>
1693 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1694 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1697 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1699 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1700 (umaddsidi4, umaddsidi_split): Likewise.
1702 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1704 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1707 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1710 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1712 2017-12-18 Richard Biener <rguenther@suse.de>
1714 PR tree-optimization/81877
1715 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1716 (outermost_indep_loop): Adjust.
1717 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1718 (can_sm_ref_p): Adjust.
1720 2017-12-18 Richard Biener <rguenther@suse.de>
1723 * tree.c (array_at_struct_end_p): Return true if the underlying
1724 object has space for at least one element in excess of what
1725 the array domain specifies.
1727 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1729 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1730 example to avoid overfull hbox.
1731 * doc/invoke.texi (Option Summary): Add missing @gol.
1732 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1735 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1736 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1738 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1739 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1740 (AArch64 Options): Document -mverbose-cost-dump.
1741 (ARM Options): Likewise, plus -mflip-thumb.
1743 2017-12-17 Martin Sebor <msebor@redhat.com>
1746 * gimple-ssa-warn-restrict.c
1747 (builtin_memref::offset_out_of_bounds): Correct the handling of
1750 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1752 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1754 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1756 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1758 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1760 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1763 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1765 * doc/invoke.texi: Fix some typos.
1767 2017-12-16 Martin Sebor <msebor@redhat.com>
1769 PR tree-optimization/78918
1770 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1771 * builtins.c (check_sizes): Rename...
1772 (check_access): ...to this. Rename function arguments for clarity.
1773 (check_memop_sizes): Adjust names.
1774 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1775 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1776 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1777 (check_strncat_sizes, expand_builtin_strncat): Same.
1778 (expand_builtin_strncpy, expand_builtin_memset): Same.
1779 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1780 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1781 (maybe_emit_sprintf_chk_warning): Same.
1782 (expand_builtin_strcpy): Adjust.
1783 (expand_builtin_stpcpy): Same.
1784 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1785 in pointer-checking forms of memcpy, memmove, and mempcpy.
1786 (gcall_to_tree_minimal, max_object_size): Define new functions.
1787 * builtins.h (max_object_size): Declare.
1788 * calls.c (alloc_max_size): Call max_object_size instead of
1789 hardcoding ssizetype limit.
1790 (get_size_range): Handle new argument.
1791 * calls.h (get_size_range): Add a new argument.
1792 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1793 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1794 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1796 (gimple_fold_builtin_memory_chk): Same.
1797 (gimple_fold_builtin_stxcpy_chk): New function.
1798 * gimple-ssa-warn-restrict.c: New source.
1799 * gimple-ssa-warn-restrict.h: New header.
1800 * gimple.c (gimple_build_call_from_tree): Propagate location.
1801 * passes.def (pass_warn_restrict): Add new pass.
1802 * tree-pass.h (make_pass_warn_restrict): Declare.
1803 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1805 (handle_builtin_strcat): Same.
1806 (strlen_optimize_stmt): Rename...
1807 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1808 stpncpy, strncpy, and their checking forms.
1810 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1812 PR rtl-optimization/82849
1813 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1814 and get_max_loop_iterations_int.
1816 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1817 Alan Hayward <alan.hayward@arm.com>
1818 David Sherwood <david.sherwood@arm.com>
1820 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1821 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1822 size as a poly_uint64.
1823 (mode_for_vector, mode_for_int_vector): Take the number of vector
1824 elements as a poly_uint64.
1825 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1826 size as a poly_uint64.
1827 (mode_for_vector, mode_for_int_vector): Take the number of vector
1828 elements as a poly_uint64.
1830 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1831 Alan Hayward <alan.hayward@arm.com>
1832 David Sherwood <david.sherwood@arm.com>
1834 * machmode.h (MACRO_MODE): New macro.
1835 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1836 * targhooks.c (default_libcall_value, default_secondary_reload)
1837 (default_memory_move_cost, default_register_move_cost)
1838 (default_class_max_nregs): Likewise.
1840 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1841 Alan Hayward <alan.hayward@arm.com>
1842 David Sherwood <david.sherwood@arm.com>
1844 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1845 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1847 * genautomata.c (main): Likewise.
1848 * genconditions.c (write_header): Likewise.
1849 * genemit.c (main): Likewise.
1850 * genextract.c (print_header): Likewise.
1851 * genopinit.c (main): Likewise.
1852 * genoutput.c (output_prologue): Likewise.
1853 * genpeep.c (main): Likewise.
1854 * genpreds.c (write_insn_preds_c): Likewise.
1855 * genrecog.c (writer_header): Likewise.
1856 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1857 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1858 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
1859 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
1860 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
1861 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
1862 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
1863 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
1864 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
1865 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
1866 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
1867 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
1868 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
1869 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
1870 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
1871 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
1872 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
1873 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
1874 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
1875 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
1876 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
1877 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
1878 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
1879 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
1880 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
1881 * config/darwin.c (IN_TARGET_CODE): Likewise.
1882 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
1883 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
1884 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
1885 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
1886 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
1887 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
1888 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
1889 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
1890 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
1891 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
1892 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
1893 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
1894 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
1895 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
1896 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
1897 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
1898 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
1899 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
1900 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
1901 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
1902 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
1903 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
1904 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
1905 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
1906 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
1907 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
1908 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
1909 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
1910 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
1911 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
1912 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
1913 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
1914 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
1915 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
1916 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
1917 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
1918 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
1919 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
1920 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
1921 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
1922 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
1923 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
1924 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
1925 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
1926 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
1927 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
1928 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
1929 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
1930 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
1931 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
1932 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
1933 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
1934 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
1935 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
1936 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
1937 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
1938 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
1939 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
1940 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
1941 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
1942 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1943 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
1944 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
1945 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
1946 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
1947 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
1948 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
1949 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
1950 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
1951 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
1952 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
1953 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1954 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
1955 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
1956 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
1957 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
1958 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
1959 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
1960 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
1961 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
1962 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
1963 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
1964 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
1965 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
1966 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
1967 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
1968 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
1969 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
1970 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
1971 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
1972 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
1973 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
1974 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
1975 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
1976 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
1977 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
1978 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
1979 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
1980 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
1981 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
1982 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
1983 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
1984 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
1985 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
1986 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
1987 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
1989 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1990 Alan Hayward <alan.hayward@arm.com>
1991 David Sherwood <david.sherwood@arm.com>
1993 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
1996 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1997 Alan Hayward <alan.hayward@arm.com>
1998 David Sherwood <david.sherwood@arm.com>
2000 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2001 * doc/md.texi (vec_series@var{m}): Document.
2002 * tree.def (VEC_SERIES_EXPR): New tree code.
2003 * tree.h (build_vec_series): Declare.
2004 * tree.c (build_vec_series): New function.
2005 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2006 * tree-pretty-print.c (dump_generic_node): Likewise.
2007 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2008 * tree-inline.c (estimate_operator_cost): Likewise.
2009 * expr.c (expand_expr_real_2): Likewise.
2010 * optabs-tree.c (optab_for_tree_code): Likewise.
2011 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2012 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2013 * expmed.c (make_tree): Handle VEC_SERIES.
2014 * optabs.def (vec_series_optab): New optab.
2015 * optabs.h (expand_vec_series_expr): Declare.
2016 * optabs.c (expand_vec_series_expr): New function.
2017 * tree-vect-generic.c (expand_vector_operations_1): Check that
2018 the operands also have vector type.
2020 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2021 Alan Hayward <alan.hawyard@arm.com>
2022 David Sherwood <david.sherwood@arm.com>
2024 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2025 (VEC_COND_EXPR): Add missing @tindex.
2026 * doc/md.texi (vec_duplicate@var{m}): Document.
2027 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2028 * tree.c (build_vector_from_val): Add stubbed-out handling of
2029 variable-length vectors, using VEC_DUPLICATE_EXPR.
2030 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2031 * cfgexpand.c (expand_debug_expr): Likewise.
2032 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2033 * tree-inline.c (estimate_operator_cost): Likewise.
2034 * tree-pretty-print.c (dump_generic_node): Likewise.
2035 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2036 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2037 (test_vec_duplicate_folding): New function.
2038 (fold_const_c_tests): Call it.
2039 * optabs.def (vec_duplicate_optab): New optab.
2040 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2041 * optabs.h (expand_vector_broadcast): Declare.
2042 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2043 vec_duplicate_optab.
2044 * expr.c (store_constructor): Try using vec_duplicate_optab for
2046 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2048 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2051 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2052 div/mod latencies a bit.
2054 2017-12-15 Jeff Law <law@redhat.com>
2056 PR tree-optimization/36550
2057 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2058 (mark_threaded_blocks): Rewrite code to avoid block copying when
2059 optimizing for size. Don't pessimize blocks which will be
2060 copied, but all the statements will be dead.
2062 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2064 PR tree-optimization/81165
2065 * tree-ssa-threadupdate.c (uses_in_bb): New.
2066 (estimate_threading_killed_stmts): New.
2067 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2068 * tree-ssa-threadedge.c
2069 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2072 2017-12-15 Jeff Law <law@redhat.com>
2074 PR tree-optimization/83410
2075 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2076 threads when parallelizing loops.
2078 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2080 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2082 * config/alpha/alpha.c (vms_attribute_table): Swap
2083 affects_type_identity and handler fields, adjust comments.
2084 * config/mips/mips.c (mips_attribute_table): Likewise.
2085 * config/visium/visium.c (visium_attribute_table): Likewise.
2086 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2087 * config/microblaze/microblaze.c (microblaze_attribute_table):
2089 * config/spu/spu.c (spu_attribute_table): Likewise.
2090 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2091 * config/arc/arc.c (arc_attribute_table): Likewise.
2092 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2093 * config/v850/v850.c (v850_handle_interrupt_attribute,
2094 v850_handle_data_area_attribute): Formatting fixes.
2095 (v850_attribute_table): Swap affects_type_identity and handler
2096 fields, adjust comments.
2097 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2098 * config/arm/arm.c (arm_attribute_table): Likewise.
2099 * config/avr/avr.c (avr_attribute_table): Likewise.
2100 * config/s390/s390.c (s390_attribute_table): Likewise.
2101 * config/sh/sh.c (sh_attribute_table): Likewise.
2102 * config/i386/i386.c (ix86_handle_cconv_attribute,
2103 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2104 (ix86_attribute_table): Swap affects_type_identity and handler
2105 fields, adjust comments.
2106 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2107 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2108 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2109 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2110 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2111 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2112 * config/rx/rx.c (rx_attribute_table): Likewise.
2113 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2114 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2115 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2116 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2117 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2118 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2119 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2120 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2121 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2122 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2123 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2125 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2127 (register_scoped_attributes, decl_attributes): Formatting fixes.
2129 PR tree-optimization/83269
2130 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2131 subtraction in arg0's type if type is signed and arg0 is unsigned.
2135 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2136 convert? on @0 instead of convert. Check type of @1, not @0.
2137 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2138 convert? on @0 instead of convert. Check type of @1, not @0.
2139 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2140 only optimize if either both @1 and @2 types are narrower
2141 precision, or both are wider or equal precision, and in the former
2142 case only if both have undefined overflow.
2144 2017-12-15 Richard Biener <rguenther@suse.de>
2147 * internal-fn.def (IFN_NOP): Add.
2148 * internal-fn.c (expand_NOP): Do nothing.
2149 * lto-streamer-in.c (input_function): Instead of removing
2150 sanitizer calls replace them with IFN_NOP calls.
2152 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2153 Alan Hayward <alan.hayward@arm.com>
2154 David Sherwood <david.sherwood@arm.com>
2156 * dse.c (store_info, read_info_type): Replace begin and end with
2158 (print_range): New function.
2159 (set_all_positions_unneeded, any_positions_needed_p)
2160 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2162 (record_store): Likewise. Optimize the case in which all positions
2164 (get_stored_val): Replace read_begin and read_end with read_offset
2166 (replace_read): Update call accordingly.
2168 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2170 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2171 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2172 (loop_cand::loop_cand): Initialize above members.
2173 (loop_cand::supported_operations): Delete.
2174 (loop_cand::can_interchange_p): Inline above function.
2175 (loop_cand::classify_simple_reduction): Record number of constant
2176 initialized simple reductions.
2177 (should_interchange_loops): New parameters. Check stmt cost of loops
2179 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2180 Update call to should_interchange_loops.
2181 (should_interchange_loop_nest): Update call to
2182 should_interchange_loops.
2184 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2187 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2188 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2189 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2191 2017-12-15 Julia Koval <julia.koval@intel.com>
2193 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2194 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
2195 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2196 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2197 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2199 2017-12-15 Julia Koval <julia.koval@intel.com>
2201 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2202 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2203 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2204 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2205 _mm_aesenc_epi128): New intrinsics.
2207 2017-12-15 Julia Koval <julia.koval@intel.com>
2209 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2210 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
2211 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2212 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2213 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2215 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2217 * gimple-ssa-strength-reduction.c (analyze_increments):
2218 Distinguish replacement costs for constant strides from those for
2221 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2223 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2225 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2227 * doc/invoke.texi: Document -Wcast-function-type.
2228 * recog.h (stored_funcptr): Change signature.
2229 * tree-dump.c (dump_node): Avoid warning.
2230 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2232 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2235 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2238 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2240 PR tree-optimization/83312
2241 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2242 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2243 "val" param, and to cope with arbitrary basic blocks.
2244 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2245 handle NULL_TREE for "val", dropping "bb" param.
2246 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2247 drop "bb" param. Handle NULL_TREE for "val".
2248 (find_case_label_for_value): Make "switch_stmt" param const.
2249 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2251 (vrp_prop::check_all_array_refs): Reimplement as...
2252 (check_array_bounds_dom_walker::before_dom_children): ...this new
2253 vfunc. Replace linear search through BB block list, excluding
2254 those with non-executable in-edges via dominator walk.
2256 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2258 * config/arm/arm.opt (mverbose-cost-dump): New option.
2259 * config/arm/arm.c (arm_rtx_costs): Use it.
2261 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2264 * reload1.c (emit_input_reload_insns): Skip debug markers.
2266 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2268 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2272 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2273 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2274 * config/sh/sh.c: Skip debug insns besides notes.
2275 * config/sh/sh.md: Likewise.
2276 * config/sh/sh_treg_combine.cc: Likewise.
2277 * config/sh/sync.md: Likewise.
2279 2017-12-14 Tom de Vries <tom@codesourcery.com>
2281 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2282 Add item for weak_undefined.
2284 2017-12-14 Richard Biener <rguenther@suse.de>
2286 PR tree-optimization/67842
2287 * sese.h (bb_in_region): Remove #if 0'ed code.
2289 2017-12-14 Richard Biener <rguenther@suse.de>
2291 PR tree-optimization/83326
2292 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2293 may_be_zero parameter and handle it by not marking the first
2294 peeled copy as not exiting the loop.
2295 (try_peel_loop): Likewise.
2296 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2297 to handle the case of constant or zero iterations and perform
2298 loop header copying on-the-fly.
2300 2017-12-14 Richard Biener <rguenther@suse.de>
2302 PR tree-optimization/83418
2303 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2304 Instead of asserting we don't get unfolded comparisons deal with
2307 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2310 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2311 basic blocks. Assert debug bind insns don't appear outside of bbs,
2312 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2315 PR tree-optimization/83198
2316 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2317 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2318 value if arg is a REAL_CST with incompatible type.
2320 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2321 Bin Cheng <bin.cheng@arm.com>
2324 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2326 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2329 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2331 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2332 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2334 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2335 Alan Hayward <alan.hayward@arm.com>
2336 David Sherwood <david.sherwood@arm.com>
2338 * poly-int.h: New file.
2339 * poly-int-types.h: Likewise.
2340 * coretypes.h: Include them.
2341 (POLY_INT_CONVERSION): Define.
2342 * target.def (estimated_poly_value): New hook.
2343 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2344 * doc/tm.texi: Regenerate.
2345 * doc/poly-int.texi: New file.
2346 * doc/gccint.texi: Include it.
2347 * doc/rtl.texi: Describe restrictions on subreg modes.
2348 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2349 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2350 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2351 * targhooks.h (default_estimated_poly_value): Declare.
2352 * targhooks.c (default_estimated_poly_value): New function.
2353 * target.h (estimated_poly_value): Likewise.
2354 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2355 (wi::unary_traits): Delete.
2356 (wi::binary_traits::signed_shift_result_type): Define for
2357 offset_int << HOST_WIDE_INT, etc.
2358 (generic_wide_int::operator <<=): Define for all types and use
2359 wi::lshift instead of <<.
2360 (wi::hwi_with_prec): Add a default constructor.
2361 (wi::ints_for): New class.
2362 (operator <<): Define for all wide-int types.
2363 (operator /): New function.
2364 (operator %): Likewise.
2365 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2366 (ASSERT_MAYBE_NE_AT): New macros.
2368 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2369 Dominik Vogt <vogt@linux.vnet.ibm.com>
2372 * emit-rtl.c (init_emit): Remove ??? comment.
2373 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2374 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2375 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2376 alignment of 3 virtual registers to BITS_PER_WORD.
2378 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2380 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2382 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2383 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2385 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2388 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2390 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2392 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2393 with the exception of debug begin stmt markers appear before
2397 * final.c (rest_of_handle_final): Call variable_tracking_main only
2398 if !flag_var_tracking.
2400 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2401 Jakub Jelinek <jakub@redhat.com>
2405 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2406 labels debug stmts that can only appear after labels.
2408 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2410 PR rtl-optimization/82398
2411 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2412 EXPR_USEFULNESS in priority comparison.
2414 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2416 PR rtl-optimization/83393
2417 * combine.c (move_deaths): If reg_stat points to a too new insn in
2418 last_death, do not use it: find the proper insn instead.
2420 2017-12-12 Jeff Law <law@redhat.com>
2422 PR tree-optimization/83298
2423 PR tree-optimization/83362
2424 PR tree-optimization/83383
2425 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2426 push_value_range a public interface. Add new argument to
2427 record_ranges_from_stmt.
2428 * gimple-ssa-evrp-analyze.c
2429 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2430 Update comments. Handle recording temporary equivalences.
2431 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2432 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2433 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2434 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2435 gimple-ssa-evrp-analyze.h.
2436 (record_temporary_equivalences_from_phis): Add new argument. When
2437 the PHI arg is an SSA_NAME, set the result's range to the range
2439 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2440 from statements too.
2441 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2442 Pass it down to children as needed.
2443 (thread_outgoing_edges): Likewise.
2444 (thread_across_edge): Likewise. Push/pop range state as needed.
2445 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2447 2017-12-12 Julia Koval <julia.koval@intel.com>
2449 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2450 (PTA_CANNONLAKE): Remove PTA_CLWB.
2452 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2455 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2456 (vcondv2div2df): Likewise.
2457 (vconduv2dfv2di): Likewise.
2459 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2461 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2462 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2463 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2464 ATTR_MATHFN_FPROUNDING_ERRNO.
2466 2017-12-12 Richard Biener <rguenther@suse.de>
2468 PR tree-optimization/83385
2469 * graphite-scop-detection.c (get_order, order): Remove.
2470 (bb_to_rpo): New global.
2472 (build_scops): Sort pbbs in RPO order.
2474 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2476 * combine.c (simplify_set): Do not transform subregs to zero_extends
2477 if the destination mode is a vector mode.
2479 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2481 PR tree-optimization/83359
2482 * tree-cfg.h (fold_loop_internal_call): Declare.
2483 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2484 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2485 (find_loop_dist_alias): New function.
2486 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2487 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2490 PR tree-optimization/80631
2491 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2492 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2493 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2494 hardcoding zero as the value if COND_EXPR is never true. For
2495 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2496 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2497 hardcoding MAX_EXPR as the reduction operation.
2498 (is_nonwrapping_integer_induction): Allow negative step.
2499 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2500 vect_create_epilog_for_reduction, if no value is suitable, don't
2501 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2503 2017-12-12 Richard Biener <rguenther@suse.de>
2505 PR tree-optimization/81889
2506 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2507 range info from the non-wrapping IV instead of just the range
2510 2017-12-12 Julia Koval <julia.koval@intel.com>
2512 * config.gcc: Add vaesintrin.h.
2513 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2515 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2516 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2518 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2519 * config/i386/immintrin.h: Include vaesintrin.h.
2520 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2521 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2522 _mm_aesdec_epi128): New intrinsics.
2524 2017-12-12 Julia Koval <julia.koval@intel.com>
2526 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2527 OPTION_MASK_ISA_VAES_UNSET): New.
2528 (ix86_handle_option): Handle -mvaes.
2529 * config/i386/cpuid.h: Define bit_VAES.
2530 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2531 * config/i386/i386-c.c (__VAES__): New.
2532 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2533 (ix86_valid_target_attribute_inner_p): Ditto.
2534 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2535 * config/i386/i386.opt: Add -mvaes.
2536 * doc/invoke.texi: Ditto.
2538 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2540 * debug.h (gcc_debug_hooks): Add inline_entry.
2541 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2542 * debug.c (do_nothing_debug_hooks): Likewise.
2543 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2544 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2545 (dwarf2_lineno_debug_hooks): Likewise.
2547 * common.opt (gstatement-frontiers): New, setting
2548 debug_nonbind_markers_p.
2549 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2550 * toplev.c (process_options): Autodetect value for debug statement
2552 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2553 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2555 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2556 markers. Integrate source bind into debug stmt expand loop.
2557 (pass_expand::execute): Check debug marker limit. Avoid deep
2558 TER and expand debug locations for debug bind insns only.
2559 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2561 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2562 * final.c (reemit_insn_block_notes): Take current block from
2563 nonbind markers. Declare note where it's first set.
2564 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2565 begin stmt markers if enabled.
2566 (notice_source_line): Handle nonbind markers. Fail if their
2567 location is unknown or that of builtins.
2568 (rest_of_handle_final): Convert begin stmt markers to notes if
2569 var-tracking didn't run.
2570 (rest_of_clean_state): Skip begin stmt markers.
2571 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2573 * function.c (allocate_struct_function): Set begin_stmt_markers.
2574 * function.h (struct function): Add debug_marker_count counter
2575 and debug_nonbind_markers flag.
2576 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2577 * gimple-low.c (lower_function_body): Adjust
2578 debug_nonbind_markers.
2579 (lower_stmt): Drop or skip gimple debug stmts.
2580 (lower_try_catch): Skip debug stmts.
2581 * gimple.c (gimple_build_debug_begin_stmt): New.
2582 (gimple_copy): Increment debug_marker_count if copying one.
2583 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2584 * gimplify.c (rexpr_location): New.
2585 (rexpr_has_location): New.
2586 (warn_switch_unreachable_r): Handle gimple debug stmts.
2587 (shortcut_cond_r): Call expr_location.
2589 (find_goto_label): New.
2590 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2592 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2594 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2595 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2596 (LANG_HOOKS_INITIALIZER): ... this.
2597 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2598 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2600 (update_ebb_live_info): Skip debug insn markers.
2601 * lra.c (debug_insn_static_data): Rename to...
2602 (debug_bind_static_data): ... this.
2603 (debug_marker_static_data): New.
2604 (lra_set_insn_recog_data): Select one of the above depending
2606 (lra_update_isn_regno_info): Don't assume debug insns have
2608 (push_insns): Skip debug insns.
2609 * lto-streamer-in.c (input_function): Drop debug stmts
2610 depending on active options. Adjust debug_nonbind_markers.
2611 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2612 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2613 begin stmt marker notes.
2614 (print_insn): Likewise.
2615 * recog.c (extract_insn): Recognize rtl for debug markers.
2616 * rtl.def (DEBUG_MARKER): New.
2617 * tree-inline.c: Include params.h.
2618 (remap_gimple_stmt): Handle nonbind markers.
2619 (maybe_move_debug_stmts_to_successors): Likewise.
2620 (copy_debug_stmt): Likewise.
2621 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2622 markers regardless of no side effects.
2623 (tsi_link_before): Don't update container's side effects when adding
2624 a begin stmt marker.
2625 (tsi_link_after): Likewise.
2626 (expr_first): Skip begin stmt markers.
2627 (expr_last): Likewise.
2628 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2629 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2630 Disregard nonbind markers.
2631 * tree.c (make_node_stat): Don't set side effects for begin stmt
2633 (build1_stat): Likewise.
2634 * tree.def (DEBUG_BEGIN_STMT): New.
2635 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2636 * var-tracking.c (delete_debug_insns): Renamed to...
2637 (delete_vta_debug_insns): ... this.
2638 (reemit_marker_as_note): New.
2639 (vt_initialize): Reemit markers.
2640 (delete_vta_debug_insns): Likewise.
2641 (vt_debug_insns_local): Reemit or delete markers.
2642 (variable_tracking_main_1): Likewise.
2643 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2644 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2645 (gimple_debug_nonbind_marker_p): New.
2646 (gimple_build_debug_bind): Adjust.
2647 (gimple_build_debug_begin_stmt): New.
2648 * doc/invoke.texi (max-debug-marker-count): New param.
2649 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2650 (debug_parameter_ref, debug_marker): New.
2651 (NOTE_INSN_BEGIN_STMT): New.
2652 (DEBUG_INSN): Describe begin stmt markers.
2654 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2655 without debug insns, we wouldn't, but clean up debug insns
2656 after a control flow insn nevertheless.
2658 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2659 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2660 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2661 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2662 (rtl_tidy_fallthru_edge): Likewise.
2663 (rtl_verify_fallthru): Likewise.
2664 (rtl_verify_bb_layout): Likewise.
2665 (skip_insns_after_block): Likewise.
2666 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2667 * dwarf2out.c: Include print-rtl.h.
2668 (dwarf2out_next_real_insn): New.
2669 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2670 Dump debug binds in asm comments.
2671 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2672 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2673 callers to use gsi_start_nondebug_bb instead.
2674 (gsi_after_labels): Skip gimple debug stmts.
2675 (gsi_start_nondebug): New.
2676 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2677 (proper_loop_form_for_interchange): Adjust.
2678 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2679 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2680 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2681 (collect_fallthrough_labels): Likewise.
2682 (should_warn_for_implicit_fallthrough): Likewise.
2683 (warn_implicit_fallthrough_r): Likewise.
2684 (expand_FALLTHROUGH_r): Likewise.
2685 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2686 (graphite_copy_stmts_from_block): Skip nonbind markers.
2687 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2688 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2689 * jump.c (clean_barriers): Skip debug insns.
2690 * omp-expand.c (expand_parallel_call): Skip debug insns.
2691 (expand_task_call): Likewise.
2692 (remove_exit_barrier): Likewise.
2693 (expand_omp_taskreg): Likewise.
2694 (expand_omp_for_init_counts): Likewise.
2695 (expand_omp_for_generic): Likewise.
2696 (expand_omp_for_static_nochunk): Likewise.
2697 (expand_omp_for_static_chunk): Likewise.
2698 (expand_omp_simd): Likewise.
2699 (expand_omp_taskloop_for_outer): Likewise.
2700 (expand_omp_taskloop_for_inner): Likewise.
2701 (expand_oacc_for): Likewise.
2702 (expand_omp_sections): Likewise.
2703 (expand_omp_single): Likewise.
2704 (expand_omp_synch): Likewise.
2705 (expand_omp_atomic_load): Likewise.
2706 (expand_omp_atomic_store): Likewise.
2707 (expand_omp_atomic_fetch_op): Likewise.
2708 (expand_omp_atomic_pipeline): Likewise.
2709 (expand_omp_atomic_mutex): Likewise.
2710 (expand_omp_target): Likewise.
2711 (grid_expand_omp_for_loop): Likewise.
2712 (grid_expand_target_grid_body): Likewise.
2713 (build_omp_regions_1): Likewise.
2714 * omp-low.c (check_combined_parallel): Skip debug stmts.
2715 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2716 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2717 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2719 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2720 (make_edges): Likewise.
2721 (cleanup_dead_labels): Likewise.
2722 (gimple_can_merge_blocks_p): Likewise.
2723 (stmt_starts_bb_p): Likewise.
2724 (gimple_block_label): Likewise.
2725 (gimple_redirect_edge_and_branch): Likewise.
2726 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2728 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2730 * tree-pretty-print (print_declaration): Omit initializer in slim
2732 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2734 (eliminate_unnecessary_stmts): Stabilize block removal order.
2735 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2736 * var-tracking.c (get_first_insn): New.
2737 (vt_emit_notes): Call it.
2738 (vt_initialize): Walk any insns before the first BB.
2739 (delete_debug_insns): Likewise.
2741 * gimple.h (enum gimple_debug_subcode): Add
2742 GIMPLE_DEBUG_BEGIN_STMT.
2743 (gimple_debug_begin_stmt_p): New.
2744 (gimple_debug_nonbind_marker_p): New.
2745 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2746 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2747 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2748 * insn-notes.def (BEGIN_STMT): New.
2749 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2750 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2751 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2752 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2753 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2754 (INSN_DEBUG_MARKER_KIND): New.
2755 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2756 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2757 (INSN_VAR_LOCATION_PTR): New.
2758 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2760 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2761 deps only if debug bind insns are enabled.
2762 (pass_expand::execute): Avoid deep TER and expand
2763 debug locations for debug bind insns only.
2764 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2765 debug stmts special handling down to debug bind stmts.
2766 * combine.c (try_combine): Narrow debug insns special handling
2767 down to debug bind insns.
2768 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2769 Narrow debug insns preexisting special handling down to debug
2771 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2772 handling down to debug bind insns.
2773 * function.c (instantiate_virtual_regs): Skip debug markers,
2774 adjust handling of debug binds.
2775 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2776 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2777 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2778 insns to debug bind insns.
2779 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2780 special handling of debug stmts to debug bind stmts.
2781 * ipa-split.c (split_function): Likewise.
2782 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2783 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2785 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2786 * regrename.c (build_def_use): Likewise.
2787 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2788 (pass_cprop_hardreg): Narrow special casing of debug insns to
2790 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2791 * reload1.c (reload): Likewise.
2792 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2793 casing of debug stmts to debug bind stmts.
2794 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2795 * ssa-iterators.h (num_imm_uses): Likewise.
2796 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2797 debug stmts to debug bind stmts.
2798 * tree-inline.c (tree_function_versioning): Narrow special casing
2799 of debug stmts to debug bind stmts.
2800 * tree-loop-distribution.c (generate_loops_for_partition):
2801 Narrow special casing of debug stmts to debug bind stmts.
2802 * tree-sra.c (analyze_access_subtree): Narrow special casing
2803 of debug stmts to debug bind stmts.
2804 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2805 stmts to debug bind stmts.
2806 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2807 casing of debug stmts to debug bind stmts.
2808 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2809 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2810 casing of debug stmts to debug bind stmts.
2811 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2813 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2814 debug stmts to debug bind stmts.
2815 (gimple_replace_ssa_lhs): Likewise.
2816 (insert_debug_temp_for_var_def): Likewise.
2817 (insert_debug_temps_for_defs): Likewise.
2818 (reset_debug_uses): Likewise.
2819 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2820 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2821 (adjust_debug_stmts): Likewise.
2822 (adjust_phi_and_debug_stmts): Likewise.
2823 (vect_do_peeling): Likewise.
2824 * tree-vect-loop.c (vect_transform_loop): Likewise.
2825 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2826 * var-tracking.c (adjust_mems): Narrow special casing of debug
2827 insns to debug bind insns.
2828 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2829 (compute_bb_dataflow, vt_find_locations): Likewise.
2830 (vt_expand_loc, emit_notes_for_changes): Likewise.
2831 (vt_init_cfa_base): Likewise.
2832 (vt_emit_notes): Likewise.
2833 (vt_initialize): Likewise.
2834 (vt_finalize): Likewise.
2836 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2837 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2838 (next_nonnote_nondebug_insn_bb): New.
2839 (prev_nonnote_nondebug_insn_bb): New.
2840 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2841 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2842 (prev_nonnote_nondebug_insn_bb): Declare.
2843 (next_nonnote_nondebug_insn_bb): Declare.
2844 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2845 * cfgrtl.c (get_last_bb_insn): Likewise.
2846 * lra.c (push_insns): Likewise.
2848 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2851 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2853 * selftest.c (selftest::test_runner::test_runner): New ctor.
2854 (selftest::test_runner::~test_runner): New dtor.
2855 * selftest.h (class selftest::test_runner): New class.
2857 2017-12-11 Carl Love <cel@us.ibm.com>
2859 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
2860 vec_extract_fp32_from_shortl]): Add #defines.
2861 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
2862 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
2863 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
2864 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
2865 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
2866 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
2867 * doc/extend.texi: Add documentation for the added builtins.
2869 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
2871 PR rtl-optimization/80693
2872 PR rtl-optimization/81019
2873 PR rtl-optimization/81020
2874 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
2875 are not mentioned in i3. Place the REG_UNUSED note on i2,
2876 possibly modified to REG_DEAD, if it did not originate in i3.
2878 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2880 * recog.c (store_data_bypass_p_1): New function.
2881 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
2882 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
2884 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
2886 PR rtl-optimization/83361
2887 * ifcvt.c (if_convert): Call fixup_partitions.
2889 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
2891 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2892 early folding of splat_u{8,16,32}.
2894 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2896 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
2897 output_operand_lossage first argument with capital letter.
2898 (aarch64_override_options): Don't start error and sorry first argument
2899 with capital letter.
2901 2017-12-11 Andi Kleen <ak@linux.intel.com>
2903 PR gcov-profile/83355
2904 * auto-profile.c (string_table::get_index_by_decl): Don't
2905 recurse when abstract origin points to itself.
2907 2017-12-11 Bin Cheng <bin.cheng@arm.com>
2909 PR tree-optimization/83320
2910 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
2911 (prune_datarefs_not_in_loop): Ditto.
2913 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
2915 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
2917 2017-12-10 Jakub Jelinek <jakub@redhat.com>
2919 PR tree-optimization/83337
2920 * gimple-loop-interchange.cc (compute_access_stride): Handle
2921 bitfield DRs properly.
2923 2017-12-09 Jakub Jelinek <jakub@redhat.com>
2925 PR tree-optimization/83338
2926 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
2927 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
2928 vector of unsigned integers to vector of signed integers.
2930 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
2932 PR rtl-optimization/83317
2933 * lra-constraints.c (process_address_1): Add insn code check.
2935 2017-12-08 Michael Matz <matz@suse.de>
2937 Fix PR tree-optimization/83323
2938 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
2939 head-controlled loops and loop BBs.
2940 * common.opt (funroll-and-jam): Remove, instead ...
2941 (floop-unroll-and-jam): ... reuse this option.
2942 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
2943 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
2944 (-floop-unroll-and-jam): ... this option.
2946 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2948 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
2949 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
2950 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
2952 2017-12-08 Richard Biener <rguenther@suse.de>
2955 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
2956 handle accesses outside of zero-sized vars.
2958 2017-12-08 Martin Jambor <mjambor@suse.cz>
2960 PR tree-optimization/83141
2961 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
2962 test for MEM_REFs implicitely changing types with padding. Remove
2964 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
2966 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2968 * config/arc/arc.c (arc_attribute_table): Add exclusions to
2970 * config/avr/avr.c (avr_attribute_table): Likewise.
2971 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2972 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2973 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2974 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
2975 of struct attribute_spec.
2976 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2978 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
2981 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
2983 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
2986 * profile-count.c (profile_count::from_gcov_type): Move from
2987 profile-count.h; handle overflow.
2988 * profile-count.h (profile_count::from_gcov_type): Move offline.
2990 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
2992 PR rtl-optimization/83304
2993 * combine.c (move_deaths): If we do not know where a register died,
2996 2017-12-08 Richard Biener <rguenther@suse.de>
2998 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2999 Provide -fopt-info-loop feedback when we interchange in a nest.
3001 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3003 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3004 for armv6 ARM CPU IDs.
3006 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3008 * common/config/arm/arm-common.c: Include <algorithm>.
3009 (INCLUDE_VECTOR): Define.
3010 (compare_opt_names): New function.
3011 (arm_rewrite_selected_arch): Only strip out extensions that can be
3012 expressed through -mfpu. Sort the remaining extensions
3015 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3017 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3018 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3019 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3020 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3022 2017-06-08 Tristan Gingold <gindold@adacore.com>
3025 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3026 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3028 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3030 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3031 of struct attribute_spec.
3033 2017-12-08 Julia Koval <julia.koval@intel.com>
3035 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3036 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3037 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3038 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3039 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3040 _mm_maskz_dpwssds_epi32): Ditto.
3042 2017-12-08 Richard Biener <rguenther@suse.de>
3044 PR tree-optimization/81303
3045 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3046 conditions try to create a comparison vector type matching
3047 the data vector type.
3048 (vectorizable_condition): Adjust.
3049 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3050 Leave invariant conditions alone in case we can vectorize those.
3052 2017-12-08 Julia Koval <julia.koval@intel.com>
3054 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3055 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3056 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3057 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3058 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3060 2017-12-08 Julia Koval <julia.koval@intel.com>
3062 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3063 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3064 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3065 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3066 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3067 _mm_maskz_dpbusds_epi32): New intrinsics.
3069 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3071 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3073 (ld<bh>io_signed): Likewise.
3074 (st<bhw>io): Likewise.
3075 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3076 SMALL_INT12 constant integer operand.
3078 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3080 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3082 (nios2_large_constant_allowed): ...this. Adjust uses.
3083 (nios2_plus_symbolic_constant_p): Rename to...
3084 (nios2_plus_large_constant_p): ...this. Adjust uses.
3085 (nios2_legitimate_address_p): Correct CONST_INT handling.
3086 (nios2_symbolic_memory_operand_p): Rename to...
3087 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3088 (nios2_large_constant_p): Check for large constant integers too.
3089 (nios2_split_large_constant): Handle constant integers.
3090 (nios2_split_symbolic_memory_operand): Rename to...
3091 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3092 (nios2_legitimize_constant_address): Handle constant integers.
3093 (r0rel_constant_p): Handle small constant integers.
3094 (nios2_print_operand_address): Handle r0-relative integer addresses.
3095 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3096 * config/nios2/nios2.md: Adjust for renamed functions.
3098 2017-12-07 Andrew Waterman <andrew@sifive.com>
3100 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3101 (TARGET_HAVE_SRODATA_SECTION): New define.
3102 (riscv_select_section): New function.
3104 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3105 Alexander Monakov <amonakov@ispras.ru>
3106 Jakub Jelinek <jakub@redhat.com>
3109 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3111 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3113 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3114 a bool success value. Don't call output_operand_lossage here.
3115 (aarch64_print_ldpstp_address): Return a bool success value.
3116 (aarch64_print_operand_address): Call output_addr_const if
3117 aarch64_print_address_internal fails.
3118 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3119 'y'; call output_operand_lossage instead. Call output_operand_lossage
3120 if aarch64_print_ldpstp_address fails.
3122 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3124 * tree-vector-builder.h
3125 (tree_vector_builder::binary_encoded_nelts): Declare.
3126 * tree-vector-builder.c
3127 (tree_vector_builder::binary_encoded_nelts): New function.
3128 * fold-const.c (negate_expr_p): Likewise.
3129 (operand_equal_p, fold_checksum_tree): Likewise.
3130 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3131 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3132 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3133 (uniform_vector_p): Likewise.
3134 * varasm.c (const_hash_1, compare_constant): Likewise.
3135 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3136 (valid_lattice_transition): Operate directly on the VECTOR_CST
3138 * ipa-icf.c: Include tree-vector-builder.h.
3139 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3140 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3142 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3144 * tree.c (build_vector): Delete.
3145 * tree.h (build_vector): Make static and move into the self-testing
3148 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3150 * vector-builder.h (vector_builder::derived): New const overload.
3151 (vector_builder::elt): New function.
3152 * tree-vector-builder.h (tree_vector_builder::type): New function.
3153 (tree_vector_builder::apply_step): Declare.
3154 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3156 * gimple-fold.h (tree_vector_builder): Declare.
3157 (gimple_build_vector): Take a tree_vector_builder instead of a
3158 type and vector of elements.
3159 * gimple-fold.c (gimple_build_vector): Likewise.
3160 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3162 (get_initial_defs_for_reduction): Likewise.
3163 (vectorizable_induction): Likewise.
3165 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3167 * tree-vector-builder.h
3168 (tree_vector_builder::new_binary_operation): Declare.
3169 * tree-vector-builder.c
3170 (tree_vector_builder::new_binary_operation): New function.
3171 * fold-const.c (fold_relational_const): Use it.
3172 (const_binop): Likewise. Check that both input vectors have
3173 the same number of elements, thus excluding things like WIDEN_SUM.
3174 Check whether it is possible to operate directly on the encodings
3177 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3179 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3180 new_unary_operation, operating only on the encoded elements.
3181 (const_unop): Likewise.
3182 (exact_inverse): Likewise.
3183 (distributes_over_addition_p): New function.
3184 (const_binop): Use tree_vector_builder and new_unary_operation
3185 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3186 on the encoded elements unless the encoding is strided and the
3187 operation does not distribute over addition.
3188 (fold_convert_const): Use tree_vector_builder and
3189 new_unary_operation. Operate only on the encoded elements
3190 for truncating integer conversions, or for non-stepped encodings.
3192 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3194 * config/sparc/sparc.c: Include tree-vector-builder.h.
3195 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3196 * expmed.c: Include tree-vector-builder.h.
3197 (make_tree): Use tree_vector_builder instead of build_vector.
3198 * fold-const.c: Include tree-vector-builder.h.
3199 (const_binop): Use tree_vector_builder instead of build_vector.
3200 (const_unop): Likewise.
3201 (native_interpret_vector): Likewise.
3202 (fold_vec_perm): Likewise.
3203 (fold_ternary_loc): Likewise.
3204 * gimple-fold.c: Include tree-vector-builder.h.
3205 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3207 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3208 (simplify_vector_constructor): Use tree_vector_builder instead
3210 * tree-vect-generic.c: Include tree-vector-builder.h.
3211 (add_rshift): Use tree_vector_builder instead of build_vector.
3212 (expand_vector_divmod): Likewise.
3213 (optimize_vector_constructor): Likewise.
3214 * tree-vect-loop.c: Include tree-vector-builder.h.
3215 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3216 of build_vector. Explicitly use a stepped encoding for
3218 * tree-vect-slp.c: Include tree-vector-builder.h.
3219 (vect_get_constant_vectors): Use tree_vector_builder instead
3221 (vect_transform_slp_perm_load): Likewise.
3222 (vect_schedule_slp_instance): Likewise.
3223 * tree-vect-stmts.c: Include tree-vector-builder.h.
3224 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3225 (vect_gen_perm_mask_any): Likewise.
3226 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3227 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3229 (build_vector_from_val): Likewise. Explicitly use a duplicate
3232 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3234 * doc/generic.texi (VECTOR_CST): Describe new representation of
3236 * vector-builder.h: New file.
3237 * tree-vector-builder.h: Likewise.
3238 * tree-vector-builder.c: Likewise.
3239 * Makefile.in (OBJS): Add tree-vector-builder.o.
3240 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3241 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3242 (tree_vector): Change the number of elements to
3243 vector_cst_encoded_nelts.
3244 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3245 (VECTOR_CST_ELTS): Delete.
3246 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3247 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3248 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3249 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3250 (VECTOR_CST_ENCODED_ELT): Likewise.
3251 (vector_cst_encoded_nelts): New function.
3252 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3253 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3254 (vector_cst_int_elt, vector_cst_elt): Declare.
3255 * tree.c: Include tree-vector-builder.h.
3256 (tree_code_size): Abort if passed VECTOR_CST.
3257 (tree_size): Update for new VECTOR_CST layout.
3258 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3259 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3260 (build_vector): Use tree_vector_builder.
3261 (vector_cst_int_elt, vector_cst_elt): New functions.
3262 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3263 encoded elements and then create the vector in the canonical form.
3264 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3265 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3266 (tree_c_tests): Call test_vector_cst_patterns.
3267 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3269 (hash_tree): Likewise.
3270 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3271 (streamer_write_tree_header): Likewise.
3272 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3273 (streamer_alloc_tree): Likewise. Update call to make_vector.
3274 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3276 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3278 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3279 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3280 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3282 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3283 Richard Biener <rguenther@suse.de>
3285 PR tree-optimization/81303
3286 * Makefile.in (gimple-loop-interchange.o): New object file.
3287 * common.opt (floop-interchange): Reuse the option from graphite.
3288 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3289 -floop-interchange and mention it for -O3.
3290 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3291 * gimple-loop-interchange.cc: New file.
3292 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3293 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3294 * passes.def (pass_linterchange): New pass.
3295 * timevar.def (TV_LINTERCHANGE): New time var.
3296 * tree-pass.h (make_pass_linterchange): New declaration.
3297 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3298 interchange. Record IV before/after increment in new parameters.
3299 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3300 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3302 (check_reduction_path): ...New function here.
3303 * tree-vectorizer.h (check_reduction_path): New declaration.
3305 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3308 PR rtl-optimization/80818
3309 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3310 always early clobbered.
3311 * lra-lives.c (process_bb_lives): Check input hard regs for early
3312 clobbered non-operand hard reg.
3314 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3317 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3318 types_compatible_p, just that TYPE_MODE is the same.
3320 2017-12-07 Martin Sebor <msebor@redhat.com>
3323 * attribs.c (empty_attribute_table): Initialize new member of
3324 struct attribute_spec.
3325 (decl_attributes): Add argument. Handle mutually exclusive
3326 combinations of attributes.
3327 (selftests::test_attribute_exclusions): New function.
3328 (selftests::attribute_c_tests): Ditto.
3329 * attribs.h (decl_attributes): Add default argument.
3330 * selftest.h (attribute_c_tests): Declare.
3331 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3332 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3334 * doc/extend.texi (Common Function Attributes): Update const and pure.
3335 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3336 of struct attribute_spec.
3337 * config/arc/arc.c (arc_attribute_table): Same.
3338 * config/arm/arm.c (arm_attribute_table): Same.
3339 * config/avr/avr.c ( avr_attribute_table): Same.
3340 * config/bfin/bfin.c (bfin_attribute_table): Same.
3341 * config/cr16/cr16.c (cr16_attribute_table): Same.
3342 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3343 * config/h8300/h8300.c (h8300_attribute_table): Same.
3344 * config/i386/i386.c (ix86_attribute_table): Same.
3345 * config/ia64/ia64.c (ia64_attribute_table): Same.
3346 * config/m32c/m32c.c (m32c_attribute_table): Same.
3347 * config/m32r/m32r.c (m32r_attribute_table): Same.
3348 * config/m68k/m68k.c (m68k_attribute_table): Same.
3349 * config/mcore/mcore.c (mcore_attribute_table): Same.
3350 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3351 * config/mips/mips.c (mips_attribute_table): Same.
3352 * config/msp430/msp430.c (msp430_attribute_table): Same.
3353 * config/nds32/nds32.c (nds32_attribute_table): Same.
3354 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3355 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3356 * config/rl78/rl78.c (rl78__attribute_table): Same.
3357 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3358 * onfig/rx/rx.c (rx_attribute_table): Same.
3359 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3360 * config/sh/sh.c (sh_attribute_table): Same.
3361 * config/sparc/sparc.c (sparc_attribute_table): Same.
3362 * config/spu/spu.c (spu_attribute_table): Same.
3363 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3364 * config/v850/v850.c (v850_attribute_table): Same.
3365 * config/visium/visium.c (visium_attribute_table): Same.
3367 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3370 * config/arm/arm.c (INCLUDE_STRING): Define.
3371 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3372 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3375 2017-12-07 Michael Matz <matz@suse.de>
3377 Add unroll and jam pass
3379 * gimple-loop-jam.c: New file.
3380 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3381 * common.opt (funroll-and-jam): New option.
3382 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3383 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3384 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3385 * passes.def: Add pass_loop_jam.
3386 * timevar.def (TV_LOOP_JAM): Add.
3387 * tree-pass.h (make_pass_loop_jam): Declare.
3388 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3389 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3390 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3391 to flow_loop_tree_node_add.
3392 (duplicate_subloops, copy_loops_to): Append to sibling list.
3393 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3394 * doc/invoke.texi (-funroll-and-jam): Document new option.
3395 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3397 2017-12-07 Richard Biener <rguenther@suse.de>
3399 PR tree-optimization/83296
3400 PR tree-optimization/67769
3401 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3402 flow sensitive info in an unrelated BB.
3403 (value_replacement): Use reset_flow_sensitive_info.
3404 (minmax_replacement): Reset flow sensitive info on the def
3405 we move. Do not reset flow sensitive info in the whole BB
3406 we move the stmt to.
3407 (abs_replacement): Likewise.
3409 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3412 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3413 rs6000_cpu to the given -mcpu=, or to the default processor.
3415 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3417 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3418 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3419 instead of rs6000_cpu_attr.
3420 (rs6000_adjust_cost): Ditto.
3421 (is_microcoded_insn): Ditto.
3422 (rs6000_adjust_priority): Ditto.
3423 (rs6000_issue_rate): Ditto.
3424 (rs6000_use_sched_lookahead): Ditto.
3425 (rs6000_use_sched_lookahead_guard): Ditto.
3426 (rs6000_sched_reorder): Ditto.
3427 (force_new_group): Ditto.
3428 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3429 (group_ending_nop): Ditto.
3431 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3433 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3434 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3435 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3436 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3438 (rs6000_reassociation_width): Ditto.
3439 (rs6000_emit_epilogue): Ditto.
3440 (rs6000_adjust_cost): Ditto.
3441 (is_microcoded_insn): Ditto.
3442 (is_cracked_insn): Ditto.
3443 (rs6000_adjust_priority): Ditto.
3444 (rs6000_sched_reorder): Ditto.
3445 (rs6000_sched_reorder2): Ditto.
3446 (insn_must_be_first_in_group): Ditto.
3447 (insn_must_be_last_in_group): Ditto.
3448 (rs6000_register_move_cost): Ditto.
3449 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3452 2017-12-07 Julia Koval <julia.koval@intel.com>
3454 * config.gcc: Add vaesintrin.h.
3455 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3457 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3458 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3460 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3461 * config/i386/immintrin.h: Include vaesintrin.h.
3462 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3463 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3464 _mm_aesdec_epi128): New intrinsics.
3466 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3468 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3469 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3470 c-family/c-spellcheck.cc.
3471 (best_macro_match::best_macro_match): Likewise.
3473 (struct edit_distance_traits<cpp_hashnode *>): Move to
3474 c-family/c-spellcheck.h.
3475 (class best_macro_match): Likewise.
3477 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3479 PR tree-optimization/83293
3480 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3481 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3482 might insert into empty bb.
3485 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3486 simplify for plus with :c added, and pointer_plus without that.
3487 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3488 with undefined overflow and the conversion is not widening,
3489 perform negation in utype and only convert to type afterwards.
3490 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3491 simplify for plus with :c added, and pointer_plus without that.
3492 If type is integral with undefined overflow and the conversion is
3493 not widening, perform minus in utype and only convert to type
3494 afterwards. Move the last pointer_diff_expr simplify into the
3497 2017-12-06 Martin Sebor <msebor@redhat.com>
3499 PR tree-optimization/82646
3500 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3501 strncpy, not maxlen.
3503 2017-12-06 Martin Sebor <msebor@redhat.com>
3505 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3508 PR tree-optimization/83075
3509 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3510 strncat/strncpy don't change length of source string.
3512 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3515 2017-11-29 Martin Aberg <maberg@gaisler.com>
3517 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3518 to prevent b2bst errata sequence.
3519 (sqrtdf2_fix): Likewise.
3521 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3523 PR tree-optimization/81945
3524 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3525 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3526 to dest_cfun has orig_loop_num set, either remap it to the new
3527 loop number if the loop got moved too, or clear it.
3529 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3531 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3533 (thunderx2t99_nothing): New insn reservation.
3534 (thunderx2t99_mrs): New insn reservation.
3535 (thunderx2t99_multiple): New insn reservation.
3536 (thunderx2t99_alu_basi): Add bfx to reservation.
3537 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3539 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3542 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3544 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3546 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3547 tree-ssa-pre.c::remove_dead_inserted_code.
3548 * tree-ssa-dce.h: New file.
3549 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3550 (remove_dead_inserted_code): Move and rename to function
3551 tree-ssa-dce.c::simple_dce_from_worklist.
3552 (pass_pre::execute): Update use.
3554 2017-12-05 Richard Biener <rguenther@suse.de>
3556 PR tree-optimization/83277
3557 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3558 to code-gen liveout vars.
3560 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3562 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3563 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3564 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3567 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3569 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3571 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3572 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3575 2017-12-05 Richard Biener <rguenther@suse.de>
3577 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3578 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3579 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3580 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3582 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3584 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3586 2017-12-05 Martin Liska <mliska@suse.cz>
3587 Jakub Jelinek <jakub@redhat.com>
3589 * doc/invoke.texi: Document the options.
3590 * flag-types.h (enum sanitize_code): Add
3591 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3592 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3593 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3594 * opts.c: Define new sanitizer options.
3595 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3596 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3598 2017-12-05 Julia Koval <julia.koval@intel.com>
3600 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3601 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3602 (ix86_handle_option): Handle -mavx512vnni.
3603 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3604 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3605 * config/i386/i386-c (__AVX512VNNI__): New.
3606 * config/i386/i386.c (ix86_target_string): Handle new option.
3607 (ix86_valid_target_attribute_inner_p): Handle new option.
3608 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3609 * config/i386/i386.opt (mavx512vnni): New option.
3611 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3614 * athlon.md: Disable for generic.
3615 * haswell.md: Enable for generic.
3616 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3617 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3619 (ix86_adjust_cost): Move generic to haswell path.
3621 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3623 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3624 instead of MEM_P in a couple more places. Fix formatting issues.
3626 2017-12-04 Jim Wilson <jimw@sifive.com>
3628 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3629 instead of GP_REG_LAST-1.
3630 (riscv_adjust_libcall_cfi_prologue): Likewise.
3631 (riscv_adjust_libcall_cri_epilogue): Likewise.
3632 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3635 2017-12-04 Luis Machado <luis.machado@linaro.org>
3637 * ipa-pure-const.c (check_decl): Add missing newline.
3638 (state_from_flags): Likewise.
3640 2017-12-04 Jeff Law <law@redhat.com>
3642 PR tree-optimizatin/78496
3643 * gimple-ssa-evrp-analyze.h
3644 (evrp_range_analyzer::get_vr_values): Simplify.
3645 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3646 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3647 and gimple-ssa-evrp-analyze.h.
3648 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3649 (simplify_stmt_for_jump_threading): Copy a blob of code from
3650 tree-vrp.c to use ranges to simplify statements.
3651 (dom_opt_dom_walker::before_dom_children): Call
3652 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3653 (dom_opt_dom_walker::after_dom_children): Similarly for
3654 evrp_range_analyzer::leave.
3655 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3658 * gimple-ssa-evrp-analyze.c
3659 (evrp_range_analyzer::extract_range_from_stmt): Always use
3660 vr_values::update_value_range so preexisting range info is
3661 medged with new range info, even if the new range is VR_VARYING.
3663 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3665 * combine.c: Adjust comment.
3666 (use_crosses_set_p): Delete.
3667 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3668 (try_combine): Ditto.
3670 2017-12-04 Richard Biener <rguenther@suse.de>
3672 PR tree-optimization/83255
3673 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3674 Re-add zero-iteration check.
3676 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3678 PR rtl-optimization/83245
3679 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3680 hard registers as earlyclobber, also if not in an asm.
3682 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3686 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3689 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3690 rs6000_cpu based on cpu_index, not tune_index.
3692 2017-12-04 Richard Biener <rguenther@suse.de>
3694 PR tree-optimization/83238
3695 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3696 code match comment, rejecting invalid SESE regions.
3698 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3700 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3701 require base operand is a REG_POINTER prior to reload on targets
3702 with non-equivalent space registers.
3704 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3706 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3707 (update_specialized_profile): Fix updating of counts.
3708 (perhaps_add_new_callers): Likewise.
3710 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3713 * x86-tune.def: Remove obsolette FIXMEs.
3714 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3715 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3716 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3718 (X86_TUNE_PAD_RETURNS): Disable for generic.
3720 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3722 PR tree-optimization/83170
3723 PR tree-optimization/83241
3724 * gimple-ssa-store-merging.c
3725 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3726 gimple_vuse (ins_stmt) in case it has changed.
3727 (imm_store_chain_info::output_merged_store): Likewise.
3729 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3733 * tree-cfg.c (pass_warn_function_return::execute): Handle
3734 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3735 with BUILTINS_LOCATION.
3739 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3740 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3741 instead of DECL_MODE.
3743 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3744 last argument to rtx pointer.
3745 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3746 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3747 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3748 depending on the chosen ISAs.
3749 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3750 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3751 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3753 * config/i386/sse.md (mov<mode>_internal): Likewise.
3754 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3756 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3758 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3760 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3762 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3763 parameters from prototype.
3764 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3765 parameters. Don't print a comment.
3766 (emit_fusion_gpr_load): Adjust.
3767 (emit_fusion_load_store): Adjust.
3768 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3769 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3770 comment on the second line.
3772 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3775 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3776 rs6000_cpu based on cpu_index, not tune_index.
3778 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3780 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3781 which_alternative instead of which_alternative + 1.
3782 (output_asm_insn): Print an extra tab if the template is short.
3784 2017-12-01 Jim Wilson <jimw@sifive.com>
3786 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3788 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3789 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3791 * doc/tm.texi: Regenerate.
3793 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3796 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3797 whether we can allocate pseudos before trying to fix an address.
3798 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3799 memory address is indexed or indirect.
3800 (floatuns_<mode>si2_hw2): Likewise.
3802 2017-12-01 Jason Merrill <jason@redhat.com>
3804 * Makefile.in (TAGS): Add c-family/*.cc.
3806 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3808 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3809 (call_value_insn): Likewise.
3810 (sibcall_insn): Likewise.
3811 (sibcall_value_insn): Likewise.
3812 (movsi_aarch64): Likewise.
3813 (movdi_aarch64): Likewise.
3814 (add_losym_): Likewise.
3815 (ldr_got_small_): Likewise.
3816 (ldr_got_small_sidi): Likewise.
3817 (ldr_got_small_28k_): Likewise.
3818 (ldr_got_small_28k_sidi): Likewise.
3819 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3820 Move output_addr_const to symbolic case. Add error check.
3822 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3824 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3825 * config/s390/s390.md: Disable MVC merging peephole if it would
3826 disable operand forwarding.
3827 (new peephole2): Split MVCs if it would turn them into up to 2
3830 2017-12-01 Richard Biener <rguenther@suse.de>
3832 PR tree-optimization/83232
3833 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3834 detection of same access. Instead of breaking the group here
3835 do not consider the duplicate. Add comment explaining real fix.
3837 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3839 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3841 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3843 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3845 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3847 * function.h (struct function): Remove cilk_frame_decl,
3848 is_cilk_function and calls_cilk_spawn fields.
3849 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3851 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3852 cilk_elemental field.
3853 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3854 * target.def: Adjust comment.
3855 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3856 Don't test cilk_elemental.
3858 PR tree-optimization/83233
3859 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
3860 bswap_stat name for the struct.
3863 * tree.h (SWITCH_BREAK_LABEL_P): Define.
3864 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
3865 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
3866 SWITCH_BREAK_LABEL_P set on the label.
3867 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
3868 added for default case if it was missing and not all cases covered.
3869 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
3870 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
3872 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
3873 to avoid -Wimplicit-fallthrough warning.
3874 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
3875 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
3877 PR tree-optimization/83221
3878 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
3880 (init_reassoc): Formatting fix.
3883 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
3884 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
3888 * tree-cfg.c: Include asan.h.
3889 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
3891 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3893 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
3895 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3897 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
3898 vpcmpeqd instruction.
3900 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3902 * config/i386/i386.c (standard_sse_constant_opcode): Fix
3903 registers type for 128bit mode.
3905 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3907 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
3909 * gengtype-state.c (read_state_pair): Likewise.
3910 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
3911 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
3912 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
3913 * ifcvt.c (noce_try_store_flag_constants): Likewise.
3914 * tree-ssa-ccp.c (ccp_finalize): Likewise.
3915 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
3916 * builtins.c (fold_builtin_3): Likewise.
3917 * graphite-scop-detection.c
3918 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
3919 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
3922 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
3925 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
3926 parameter. Return true if ignore_decl_rtl_set_p is true and other
3927 conditions are satisfied.
3928 * asan.h (asan_protect_global): Add new parameter.
3929 * varasm.c (categorize_decl_for_section): Pass true as second parameter
3930 to asan_protect_global calls.
3932 2017-11-30 Jim Wilson <jimw@sifive.com>
3934 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
3935 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
3936 -mstrict-align, add info on default value. Delete redundant lines for
3937 -mabi. Add missing -mexplicit-relocs docs.
3939 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3941 * config/arc/arc.md (trap): New pattern.
3943 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3945 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
3946 instruction to end into a delay slot.
3947 * config/arc/arc.md (cond_delay_insn): Check if the instruction
3948 can be placed into a delay slot against reg_note.
3950 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3952 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
3953 labels number of usages.
3955 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3957 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
3959 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
3961 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3964 * internal-fn.c (expand_mul_overflow): Optimize unsigned
3965 multiplication by power of 2 constant into two shifts + comparison.
3967 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
3970 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
3972 2017-11-30 Richard Biener <rguenther@suse.de>
3974 PR tree-optimization/83202
3975 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3976 allow_peel argument and guard peeling.
3977 (canonicalize_loop_induction_variables): Likewise.
3978 (canonicalize_induction_variables): Pass false.
3979 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
3980 peeling from cunrolli.
3982 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3984 * combine.c (try_combine): Print a message to dump file whenever
3985 I0, I1, or I2 cannot be combined into I3.
3987 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3989 PR rtl-optimization/83156
3990 PR rtl-optimization/82621
3991 * combine.c (try_combine): Don't split an I2 if one of the dests is
3992 set again before I3. Allow unused dests.
3994 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3996 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
3998 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4000 PR rtl-optimization/80818
4001 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4002 recursively. Use insn code for clobber.
4003 (lra_set_insn_recog_data): Pass the new arg to
4004 collect_non_operand_hard_regs.
4005 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4007 (lra_update_insn_regno_info): Pass insn to
4008 add_regs_to_insn_regno_info.
4010 2017-11-29 Jim Wilson <jimw@sifive.com>
4011 Andrew Waterman <andrew@sifive.com>
4013 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4014 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4015 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4016 use SINGLE_SHIFT_COST.
4017 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4018 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4020 2017-11-29 Julia Koval <julia.koval@intel.com>
4022 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4023 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4024 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4025 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4026 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4027 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4028 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4029 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4030 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4031 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4032 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4033 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4034 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4035 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4036 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4037 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4038 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4039 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4040 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4041 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4042 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4043 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4044 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4045 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4046 __builtin_ia32_vpshldv_v2di_mask,
4047 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4048 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4049 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4051 2017-11-29 Julia Koval <julia.koval@intel.com>
4053 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4054 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4055 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4056 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4057 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4058 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4059 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4060 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4061 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4062 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4063 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4064 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4065 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4066 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4067 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4068 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4069 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4070 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4071 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4072 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4073 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4075 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4077 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4078 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4081 2017-11-29 Martin Aberg <maberg@gaisler.com>
4083 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4084 to prevent b2bst errata sequence.
4085 (sqrtdf2_fix): Likewise.
4087 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4089 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4090 (div_sqrt_insn_p): New function.
4091 (sparc_do_work_around_errata): Insert NOP instructions to
4092 prevent sequences that could trigger the TN-0013 errata for
4093 certain LEON3 processors.
4094 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4095 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4096 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4097 (in_branch_delay): Prevent div and sqrt in delay slot if
4099 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4101 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4103 * config/sparc/sparc.c (atomic_insn_p): New function.
4104 (sparc_do_work_around_errata): Insert NOP instructions to
4105 prevent sequences that could trigger the TN-0010 errata for
4107 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4108 instruction referable in atomic_insns_p.
4110 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4112 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4113 (atomic_compare_and_swap_leon3_1): Likewise.
4116 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4118 * config/sparc/sparc.c (fpop_insn_p): New function.
4119 (sparc_do_work_around_errata): Insert NOP instructions to
4120 prevent sequences that could trigger the TN-0012 errata for
4122 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4123 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4124 (in_branch_annul_delay): Prevent floating-point instructions
4125 in delay slot of annulled integer branch.
4127 2017-11-29 Richard Biener <rguenther@suse.de>
4129 PR tree-optimization/83202
4130 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4132 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4133 nodes vectorized to the same stmts multiple times.
4134 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4135 (vect_analyze_slp_instance): Adjust.
4136 (scalar_stmts_to_slp_tree_map_t): New typedef.
4137 (vect_schedule_slp_instance): Add a map recording the SLP node
4138 representing the vectorized stmts for a set of scalar stmts.
4139 Avoid code-generating redundancies.
4140 (vect_schedule_slp): Allocate map and pass it down.
4142 2017-11-29 Nathan Sidwell <nathan@acm.org>
4145 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4148 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4150 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4151 cases for printing LDP/STP memory addresses.
4152 (aarch64_print_address_internal): Renamed from
4153 aarch64_print_operand_address, added parameter, add Pmode check.
4154 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4155 (aarch64_print_operand_address): Indirect to
4156 aarch64_print_address_internal.
4157 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4158 'y' operand output specifier.
4160 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4163 * tree.c (build_simple_mem_ref_loc): Handle
4164 get_addr_base_and_unit_offset returning a MEM_REF.
4167 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4170 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4173 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4175 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4177 (layout::move_to_column): Likewise.
4179 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4182 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4183 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4185 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4186 Martin Jambor <mjambor@suse.cz>
4189 * tree.h (expr_type_first_operand_type_p): Declare
4190 * tree.c (expr_type_first_operand_type_p): New function.
4191 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4192 (ipa_value_from_jfunc): Adjust declaration.
4193 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4194 Use it as result type for arithmetics, unless it is NULL in which case
4195 be more conservative.
4196 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4197 ipa_get_jf_pass_through_result.
4198 (propagate_vals_across_pass_through): Likewise.
4199 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4200 is to propagate_vals_across_pass_through.
4201 (propagate_constants_across_call): Pass PARM_TYPE to
4202 propagate_scalar_across_jump_function.
4203 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4204 ipa_value_from_jfunc.
4205 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4206 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4207 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4208 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4209 target_type, pass it to ipa_value_from_jfunc.
4210 (update_indirect_edges_after_inlining): Pass parameter type to
4211 try_make_edge_direct_simple_call.
4213 2017-11-28 Jeff Law <law@redhat.com>
4215 * gimple-ssa-evrp-analyze.c
4216 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4217 refine ranges if scev_initialized_p returns true.
4218 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4220 2017-11-28 Julia Koval <julia.koval@intel.com>
4222 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4223 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4224 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4225 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4226 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4227 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4228 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4229 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4230 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4231 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4232 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4233 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4234 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4235 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4236 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4237 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4238 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4239 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4240 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4241 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4242 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4243 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4245 2017-11-28 Julia Koval <julia.koval@intel.com>
4247 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4248 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4249 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4250 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4251 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4252 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4253 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4254 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4255 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4256 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4257 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4258 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4259 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4260 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4261 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4262 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4263 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4264 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4265 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4266 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4267 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4268 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4269 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4270 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4271 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4272 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4273 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4274 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4275 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4276 __builtin_ia32_vpshld_v2di_mask): New builtins.
4277 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4278 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4280 2017-11-28 Richard Biener <rguenther@suse.de>
4282 PR tree-optimization/80776
4283 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4285 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4287 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4288 If the incoming edge is an effective fallthru because the other
4289 edge only reaches a __builtin_unreachable () then record ranges
4290 derived from the controlling condition in SSA info.
4291 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4292 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4294 2017-11-28 Olivier Hainque <hainque@adacore.com>
4296 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4297 on mingw build hosts.
4299 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4302 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4304 2017-11-28 Richard Biener <rguenther@suse.de>
4307 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4308 copies generated from memcpy use a character array as reference
4311 2017-11-28 Julia Koval <julia.koval@intel.com>
4312 Sebastian Peryt <sebastian.peryt@intel.com>
4314 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4315 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4316 cilk-common.o, cilk.h, cilk-common.c): Remove.
4318 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4319 * builtins.c (is_builtin_name): Remove cilkplus condition.
4320 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4321 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4322 cilk-builtins.def, cilkplus.def): Remove.
4323 * cif-code.def (CILK_SPAWN): Remove.
4324 * cilk-builtins.def: Delete.
4325 * cilk-common.c: Ditto.
4327 * cilkplus.def: Ditto.
4328 * config/darwin.h (fcilkplus): Delete.
4329 * cppbuiltin.c: Ditto.
4330 * doc/extend.texi: Remove cilkplus doc.
4331 * doc/generic.texi: Ditto.
4332 * doc/invoke.texi: Ditto.
4333 * doc/passes.texi: Ditto.
4334 * gcc.c (fcilkplus): Remove.
4335 * gengtype.c (cilk.h): Remove.
4336 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4338 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4340 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4342 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4343 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4344 cilkplus conditions.
4345 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4346 inline_read_section): Ditto.
4347 * ipa-inline-analysis.c (cilk.h): Remove.
4348 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4349 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4350 append_linker_options): Remove condition for fcilkplus.
4351 * lto/lto-lang.c (cilk.h): Remove.
4352 (lto_init): Remove condition for fcilkplus.
4353 * omp-expand.c (expand_cilk_for_call): Delete.
4354 (expand_omp_taskreg, expand_omp_for_static_chunk,
4355 expand_omp_for): Remove cilkplus
4357 (expand_cilk_for): Delete.
4358 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4359 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4360 execute_lower_omp, diagnose_sb_0): Ditto.
4361 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4362 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4363 * tree-nested.c: Ditto.
4364 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4365 (dump_generic_node): Ditto.
4366 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4367 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4369 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4371 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4373 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4374 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4375 for vector mode and !TARGET_SIMD.
4377 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4379 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4381 * tree.h (SWITCH_LABELS): Remove.
4382 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4383 assert SWITCH_BODY is non-NULL.
4384 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4386 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4388 PR tree-optimization/80788
4389 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4390 has TREE_OVERFLOW set, call drop_tree_overflow.
4392 2017-11-28 Richard Biener <rguenther@suse.de>
4394 PR tree-optimization/83158
4395 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4397 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4400 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4401 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4403 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4405 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4406 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4408 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4411 * varasm.c (compare_constant): Take the mode of the constants into
4412 account when comparing floating point constants.
4414 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4416 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4417 from explicit instantiation of debug_helper.
4418 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4420 2017-11-27 Richard Biener <rguenther@suse.de>
4422 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4425 2017-11-27 Richard Biener <rguenther@suse.de>
4427 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4428 (type_hash_canon): Also clear int_cst_hash_table entry for
4430 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4432 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4434 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4435 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4436 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4438 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4440 * hash-map.h (gt_cleare_cache): Avoid UB.
4442 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4444 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4445 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4446 (decide_unroll_runtime_iterations): Likewise.
4447 (decide_unroll_stupid): Likewise.
4449 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4452 * config/i386/i386.md: Add a loop with incssp.
4454 2017-11-27 Martin Jambor <mjambor@suse.cz>
4456 PR tree-optimization/81248
4457 * tree-sra.c (splice_param_accesses): Remove size check.
4458 (decide_one_param_reduction): Fix size check.
4459 * gimple-pretty-print.c (dump_profile): Silence warning.
4460 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4462 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4465 * dbxout.c (lastlineno): New variable.
4466 (dbx_debug_hooks): Use dbxout_switch_text_section as
4467 switch_text_section debug hook.
4468 (dbxout_function_end): Switch to current_function_section
4469 rather than function_section. If crtl->has_bb_partition,
4470 output just one N_FUN, depending on in_cold_section_p.
4471 (dbxout_source_line): Remember last lineno in lastlineno.
4472 (dbxout_switch_text_section): New function.
4473 (dbxout_function_decl): Adjust dbxout_block caller.
4474 (dbx_block_with_cold_children): New function.
4475 (dbxout_block): Return true if any LBRAC/RBRAC have been
4476 emitted. Use dbx_block_with_cold_children at depth == 0
4477 in second partition. Add PARENT_BLOCKNUM argument, pass
4478 it optionally adjusted to children. Output LBRAC/RBRAC
4479 around recursive call only if the block is in the current
4480 partition, if not and anything was output, emit empty
4482 * final.c (final_scan_insn): Compute cold_function_name
4483 before calling switch_text_section debug hook. Call
4484 that hook even if dwarf2out_do_frame if not emitting
4488 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4489 TREE_READONLY decls.
4491 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4493 PR rtl-optimization/82488
4494 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4496 2017-11-26 Julia Koval <julia.koval@intel.com>
4498 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4500 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4503 2017-11-26 Julia Koval <julia.koval@intel.com>
4505 * config/i386/driver-i386.c (host_detect_local_cpu):
4506 Detect skylake-avx512.
4508 2017-11-26 Julia Koval <julia.koval@intel.com>
4510 * config.gcc: Add -march=cannonlake.
4511 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4512 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4513 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4514 (PTA_CANNONLAKE): New.
4515 (processor_target_table): Add cannonlake.
4516 (ix86_option_override_internal): Ditto.
4517 (fold_builtin_cpu): Ditto.
4518 (get_builtin_code_for_version): Handle cannonlake.
4519 (M_INTEL_COREI7_CANNONLAKE): New.
4520 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4521 * doc/invoke.texi: Add -march=cannonlake.
4523 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4525 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4526 (try_init_one_plugin): Alternative implementation for MinGW.
4527 * Makefile.in (plugin_implib): New.
4528 (gengtype-lex.c): Fix broken AIX workaround.
4529 * configure: Regenerate.
4530 * doc/plugins.texi: Document support for MinGW.
4532 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4534 PR rtl-optimization/81553
4535 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4536 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4537 is a shift where C1 has different mode than the whole shift, use C1's
4538 mode for MULT rather than the shift's mode.
4541 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4542 builtins not enabled in the currently selected ISA.
4544 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4546 PR tree-optimization/71026
4547 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4548 (insert_reciprocals): Change to insert reciprocals before a division
4549 by a square and to insert the square of a reciprocal.
4550 (execute_cse_reciprocals_1): Change to consider division by a square.
4551 (register_division_in): Add importance parameter.
4553 2017-11-24 Richard Biener <rguenther@suse.de>
4555 PR tree-optimization/82402
4556 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4557 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4559 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4561 * match.pd (0-ptr): New transformation.
4563 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4566 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4567 when skipping checking.
4569 2017-11-24 Richard Biener <rguenther@suse.de>
4571 PR tree-optimization/83128
4572 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4573 (vn_reference_lookup_3): Likewise.
4575 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4578 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4579 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4582 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4583 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4584 check for the latter with gimple_call_builtin_p. Do not handle
4585 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4587 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4589 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4592 2017-11-23 Julia Koval <julia.koval@intel.com>
4594 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4595 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4596 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4597 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4598 _mm512_maskz_expandloadu_epi16): New intrinsics.
4599 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4600 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4601 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4602 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4603 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4604 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4605 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4606 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4607 _mm256_maskz_expandloadu_epi8): New intrinsics.
4608 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4609 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4610 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4611 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4612 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4613 * config/i386/sse.md (VI248_VLBW): New iterator.
4614 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4616 2017-11-23 Julia Koval <julia.koval@intel.com>
4618 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4619 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4620 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4621 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4622 _mm512_mask_compressstoreu_epi16): New.
4623 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4624 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4625 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4626 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4627 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4628 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4629 _mm256_mask_compressstoreu_epi8): New.
4630 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4631 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4632 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4633 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4634 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4635 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4636 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4637 __builtin_ia32_compresshi128_mask,
4638 __builtin_ia32_compressstoreuqi512_mask,
4639 __builtin_ia32_compressstoreuhi512_mask,
4640 __builtin_ia32_compressstoreuqi256_mask,
4641 __builtin_ia32_compressstoreuqi128_mask,
4642 __builtin_ia32_compressstoreuhi256_mask,
4643 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4644 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4646 (ix86_expand_special_args_builtin): Handle new types.
4647 (s4fma_expand): Handle new builtin array.
4648 * config/i386/immintrin.h: Include new headers.
4649 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4650 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4652 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4654 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4655 fixes. Declare temp and g variables at the top in order to avoid
4656 {} in most of the cases.
4658 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4660 * match.pd (ptr-0): New transformation.
4662 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4664 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4665 (struct addr_mode_cost_table): New.
4666 (struct tune_params): Add field addr_mode_costs.
4667 * config/arm/arm.c (generic_addr_mode_costs): New.
4668 (arm_slowmul_tune): Initialise addr_mode_costs field.
4669 (arm_fastmul_tune): Likewise.
4670 (arm_strongarm_tune): Likewise.
4671 (arm_xscale_tune): Likewise.
4672 (arm_9e_tune): Likewise.
4673 (arm_marvell_pj4_tune): Likewise.
4674 (arm_v6t2_tune): Likewise.
4675 (arm_cortex_tune): Likewise.
4676 (arm_cortex_a8_tune): Likewise.
4677 (arm_cortex_a7_tune): Likewise.
4678 (arm_cortex_a15_tune): Likewise.
4679 (arm_cortex_a35_tune): Likewise.
4680 (arm_cortex_a53_tune): Likewise.
4681 (arm_cortex_a57_tune): Likewise.
4682 (arm_exynosm1_tune): Likewise.
4683 (arm_xgene1_tune): Likewise.
4684 (arm_cortex_a5_tune): Likewise.
4685 (arm_cortex_a9_tune): Likewise.
4686 (arm_cortex_a12_tune): Likewise.
4687 (arm_cortex_a73_tune): Likewise.
4688 (arm_v7m_tune): Likewise.
4689 (arm_cortex_m7_tune): Likewise.
4690 (arm_v6m_tune): Likewise.
4691 (arm_fa726te_tune): Likewise.
4692 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4694 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4696 * config/arm/arm.c (arm_mem_costs): New function.
4697 (arm_rtx_costs_internal): Use arm_mem_costs.
4699 2017-11-23 Mark Wielaard <mark@klomp.org>
4701 * dwarf2out.c (init_sections_and_labels): Use generation to create
4702 unique ranges_section_label and ranges_base_label. Return generation.
4703 (output_rnglists): Add generation argument. Use generation to create
4704 unique ranges labels.
4705 (dwarf2out_finish): Get generation from init_sections_and_labels
4706 and pass generation to output_rnglists.
4708 2017-11-23 Mike Stump <mikestump@comcast.net>
4709 Eric Botcazou <ebotcazou@adacore.com>
4711 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4712 * cfgloop.h (struct loop): Add unroll field.
4713 * function.h (struct function): Add has_unroll bitfield.
4714 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4715 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4716 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4717 (pass_rtl_unroll_loops::gate): Likewise.
4718 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4719 for which loop->unroll==1.
4720 (decide_unroll_constant_iterations): Use note for consistency and
4721 take loop->unroll into account. Return early if loop->unroll is set.
4722 Fix thinko in existing test.
4723 (decide_unroll_runtime_iterations): Use note for consistency and
4724 take loop->unroll into account.
4725 (decide_unroll_stupid): Likewise.
4726 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4727 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4728 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4730 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4731 (print_loop): Print loop->unroll if set.
4732 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4733 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4734 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4736 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4737 loop->unroll is set and smaller than the trip count. Otherwise bypass
4738 entirely the heuristics if loop->unroll is set. Remove dead note.
4739 Fix off-by-one bug in other note.
4740 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4741 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4743 (tree_unroll_loops_completely): Make static.
4744 (pass_complete_unroll::execute): Use correct type for variable.
4745 (pass_complete_unrolli::execute): Fix formatting.
4746 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4748 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4750 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4751 enable when TARGET_PREFER_AVX128 is set.
4753 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4755 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4757 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4758 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4760 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4762 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4763 frequencies. Fix estimation of aggregate parameters.
4765 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4767 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4770 2017-11-23 Tom de Vries <tom@codesourcery.com>
4772 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4774 2017-11-23 Tom de Vries <tom@codesourcery.com>
4776 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4779 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4782 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4783 sibcall_value_pcrel_fdpic): Use local variable instead of
4785 (calli_tbr_rel): Add missing operand 2.
4786 (call_valuei_tbr_rel): Add missing operand 3.
4788 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4791 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4792 bitpos/bitsize covering the whole destination, use store_expr only if
4793 the complex mode is the same. Otherwise, use expand_normal and if
4794 it returns CONCAT, subreg each part separately instead of trying to
4795 subreg the whole result.
4797 2017-11-23 Richard Biener <rguenther@suse.de>
4799 PR tree-optimization/23094
4800 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4801 come here from walking over backedges in the first iteration.
4802 (vn_reference_lookup_3): Skip clobbers that store the same value.
4804 2017-11-23 Richard Biener <rguenther@suse.de>
4806 PR tree-optimization/81403
4807 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4808 a block we need a leader relative to.
4809 (phi_translate_1): For nary processing require a leader from
4810 get_representative_for given we run expression simplification
4811 using match-and-simplify. Remove previous fix.
4813 2017-11-22 Jeff Law <law@redhat.com>
4815 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4816 Use new method allocate_value_range rather than accessing the
4817 vrp_value_range_pool data member directly.
4818 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4819 to use extract_range_from_stmt method to avoid need for
4820 extract_range_from_assignment method.
4821 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4822 method rather than setting values_propgated data member directly.
4823 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4824 and values propagated data members and extract_range_from_assignment
4825 method. Reorder private data members to conform to standards.
4826 Add new methods set_lattice_propagation_complete and
4827 allocate_value_range.
4829 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4831 PR rtl-optimization/83030
4832 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4833 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4834 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4835 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4836 CROSSING_JUMP_P flag is already set before setting it.
4837 * cfgrtl.c (fixup_partition_crossing): Likewise.
4838 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4841 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4843 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4844 Handle the case where both arguments are using gen_const_vec_series.
4846 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4849 * pretty-print.c (pp_format): Move quoting implementation to
4850 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4851 to pass address of "quote" local.
4852 (pp_begin_quote): New function.
4853 (pp_end_quote): New function.
4854 * pretty-print.h (printer_fn): Convert penultimate param from bool
4856 (pp_begin_quote): New decl.
4857 (pp_end_quote): New decl.
4858 * tree-diagnostic.c (default_tree_printer): Convert penultimate
4859 param from bool to bool *.
4860 * tree-diagnostic.h (default_tree_printer): Likewise.
4862 2017-11-22 Jeff Law <law@redhat.com>
4864 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
4865 Initialize vr_values.
4866 (evrp_range_analyzer::try_find_new_range): Call methods attached to
4867 vr_values via vr_values class instance rather than delegators.
4868 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
4869 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4870 (evrp_range_analyzer::push_value_range): Likewise.
4871 (evrp_range_analyzer::pop_value_range): Likewise.
4872 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
4873 most delegators. Those remaining are exposed as public interfaces.
4874 Make vr_values a pointer and private.
4875 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
4877 (evrp_range_analyzer::get_vr_value): New method.
4878 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
4879 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
4880 (class evrp_dom_walker): Attach evrp_folder class, initialize
4881 it in the ctor. Remove temporary delegators.
4882 (evrp_dom_walker::before_dom_children): Call methods in attached
4883 evrp_range_analyzer class via class instance pointer. Use
4884 free value_range_constant_singleton to remove need for
4885 op_with_constant_singleton_value delegator method. Do not
4886 create a vrp_prop class instance for every call! Narrow
4887 scope of a couple variables.
4888 (evrp_dom_walker::cleanup): Call methods in attached
4889 evrp_range_analyzer class via class instance pointer.
4890 * vr-values.h (class vr_values): Privatize many methods and
4893 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4895 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
4896 * cfgexpand.c (expand_debug_expr): Remove handling for them.
4897 * expr.c (expand_expr_real_2): Likewise.
4898 * fold-const.c (const_unop): Likewise.
4899 * optabs-tree.c (optab_for_tree_code): Likewise.
4900 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
4901 * tree-inline.c (estimate_operator_cost): Likewise.
4902 * tree-pretty-print.c (dump_generic_node): Likewise.
4903 (op_code_prio): Likewise.
4904 (op_symbol_code): Likewise.
4905 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
4906 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
4907 * internal-fn.c (direct_internal_fn_optab): New function.
4908 (direct_internal_fn_array, direct_internal_fn_supported_p
4909 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
4910 * fold-const-call.c (fold_const_reduction): New function.
4911 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
4913 * tree-vect-loop.c: Include internal-fn.h.
4914 (reduction_code_for_scalar_code): Rename to...
4915 (reduction_fn_for_scalar_code): ...this and return an internal
4917 (vect_model_reduction_cost): Take an internal_fn rather than
4919 (vect_create_epilog_for_reduction): Likewise. Build calls rather
4921 (vectorizable_reduction): Use internal functions rather than tree
4922 codes for the reduction operation. Update calls to the functions
4924 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
4925 Use calls to internal functions rather than REDUC tree codes.
4926 * config/aarch64/aarch64-simd.md: Update comment accordingly.
4928 2017-11-22 Olivier Hainque <hainque@adacore.com>
4930 * config/vxworks.c (vxworks_override_options): Pick default
4931 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
4932 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
4933 DWARF_GNAT_ENCODINGS_DEFAULT.
4934 * config/vxworksae.h: Likewise.
4936 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
4938 PR tree-optimization/83104
4939 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
4942 2017-11-22 Marek Polacek <polacek@redhat.com>
4943 H.J. Lu <hongjiu.lu@intel.com>
4944 Jason Merrill <jason@redhat.com>
4949 * calls.c (initialize_argument_information): Call
4950 warn_parameter_passing_abi target hook.
4951 (store_one_arg): Use 0 for empty record size. Don't push 0 size
4952 argument onto stack.
4953 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
4954 * common.opt: Update -fabi-version description.
4955 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
4956 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
4958 (ix86_is_empty_record): New function.
4959 (ix86_warn_parameter_passing_abi): New function.
4960 (TARGET_EMPTY_RECORD_P): Redefine.
4961 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
4962 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
4963 * doc/tm.texi: Regenerated.
4964 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
4965 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
4966 * dwarf2out.c (get_ultimate_context): Move to tree.c.
4967 * explow.c (hard_function_value): Call arg_int_size_in_bytes
4968 instead of int_size_in_bytes.
4969 * expr.c (copy_blkmode_to_reg): Likewise.
4970 * function.c (aggregate_value_p): Return 0 for empty types.
4971 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
4973 (locate_and_pad_parm): Call arg size_in_bytes instead
4975 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
4976 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
4977 * target.def (empty_record_p, warn_parameter_passing_abi): New target
4979 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
4980 (std_gimplify_va_arg_expr): Skip empty records. Call
4981 arg_size_in_bytes instead size_in_bytes.
4982 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
4983 * tree-core.h (tree_type_common): Add empty_flag.
4984 (tree_decl_common): Update comments.
4985 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
4987 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4988 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
4990 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4991 * tree.c (default_is_empty_type): New function.
4992 (default_is_empty_record): New function.
4993 (arg_int_size_in_bytes): New function.
4994 (arg_size_in_bytes): New function.
4995 (get_ultimate_context): New function.
4996 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
4997 TRANSLATION_UNIT_WARN_EMPTY_P.
4998 (default_is_empty_record, arg_int_size_in_bytes,
4999 arg_size_in_bytes, get_ultimate_context): Declare.
5001 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5003 * config/arm/arm.c (cmse_clear_registers): New function.
5004 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5005 code by call to cmse_clear_registers.
5006 (cmse_nonsecure_entry_clear_before_return): Likewise.
5008 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5010 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5011 (vdot_s32, vdotq_s32): New.
5012 (vdot_lane_u32, vdotq_lane_u32): New.
5013 (vdot_lane_s32, vdotq_lane_s32): New.
5016 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5019 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5020 for unsigned values with fewer HWIs than the precision.
5021 (test_overflow): New function.
5022 (wide_int_cc_tests): Call it.
5024 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5026 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5027 has the same mode class as Pmode.
5029 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5031 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5032 padding_bits_to_clear_ptr.
5033 (cmse_nonsecure_entry_clear_before_return): Likewise.
5035 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5037 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5038 auto_sbitap instead of integer bitfield to control register needing
5041 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5043 PR tree-optimization/83044
5044 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5045 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5046 that need the upper bound. Subtract offset from
5047 get_addr_base_and_unit_offset only if positive and subtract it
5048 before division by eltsize rather than after it.
5051 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5052 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5053 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5057 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5058 expand_binop with *mul_widen_optab, make sure at least one of the
5059 operands doesn't have VOIDmode.
5062 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5064 PR rtl-optimization/82044
5065 PR tree-optimization/82042
5066 * dse.c (record_store): Check for overflow.
5067 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5068 add_wild_read instead of clear_rhs_from_active_local_stores on
5069 overflow and log it into dump_file.
5071 2017-11-22 Richard Biener <rguenther@suse.de>
5073 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5074 fake edges to exit when looking for a place to insert.
5075 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5077 (insert_into_preds_of_block): Commit edge insertion immediately,
5078 assert that doesn't require new BBs.
5079 (fini_pre): Release expressions.
5080 (pass_pre::execute): Shuffle things around a bit, if the fn
5081 is too large do not compute AVAIL either as this is really the
5084 2017-11-22 Richard Biener <rguenther@suse.de>
5086 PR tree-optimization/83089
5087 * tree-if-conv.c (pass_if_conversion::execute): If anything
5088 changed reset SCEV and free the number of iteration estimates.
5090 2017-11-21 Martin Sebor <msebor@redhat.com>
5092 PR tree-optimization/82945
5093 * calls.h (warn_nonstring_bound): Remove unused function.
5095 2017-11-21 Martin Sebor <msebor@redhat.com>
5097 PR tree-optimization/82945
5098 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5099 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5100 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5102 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5103 * calls.h (get_attr_nonstring_decl): Declare new function.
5104 * doc/extend.texi (attribute nonstring): Update.
5105 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5106 get_attr_nonstring_decl and handle it.
5107 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5108 detection of nul-termination.
5109 (strlen_to_stridx): Change to a pointer.
5110 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5111 (pass_strlen::execute): Same.
5113 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5115 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5116 for the new option -mprefer-vector-width=[none|128|256|512].
5117 * config/i386/i386.c (ix86_target_string): remove old style options
5118 -mprefer-avx256 and make -mprefer-avx128 as alias.
5119 (ix86_option_override_internal): Apply defaults for the
5120 -mprefer-vector-width=[128|256] option.
5121 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5122 Implement macros to work with -mprefer-vector-width=.
5123 * config/i386/i386.opt: Implemented option
5124 -mprefer-vector-width=[none|128|256|512].
5125 * doc/invoke.texi: Documentation for
5126 -mprefer-vector-width=[none|128|256|512].
5128 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5130 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5132 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5134 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5136 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5138 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5140 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5141 * expr.c (expand_expr_real_2): Likewise.
5142 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5143 fold_binary_loc): Likewise.
5144 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5145 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5146 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5147 MINUS_EXPR transformations.
5148 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5149 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5150 * tree-inline.c (estimate_operator_cost): Likewise.
5151 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5152 op_symbol_code): Likewise.
5153 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5154 * vr-values.c (extract_range_from_binary_expr): Likewise.
5155 * varasm.c (initializer_constant_valid_p_1): Likewise.
5156 * tree.def: New tree code POINTER_DIFF_EXPR.
5158 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5160 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5161 integer suffix to movbe mnemonic.
5162 (*bswaphi2_movbe): Ditto.
5163 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5165 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5168 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5169 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5170 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5172 2017-11-21 Martin Liska <mliska@suse.cz>
5174 * tree-inline.c (expand_call_inline): Remove not needed
5177 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5178 Jakub Jelinek <jakub@redhat.com>
5181 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5182 Remove static keyword from f variable.
5184 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5186 PR tree-optimization/83086
5187 * gimple-ssa-store-merging.c
5188 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5189 rather than n.base_addr.
5191 2017-11-21 Martin Liska <mliska@suse.cz>
5193 PR rtl-optimization/82044
5194 PR tree-optimization/82042
5195 * dse.c (check_mem_read_rtx): Check for overflow.
5197 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5199 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5202 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5204 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5205 and its corresponding call arg location note.
5207 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5208 Andrew Burgess <andrew.burgess@embecosm.com>
5210 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5212 (arc_return_slot_offset): Likewise.
5213 (arc_eh_return_address_location): New declaration.
5214 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5215 (MUST_SAVE_REGISTER): Add exception handler case.
5216 (MUST_SAVE_RETURN_ADDR): Likewise.
5217 (arc_frame_pointer_required): Likewise.
5218 (arc_frame_pointer_needed): New function.
5219 (arc_compute_frame_size): Changed.
5220 (arc_expand_prologue): Likewise.
5221 (arc_expand_epilogue): Likewise.
5222 (arc_initial_elimination_offset): Likewise.
5223 (arc_return_slot_offset): Delete.
5224 (arc_eh_return_address_location): New function.
5225 (arc_builtin_setjmp_frame_value): Likewise.
5226 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5227 (EH_RETURN_STACKADJ_RTX): Define.
5228 (EH_RETURN_HANDLER_RTX): Likewise.
5229 * config/arc/arc.md (eh_return): Delete.
5231 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5233 * print-tree.h (debug_vec_tree): Remove prototype.
5234 * gdbinit.in (pvt): Remove macro.
5236 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5238 PR tree-optimization/83047
5239 * gimple-ssa-store-merging.c
5240 (imm_store_chain_info::output_merged_store): If the loads with the
5241 same vuse are in different basic blocks, for load_gsi pick a load
5242 location that is dominated by the other loads.
5245 * config/i386/i386.c (ix86_memmodel_check): Start
5246 -Winvalid-memory-model diagnostics with lowercase letter.
5249 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5250 set in_cold_section_p to the partition containing loc_list->first.
5251 When seeing loc_list->last_before_switch node, update secname and
5252 perform range_across_switch second partition handling only after that.
5255 * run-rtl-passes.c: Include debug.h.
5256 (run_rtl_passes): Call debug_hooks->assembly_start.
5257 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5261 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5262 OPTAB_DIRECT in calls to expand_simple_binop.
5264 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5267 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5269 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5272 * spellcheck.h (best_match::blithely_get_best_candidate): New
5275 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5277 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5278 with lower case letter.
5280 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5282 * config/i386/i386.md (bswaphi2): New expander.
5283 (*bswaphi2_movbe): New insn pattern.
5284 (bswaphi -> rorhi pepehole2): New peephole pattern.
5286 2017-11-20 Jeff Law <law@redhat.com>
5288 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5289 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5290 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5291 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5292 gimple-ssa-evrp-analyze.h.
5294 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5295 set BB_VISITED here.
5296 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5298 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5300 * tree-predcom.c: Add general comment on Store-Store chains.
5301 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5303 (get_chain_last_ref_at): Rename into...
5304 (get_chain_last_write_at): ...this.
5305 (get_chain_last_write_before_load): New function.
5306 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5307 CT_STORE_STORE when write reference is added.
5308 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5309 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5310 (initialize_root_vars_store_elim_1): Ditto.
5311 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5312 definition is created.
5313 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5314 chain by replacing it with dominant stored value.
5316 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5318 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5320 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5322 * vr-values.c (extract_range_from_binary_expr): Use a full range
5325 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5327 * config/arm/arm.md (R4_REGNUM): Define constant.
5328 (nonsecure_call_internal): Remove r4 clobber.
5329 (nonsecure_call_value_internal): Likewise.
5330 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5331 clobber and resequence match_operands.
5332 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5333 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5334 (nonsecure_call_value_reg_thumb2): Likewise.
5336 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5338 PR tree-optimization/78821
5339 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5340 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5341 to the address of the base rather than the base itself.
5342 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5343 (find_bswap_or_nop_finalize): New function.
5344 (find_bswap_or_nop): Use it.
5345 (bswap_replace): Return a tree rather than bool, change first
5346 argument from gimple * to gimple_stmt_iterator, allow inserting
5347 into an empty sequence, allow ins_stmt to be NULL - then emit
5348 all stmts into gsi. Fix up MEM_REF address gimplification.
5349 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5350 (struct store_immediate_info): Add N and INS_STMT non-static
5352 (store_immediate_info::store_immediate_info): Initialize them
5353 from newly added ctor args.
5354 (merged_store_group::apply_stores): Formatting fixes. Sort by
5356 (stmts_may_clobber_ref_p): For stores call also
5357 refs_anti_dependent_p.
5358 (gather_bswap_load_refs): New function.
5359 (imm_store_chain_info::try_coalesce_bswap): New method.
5360 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5361 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5362 (imm_store_chain_info::output_merged_store): Fail if number of
5363 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5364 and NOP_EXPR rhs_code.
5365 (pass_store_merging::process_store): Compute n and ins_stmt, if
5366 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5367 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5369 (pass_store_merging::execute): Calculate dominators.
5371 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5372 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5373 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5374 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5375 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5376 class pass_optimize_bswap, bswap_replace,
5377 pass_optimize_bswap::execute): Moved to ...
5378 * gimple-ssa-store-merging.c: ... this file.
5379 Include optabs-tree.h.
5380 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5381 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5382 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5383 anonymous namespace, remove static keywords.
5384 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5385 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5387 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5390 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5392 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5394 * vec.h (debug_helper): New function.
5395 (DEFINE_DEBUG_VEC): New macro.
5396 * hash-set.h (debug_helper): New function.
5397 (DEFINE_DEBUG_HASH_SET): New macro.
5398 * cfg.c (debug_slim (edge)): New function.
5399 Call DEFINE_DEBUG_VEC for edges.
5400 Call DEFINE_DEBUG_HASH_SET for edges.
5401 * cfghooks.c (debug_slim (basic_block)): New function.
5402 Call DEFINE_DEBUG_VEC for basic blocks.
5403 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5404 * print-tree.c (debug_slim): New function to handle trees.
5405 Call DEFINE_DEBUG_VEC for trees.
5406 Call DEFINE_DEBUG_HASH_SET for trees.
5407 (debug (vec<tree, va_gc>) &): Remove.
5408 (debug (<vec<tree, va_gc>) *): Remove.
5409 * print-rtl.c (debug_slim): New function to handle const_rtx.
5410 Call DEFINE_DEBUG_VEC for rtx_def.
5411 Call DEFINE_DEBUG_VEC for rtx_insn.
5412 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5413 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5414 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5415 (debug (vec<rtx_insn *> *ptr): Remove.
5416 (debug_insn_vector): Remove.
5417 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5419 2017-11-20 Tom de Vries <tom@codesourcery.com>
5421 PR rtl-optimization/82020
5422 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5423 IF_THEN_ELSE condition.
5425 2017-11-19 Jeff Law <law@redhat.com>
5427 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5428 of degenerates resulting from ignoring an edge.
5430 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5433 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5435 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5438 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5439 for uninitialied counts.
5441 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5444 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5446 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5449 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5451 2017-11-19 Tom de Vries <tom@codesourcery.com>
5453 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5456 2017-11-19 Tom de Vries <tom@codesourcery.com>
5458 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5461 2017-11-19 Tom de Vries <tom@codesourcery.com>
5463 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5466 2017-11-19 Tom de Vries <tom@codesourcery.com>
5468 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5470 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5472 2017-11-19 Tom de Vries <tom@codesourcery.com>
5474 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5476 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5477 semicolon after MCORE_EXPORT_NAME call.
5479 2017-11-19 Tom de Vries <tom@codesourcery.com>
5482 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5483 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5485 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5487 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5488 (cgraph_edge::clone): Cleanup updating of profile.
5489 * ipa-cp.c (update_profiling_info): Likewise.
5490 * ipa-inline-transform.c (inline_transform): Likewise.
5491 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5492 * ipa-split.c (execute_split_functions): Do not split when function
5494 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5495 * profile-count.c (profile_count::dump): Add global0.
5496 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5498 (profile_count::to_sreal_scale): Likewise.
5499 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5500 (profile_count::combine_with_ipa_count): New function.
5501 * profile-count.h (profile_guessed_global0adjusted): New.
5502 (profile_count::adjusted_zero): New.
5503 (profile_count::global0adjusted): New.
5504 (profile_count::combine_with_ipa_count): New.
5505 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5506 correct profile of return block of split functions.
5507 (copy_cfg_body): Remove unused profile_count.
5508 (copy_body): Likewise.
5509 (expand_call_inline): Update.
5510 (tree_function_versioning): Update.
5512 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5514 * hash-set.h (hash_set::empty): New.
5515 * tree-ssa-threadbackward.h: Delete.
5516 * tree-ssa-threadbackward.c (class thread_jumps): New.
5517 Move max_threaded_paths into class.
5518 (fsm_find_thread_path): Remove arguments that are now in class.
5519 (profitable_jump_thread_path): Rename to...
5520 (thread_jumps::profitable_jump_thread_path): ...this.
5521 (convert_and_register_jump_thread_path): Rename to...
5522 (thread_jumps::convert_and_register_current_path): ...this.
5523 (check_subpath_and_update_thread_path): Rename to...
5524 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5525 (register_jump_thread_path_if_profitable): Rename to...
5526 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5527 (handle_phi): Rename to...
5528 (thread_jumps::handle_phi): ...this.
5529 (handle_assignment): Rename to...
5530 (thread_jumps::handle_assignment): ...this.
5531 (fsm_find_control_statement_thread_paths): Rename to...
5532 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5533 (find_jump_threads_backwards): Rename to...
5534 (thread_jumps::find_jump_threads_backwards): ...this.
5535 Initialize path local data.
5536 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5537 from within thread_jumps class.
5538 (pass_early_thread_jumps::execute): Same.
5540 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5542 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5544 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5546 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5547 to canonical location.
5549 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5552 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5554 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5556 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5558 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5559 rcrt1.o%s/grcrt1.o%s for -static-pie.
5561 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5563 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5564 ix86_shift_rotate_cost): Break out from ...
5565 (ix86_rtx_costs): ... here.
5566 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5569 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5571 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5572 when entry block was promoted unlikely.
5573 (estimate_bb_frequencies): Increase frequency scale.
5574 * profile-count.h (profile_count): Export precision info.
5576 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5578 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5579 disturbin profile of entry block.
5581 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5583 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5586 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5588 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5589 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5591 2017-11-17 Jeff Law <law@redhat.com>
5593 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5594 from evrp_dom_walker class. Various methods moved into new class.
5595 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5596 (evrp_range_analyzer::enter): New method.
5597 (evrp_range_analyzer::leave): New method.
5598 (evrp_dom_walker): Remove delegators no longer needed by this class.
5599 Replace vr_values data member with evrp_range_analyzer
5601 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5602 method extracted from evrp_dom_walker::before_dom_children.
5603 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5604 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5606 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5607 Add private copy constructor and move assignment operators.
5608 Privatize methods and class data where trivially possible.
5609 (evrp_dom_walker::cleanup): New function, extracted from
5610 execute_early_vrp. Simplify access to class data.
5612 * vr-values.h (get_output_for_vrp): Prototype.
5613 * vr-values.c (get_output_for_vrp): New function extracted from
5614 vrp_visit_assignment_or_call and extract_range_from_stmt.
5615 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5617 2017-11-17 Luis Machado <luis.machado@linaro.org>
5619 * config/aarch64/aarch64.c
5620 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5621 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5622 tune_params::AUTOPREFETCHER_WEAK.
5624 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5627 * config/arm/arm.c (arm_valid_target_attribute_rec):
5628 Parse "arch=" and "+<ext>".
5629 (arm_valid_target_attribute_tree): Re-init global options.
5630 (arm_option_override): Make non-static.
5631 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5632 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5633 (__ARM_FEATURE_CRC32): Support undef.
5634 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5635 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5637 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5639 * gdbinit.in (break-on-diagnostic): New command.
5641 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5643 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5644 return to indirect jump for EH return if control-flow
5645 protection is enabled. Change explicit 'false' argument in
5646 pro_epilogue_adjust_stack with a value of
5648 * config/i386/i386.md (simple_return_indirect_internal):
5649 Remove SImode restriction to support 64-bit.
5651 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5653 * combine.c (added_notes_insn): New.
5654 (try_combine): Handle added_notes_insn like added_links_insn.
5655 Rewrite return value code.
5656 (distribute_notes): Set added_notes_insn to the earliest insn we added
5659 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5661 PR rtl-optimization/82621
5662 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5663 dest of one of those SETs is unused.
5665 2017-11-17 Richard Biener <rguenther@suse.de>
5668 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5669 * tree-pretty-print.c (dump_generic_node): Handle
5670 annot_expr_parallel_kind.
5671 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5672 * gimplify.c (gimple_boolify): Likewise.
5674 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5676 * config.gcc (extra_headers): Add cet.h for x86 targets.
5677 * config/i386/cet.h: New file.
5678 * doc/install.texi: Add --enable-cet/--disable-cet.
5680 2017-11-17 Richard Biener <rguenther@suse.de>
5682 PR tree-optimization/83017
5683 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5684 (gen_parallel_loop): Properly count iterations.
5685 (parallelize_loops): Handle loop->can_be_parallel independent
5686 of flag_loop_parallelize_all. Make static profitability test match
5688 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5689 * invoke.texi (parloops-min-per-thread): Document.
5691 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5693 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5694 upstreaming review comments.
5696 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5698 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5699 with fast unaligned access.
5700 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5702 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5704 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5705 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5707 2017-11-17 Richard Biener <rguenther@suse.de>
5709 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5710 folding of references.
5712 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5715 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5716 of replacing call to strncmp with corresponding call to strcmp when
5719 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5721 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5722 option prefer-avx256 for skylake-avx512 configuration.
5723 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5724 (get_builtin_code_for_version): Ditto.
5726 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5727 Monk Chiang <sh.chiang04@gmail.com>
5729 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5730 (FIXED_REGISTERS): Reserve more register numbers.
5731 (CALL_USED_REGISTERS): Likewise.
5732 (REG_ALLOC_ORDER): Likewise.
5733 (REG_CLASS_CONTENTS): Likewise.
5734 (REGISTER_NAMES): Likewise.
5736 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5737 Kito Cheng <kito.cheng@gmail.com>
5739 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5741 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5743 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5745 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5746 parameter list for vec_splats.
5748 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5750 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5752 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5754 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5757 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5758 with the required version of automake 1.11.6.
5760 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5762 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5764 (DU_C2_3_power9): Correct reservation combinations.
5765 (FP_DIV_power9, VEC_DIV_power9): New.
5766 (power9-alu): Split out rotate/shift...
5767 (power9-rot): ...to here, correct dispatch resource.
5768 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5770 (power9-fp): Correct latency.
5771 (power9-sdiv): Add div/sqrt resource.
5772 (power9-ddiv): Correct latency, add div/sqrt resource.
5773 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5774 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5776 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5778 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5780 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5781 switch statement mapping KF built-ins to TF built-ins if we don't
5782 have the proper ISA 3.0 assembler support.
5784 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5786 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5787 (lower_emutls_function_body): Do not compute it.
5789 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5791 * ipa-split.c (split_bb_info): Turn time to sreal.
5792 (split_point): Likewise.
5793 (dump_split_point): Likewise.
5794 (fine_split_points): Likewise.
5795 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5797 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5799 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5802 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5804 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5806 (expensive_function_p): Remove useless assert.
5807 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5809 2017-11-16 Martin Sebor <msebor@redhat.com>
5811 PR tree-optimization/82588
5812 PR tree-optimization/82583
5813 * tree-vrp.c (check_array_ref): Handle flexible array members,
5814 string literals, and inner indices.
5815 (search_for_addr_array): Add detail to diagnostics.
5817 2017-11-16 Nathan Sidwell <nathan@acm.org>
5821 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5822 (SET_DECL_ASSEMBLER_NAME): Forward to
5823 overwrite_decl_assembler_name.
5824 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5825 (overwrite_decl_assembler_name): Declare.
5826 * tree.c (overwrite_decl_assembler_name): New.
5827 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5828 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5829 (LANG_HOOKS_INITIALIZER): Add it.
5830 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5831 * langhooks.c (lhd_set_decl_assembler_name): Use
5832 SET_DECL_ASSEMBLER_NAME.
5833 (lhd_overwrite_decl_assembler_name): Default implementation.
5835 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5836 Jackson Woodruff <jackson.woodruff@arm.com>
5838 PR tree-optimization/71026
5839 * match.pd: Canonicalize constant multiplies in division.
5841 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5843 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5844 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5845 Remove OPT_fomit_frame_pointer entry.
5846 * common/config/arc/arc-common.c: Likewise.
5847 * common/config/arm/arm-common.c: Likewise.
5848 * common/config/avr/avr-common.c: Likewise.
5849 * common/config/c6x/c6x-common.c: Likewise.
5850 * common/config/cr16/cr16-common.c: Likewise.
5851 * common/config/cris/cris-common.c: Likewise.
5852 * common/config/epiphany/epiphany-common.c: Likewise.
5853 * common/config/fr30/fr30-common.c: Likewise.
5854 * common/config/frv/frv-common.c: Likewise.
5855 * common/config/ia64/ia64-common.c: Likewise.
5856 * common/config/iq2000/iq2000-common.c: Likewise.
5857 * common/config/lm32/lm32-common.c: Likewise.
5858 * common/config/m32r/m32r-common.c: Likewise.
5859 * common/config/mcore/mcore-common.c: Likewise.
5860 * common/config/microblaze/microblaze-common.c: Likewise.
5861 * common/config/mips/mips-common.c: Likewise.
5862 * common/config/mmix/mmix-common.c: Likewise.
5863 * common/config/mn10300/mn10300-common.c: Likewise.
5864 * common/config/nios2/nios2-common.c: Likewise.
5865 * common/config/pa/pa-common.c: Likewise.
5866 * common/config/pdp11/pdp11-common.c: Likewise.
5867 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
5868 * common/config/riscv/riscv-common.c: Likewise.
5869 * common/config/rs6000/rs6000-common.c: Likewise.
5870 * common/config/rx/rx-common.c: Likewise.
5871 * common/config/s390/s390-common.c: Likewise.
5872 * common/config/sh/sh-common.c: Likewise.
5873 * common/config/sparc/sparc-common.c: Likewise.
5874 * common/config/tilegx/tilegx-common.c: Likewise.
5875 * common/config/tilepro/tilepro-common.c: Likewise.
5876 * common/config/v850/v850-common.c: Likewise.
5877 * common/config/visium/visium-common.c: Likewise.
5878 * common/config/xstormy16/xstormy16-common.c: Likewise.
5879 * common/config/xtensa/xtensa-common.c: Likewise.
5880 * invoke.texi (-fomit-frame-pointer): Update documentation.
5882 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5884 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
5886 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5888 * cfg.c (scale_bbs_frequencies_int,
5889 cale_bbs_frequencies_gcov_type): Remove.
5890 * cfg.h (scale_bbs_frequencies_int,
5891 cale_bbs_frequencies_gcov_type): Remove.
5893 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5895 * tree-ssa-loop-manip.c
5896 (scale_dominated_blocks_in_loop): Update to profile counts.
5897 (tree_transform_and_unroll_loop): Likewise.
5899 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5901 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
5902 scale_bbs_frequencies_int.
5904 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5906 * final.c (compute_alignments): Use counts rather than frequencies.
5908 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5910 * cfgloopanal.c: Include sreal.h
5911 (average_num_loop_insns): Use counts and sreal for accounting.
5913 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5915 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
5918 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5919 Kito Cheng <kito.cheng@gmail.com>
5921 * config/nds32/constraints.md: Provide more constraints.
5922 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
5923 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
5924 support constraints usage.
5926 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5928 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
5930 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5931 Kito Cheng <kito.cheng@gmail.com>
5933 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
5934 * config/nds32/nds32.opt: Refine the layout.
5935 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
5936 TARGET_EXT_STRING): Support new options.
5937 * config/nds32/nds32.h: Likewise.
5938 * config/nds32/nds32.md: Likewise.
5939 * config/nds32/nds32-predicates.c: Likewise.
5940 * config/nds32/constraints.md: Likewise.
5941 * common/config/nds32/nds32-common.c: Likewise.
5943 2017-11-16 Julia Koval <julia.koval@intel.com>
5946 * config/i386/gfniintrin.h: Add sse check.
5947 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
5949 2017-11-16 Julia Koval <julia.koval@intel.com>
5951 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
5952 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
5953 (ix86_handle_option): Handle -mavx512vbmi2.
5954 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
5955 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
5956 * config/i386/i386-c.c (__AVX512VBMI2__): New.
5957 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
5958 (ix86_valid_target_attribute_inner_p): Ditto.
5959 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
5960 * config/i386/i386.opt (mavx512vbmi2): New option.
5961 * doc/invoke.texi: Add new option.
5963 2017-11-16 Julia Koval <julia.koval@intel.com>
5965 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
5966 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
5967 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
5968 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
5969 _mm512_gf2p8mul_epi8): New intrinsics.
5970 * config/i386/i386-builtin-types.def
5971 (V64QI_FTYPE_V64QI_V64QI): New type.
5972 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
5973 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
5974 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
5975 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
5976 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
5977 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
5979 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
5981 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
5982 explicitly as a stream of bytes.
5984 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5986 * config/rs6000/altivec.h (vec_xst_be): New #define.
5987 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
5988 and externalize from *altivec_vperm_<mode>_internal.
5989 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
5991 (XL_BE_V8HI): Likewise.
5992 (XL_BE_V4SI): Likewise.
5993 (XL_BE_V4SI): Likewise.
5994 (XL_BE_V2DI): Likewise.
5995 (XL_BE_V4SF): Likewise.
5996 (XL_BE_V2DF): Likewise.
5997 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
5998 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
5999 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6000 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6001 VSX_BUILTIN_VEC_XST_BE.
6002 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6003 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6005 (altivec_init_builtins): Replace conditional calls to def_builtin
6006 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6007 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6008 with unconditional calls. Remove calls to def_builtin for
6009 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6010 __builtin_vec_xst_be.
6011 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6012 to define_expand, and add alternate RTL generation for P8.
6013 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6014 vsx_ld_elemrev_v8hi.
6015 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6016 add alternate RTL generation for P8.
6017 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6018 vsx_ld_elemrev_v16qi.
6019 (vsx_st_elemrev_v8hi): Convert define_insn
6020 to define_expand, and add alternate RTL generation for P8.
6021 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6022 vsx_st_elemrev_v8hi.
6023 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6024 add alternate RTL generation for P8.
6025 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6026 vsx_st_elemrev_v16qi.
6028 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6031 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6032 TARGET_AVX512ER check.
6033 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6034 neither -mzeroupper nor -mno-zeroupper is used and
6035 TARGET_EMIT_VZEROUPPER is set.
6036 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6037 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6039 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6041 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6042 folding of vector compares.
6043 (fold_build_vec_cmp): New helper function.
6044 (fold_compare_helper): New helper function.
6045 (builtin_function_type): Add compare builtins to the list of functions
6046 having unsigned arguments. Cosmetic updates to comment indentation.
6047 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6048 the not+eq combination.
6050 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6052 PR tree-optimization/82726
6053 PR tree-optimization/70754
6054 * tree-predcom.c (order_drefs_by_pos): New function.
6055 (combine_chains): Move code setting has_max_use_after to...
6056 (try_combine_chains): ...here. New parameter. Sort combined chains
6057 according to position information.
6058 (tree_predictive_commoning_loop): Update call to above function.
6059 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6061 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6063 PR tree-optimization/82726
6065 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6067 PR tree-optimization/70754
6068 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6069 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6070 combined stmt before it if not NULL.
6071 (combine_chains): Process refs reversely and compute dominance point
6075 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6077 PR tree-optimization/79663
6078 * tree-predcom.c (combine_chains): Process refs in reverse order
6079 only for ZERO length chains, and add explaining comment.
6081 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6083 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6084 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6085 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6087 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6089 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6091 2017-11-15 Martin Liska <mliska@suse.cz>
6093 * tree-cfg.c (pass_warn_function_return::execute):
6094 Compare warn_return_type for greater than zero.
6096 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6100 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6101 to return true on Xeon and not on Xeon Phi.
6102 (ix86_check_avx256_register): Changed to ...
6103 (ix86_check_avx_upper_register): ... this. Add extra check for
6104 VALID_AVX512F_REG_OR_XI_MODE.
6105 (ix86_avx_u128_mode_needed): Changed
6106 ix86_check_avx256_register to ix86_check_avx_upper_register.
6107 (ix86_check_avx256_stores): Changed to ...
6108 (ix86_check_avx_upper_stores): ... this. Changed
6109 ix86_check_avx256_register to ix86_check_avx_upper_register.
6110 (ix86_avx_u128_mode_after): Changed
6111 avx_reg256_found to avx_upper_reg_found. Changed
6112 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6113 (ix86_avx_u128_mode_entry): Changed
6114 ix86_check_avx256_register to ix86_check_avx_upper_register.
6115 (ix86_avx_u128_mode_exit): Ditto.
6116 * config/i386/i386.h: (host_detect_local_cpu): New define.
6118 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6120 * config/arm/xgene1.md (xgene1): Split into automatons
6121 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6122 (xgene1_f_load): Adjust reservations and/or types.
6123 (xgene1_f_store): Likewise.
6124 (xgene1_load_pair): Likewise.
6125 (xgene1_store_pair): Likewise.
6126 (xgene1_fp_load1): Likewise.
6127 (xgene1_load1): Likewise.
6128 (xgene1_store1): Likewise.
6129 (xgene1_move): Likewise.
6130 (xgene1_alu): Likewise.
6131 (xgene1_simd): Likewise.
6132 (xgene1_bfm): Likewise.
6133 (xgene1_neon_load1): Likewise.
6134 (xgene1_neon_store1): Likewise.
6135 (xgene1_neon_logic): Likewise.
6136 (xgene1_neon_st1): Likewise.
6137 (xgene1_neon_ld1r): Likewise.
6138 (xgene1_alu_cond): Added.
6139 (xgene1_shift_reg): Likwise.
6140 (xgene1_bfx): Likewise.
6141 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6143 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6146 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6148 (can_widen_mult_without_libcall): New function.
6149 (expand_mul_overflow): If only checking unsigned mul overflow,
6150 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6151 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6152 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6153 (expand_DIVMOD): Formatting fix.
6154 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6155 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6156 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6158 PR tree-optimization/82977
6159 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6160 constructed temporary to strlen_to_stridx.put.
6162 2017-11-15 Martin Liska <mliska@suse.cz>
6164 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6165 * configure: Regenerate.
6167 2017-11-15 Martin Liska <mliska@suse.cz>
6170 * config/sh/sh-mem.cc: Use proper probability for
6173 2017-11-14 Jeff Law <law@redhat.com>
6175 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6176 the red zone for stack_clash_protection_final_dynamic_probe targets
6177 when the total dynamic stack size is zero bytes.
6179 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6180 blocks is post order.
6182 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6184 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6185 * final.c (rest_of_clean_state): Set it for the
6186 -fcompare-debug dump.
6187 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6188 class when TDF_COMPARE_DEBUG is set.
6190 * dwarf2out.c (gen_producer_string): Discard
6193 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6196 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6197 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6198 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6199 (__TGMATH_REAL_2_3): Remove macros.
6200 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6201 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6203 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6205 (remquo): Define using __TGMATH_REAL_3.
6207 2017-11-14 Jeff Law <law@redhat.com>
6209 * vr-values.c: New file with contents extracted from tree-vrp.c.
6210 * Makefile.in (OBJS): Add vr-values.o
6211 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6212 (set_value_range, set_and_canonicalize_value_range): Likewise.
6213 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6214 (value_range_constant_singleton, symbolic_range_p): Likewise.
6215 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6216 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6217 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6218 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6219 (find_case_label_range, find_case_label_index): Likewise.
6220 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6221 (range_int_cst_singleton_p, value_inside_range): Likewise.
6222 (get_single_symbol): Likewise.
6223 (switch_update): Move structure definition here.
6224 (to_remove_edges, to_update_switch_stmts): Provide externs.
6225 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6226 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6227 (vrp_val_is_min, set_value_range): Likewise.
6228 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6229 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6230 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6231 (range_is_nonnull, range_int_cst_p): Likewwise.
6232 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6233 (get_single_symbol, operand_less_p): Likewise
6234 (compare_values_warnv, compare_values): Likewise.
6235 (value_inside_range, value_range_constant_singleton): Likewise.
6236 (zero_nonzero_bitgs_from_vr): Likewise.
6237 (extract_range_from_binary_expr_1): Likewise.
6238 (overflow_comparison_p): Likewise.
6239 (to_remove_edges, to_update_switch_stmts): Likewise.
6240 (find_case_label-index, find_case_label_range): Likewise.
6241 (switch_update, set_value_range_to_nonnegative): Remove.
6242 (set_value_range_to_truthvalue): Likewise.
6243 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6244 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6245 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6246 (find_case_label_ranges, test_for_singularity): Likewise.
6247 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6248 (x_vr_values): Move to its remaining use site.
6250 2017-11-10 Jeff Law <law@redhat.com>
6252 * vr-values.h (VR_INITIALIZER): Move #define here.
6253 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6254 * Makefile.in (OBJS): Add tree-evrp.o
6255 * tree-vrp.h (assert_info): Move structure definition here.
6256 (set_value_range_to_varying): Prototype.
6257 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6258 (infer_value_range, register_edge_assert_for): Likewise.
6259 (stmt_interesting_for_vrp): Likewise.
6260 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6261 (set_value_range_to_varying): No longer static.
6262 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6263 (infer_value_range, register_edge_assert_for): Likewise.
6265 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6267 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6268 generate the XXBRD instruction.
6270 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6271 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6272 as being compatible if -mabi=ieeelongdouble.
6273 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6274 to setup float128 built-ins with hardware support.
6275 (BU_FLOAT128_HW_2): Likewise.
6276 (BU_FLOAT128_HW_3): Likewise.
6277 (BU_FLOAT128_HW_VSX_1): Likewise.
6278 (BU_FLOAT128_HW_VSX_2): Likewise.
6279 (scalar_extract_expq): Change float128 built-in functions to
6280 accommodate having both KFmode and TFmode functions. Use the
6281 KFmode variant as the default.
6282 (scalar_extract_sigq): Likewise.
6283 (scalar_test_neg_qp): Likewise.
6284 (scalar_insert_exp_q): Likewise.
6285 (scalar_insert_exp_qp): Likewise.
6286 (scalar_test_data_class_qp): Likewise.
6287 (sqrtf128_round_to_odd): Delete processing the round to odd
6288 built-in functions as special built-in functions, and define them
6289 as float128 built-ins. Use the KFmode variant as the default.
6290 (truncf128_round_to_odd): Likewise.
6291 (addf128_round_to_odd): Likewise.
6292 (subf128_round_to_odd): Likewise.
6293 (mulf128_round_to_odd): Likewise.
6294 (divf128_round_to_odd): Likewise.
6295 (fmaf128_round_to_odd): Likewise.
6296 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6297 support for KFmode and TFmode xststdcqp calls.
6298 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6299 point, switch the built-in handlers for the get/set float128
6300 exponent, get float128 mantissa, float128 test built-ins, and the
6301 float128 round to odd built-in functions. Eliminate creating the
6302 float128 round to odd built-in functions as special built-ins.
6303 (rs6000_init_builtins): Eliminate special creation of the float128
6304 round to odd built-in functions.
6305 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6306 function insns to support both TFmode and KFmode variants.
6307 (xsxsigqp_<mode>): Likewise.
6308 (xsiexpqpf_<mode>): Likewise.
6309 (xsiexpqp_<mode>): Likewise.
6310 (xststdcqp_<mode>): Likewise.
6311 (xststdcnegqp_<mode>): Likewise.
6312 (xststdcqp_<mode>): Likewise.
6314 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6316 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6317 unused path_in_freq_ptr parameter.
6318 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6320 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6322 * ipa-inline.c (edge_badness): Dump sreal frequency.
6323 (compute_inlined_call_time): Match natural implementaiton ...
6324 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6325 forgotten division by CGRAPH_FREQ_BASE.
6327 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6329 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6330 Solaris 11. Update comment.
6331 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6333 * config/sol2.h (STARTFILE_SPEC): Likewise.
6334 * configure: Regenerate.
6336 2017-11-14 Carl Love <cel@us.ibm.com>
6338 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6339 le_ and be_ prefixes to swap* variables. Remove
6340 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6343 2017-11-14 Jason Merrill <jason@redhat.com>
6345 Support GTY((cache)) on hash_map.
6346 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6347 (ggc_cache_remove): Override it instead of ggc_mx.
6348 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6349 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6350 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6351 (simple_cache_map_traits): Override maybe_mx.
6352 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6353 (hash_map): Friend gt_cleare_cache.
6354 (gt_cleare_cache): New.
6355 * tree.h (tree_cache_traits): New hash_map traits class.
6356 (tree_cache_map): New typedef.
6358 2017-11-14 Richard Biener <rguenther@suse.de>
6360 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6361 paramter and handling.
6362 (cleanup_control_flow_bb): Likewise.
6363 (cleanup_control_flow_pre): New helper performing a DFS walk
6364 to call cleanup_control_flow_bb in PRE order.
6365 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6366 via cleanup_control_flow_pre.
6368 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6370 * config/aarch64/aarch64-simd.md
6371 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6372 (*aarch64_simd_bsl<mode>_alt): Likewise.
6373 (aarch64_simd_bsldi_internal): New.
6374 (aarch64_simd_bsldi_alt): Likewise.
6376 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6378 * tracer.c (better_p): Do not compare frequencies.
6379 * reg-stack.c (better_edge): Likewise.
6380 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6383 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6385 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6386 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6387 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6388 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6389 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6390 * predict.c (maybe_hot_frequency_p): Inline to ...
6391 (maybe_hot_count_p): ... here; rewrite to counts.
6392 (counts_to_freqs): Rename to ...
6393 (update_max_bb_count): ... this one.
6394 (expensive_function_p): Use counts.
6395 (estimate_bb_frequencies): Update.
6396 (rebuild_frequencies): Update.
6397 * predict.h (counts_to_freqs): Rename to ...
6398 (update_max_bb_count): ... this one.
6399 * profile.c (compute_branch_probabilities): Add debug info
6400 * tree-inline.c (expand_call_inline): Update debug info.
6401 (optimize_inline_calls): Use update_max_bb_count..
6402 (tree_function_versioning): Use update_max_bb_count..
6403 * value-prof.c (gimple_value_profile_transformations):
6404 Do not use update_max_bb_count.
6406 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6408 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6409 always use frequencies.
6411 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6413 * bb-reorder.c: Remove frequencies from comments.
6414 (better_edge_p): Use profile counts.
6415 (find_traces): Dump profile counts.
6416 (rotate_loop): Use profile counts.
6417 (find_traces_1_round): Likewise.
6418 (connect_better_edge_p): Use counts instead of probabilities for
6420 (copy_bb_p): Drop early check for non-0 frequency.
6421 (sanitize_hot_paths): Update comments.
6423 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6425 * ipa-split.c (struct split_point): Add count.
6426 (consider_split): Do not compute incoming frequency; compute incoming
6427 count and store it to split_point.
6428 (split_function): Set count of the call to split part correctly.
6430 2017-11-13 Carl Love <cel@us.ibm.com>
6432 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6434 2017-11-13 Tom Tromey <tom@tromey.com>
6436 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6438 2017-11-13 Carl Love <cel@us.ibm.com>
6440 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6441 Add support for builtins:
6442 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6443 vector {un,}signed {char,int,short},
6444 vector {un,}signed {char,int,short}) arguments.
6445 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6446 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6447 Add BU_P9V_AV_2 expansions for the builtins.
6448 * config/rs6000/altivec.h (vec_first_match_index,
6449 vec_first_mismatch_index, vec_first_match_or_eos_index,
6450 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6451 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6452 new extern declaration.
6453 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6454 * config/rs6000/vsx.md (first_match_index_<mode>,
6455 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6456 first_mismatch_or_eos_index_<mode>): Add define expand.
6457 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6458 * doc/extend.texi: Update the built-in documenation file for the new
6461 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6463 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6464 into the min/max operations for _Float<N> and _Float<N>X types.
6466 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6469 * dwarf2out.c (do_eh_frame): New static variable.
6470 (dwarf2out_begin_prologue): Set it.
6471 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6473 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6475 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6477 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6479 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6482 * debug.h (dwarf2out_do_eh_frame): Declare.
6483 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6484 (dwarf2out_do_frame): Use it.
6485 (dwarf2out_do_cfi_asm): Likewise.
6486 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6487 (dwarf2out_assembly_start): Likewise.
6488 (dwarf2out_begin_prologue): Fix comment.
6489 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6490 if the target needs either debug or unwind DWARF2 info.
6491 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6492 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6493 -ftrapping-math, -ftrapv and -fwrapv.
6495 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6497 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6498 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6499 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6500 (estimate_edge_size_and_time): Likewise.
6501 (ipa_merge_fn_summary_after_inlining): Likewise.
6502 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6503 (compute_uninlined_call_time): Use sreal_frequency.
6504 (compute_inlined_call_time): Likewise.
6505 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6506 * profile-count.c: Include sreal.h.
6507 (profile_count::to_sreal_scale): New.
6508 * profile-count.h: Forward declare sreal.
6509 (profile_count::to_sreal_scale): Declare.
6511 2017-11-13 Nathan Sidwell <nathan@acm.org>
6513 * diagnostic.c (maybe_line_and_column): New.
6514 (diagnostic_get_location_text): Use it.
6515 (diagnostic_report_current_module): Likewise.
6516 (test_diagnostic_get_location_text): Add tests.
6518 2017-11-13 Luis Machado <luis.machado@linaro.org>
6520 * doc/md.texi (Specifying processor pipeline description): Fix
6521 incorrect latency for the div instruction example.
6523 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6525 PR tree-optimization/78821
6526 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6527 that bit_not_p is the same.
6528 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6529 (split_group): Count precisely bit_not_p bits in each statement.
6530 (invert_op): New function.
6531 (imm_store_chain_info::output_merged_store): Use invert_op to
6532 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6533 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6535 2017-11-13 Martin Liska <mliska@suse.cz>
6537 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6538 (struct source_info): Likewise.
6539 (add_branch_counts): Likewise.
6540 (add_line_counts): Likewise.
6541 (function_summary): Likewise.
6542 (output_intermediate_line): Likewise.
6543 (generate_results): Likewise.
6545 2017-11-13 Martin Liska <mliska@suse.cz>
6547 * gcov.c (struct block_info): Remove typedef for block_t.
6548 (struct line_info): Likewise.
6549 (line_info::has_block): Likewise.
6550 (EXIT_BLOCK): Likewise.
6551 (unblock): Likewise.
6552 (circuit): Likewise.
6553 (get_cycles_count): Likewise.
6554 (process_file): Likewise.
6555 (read_graph_file): Likewise.
6556 (solve_flow_graph): Likewise.
6557 (find_exception_blocks): Likewise.
6558 (add_line_counts): Likewise.
6559 (accumulate_line_info): Likewise.
6560 (output_line_details): Likewise.
6562 2017-11-13 Martin Liska <mliska@suse.cz>
6564 * gcov.c (struct arc_info): Remove typedef for arc_t.
6565 (struct line_info): Likewise.
6566 (add_branch_counts): Likewise.
6567 (output_branch_count): Likewise.
6568 (function_info::~function_info): Likewise.
6569 (circuit): Likewise.
6570 (output_intermediate_line): Likewise.
6571 (read_graph_file): Likewise.
6572 (solve_flow_graph): Likewise.
6573 (find_exception_blocks): Likewise.
6574 (add_line_counts): Likewise.
6575 (accumulate_line_info): Likewise.
6576 (output_line_details): Likewise.
6577 (output_function_details): Likewise.
6579 2017-11-13 Martin Liska <mliska@suse.cz>
6581 * gcov.c (struct function_info): Remove typedef for function_t.
6582 (struct source_info): Likewise.
6583 (source_info::get_functions_at_location): Likewise.
6584 (solve_flow_graph): Likewise.
6585 (find_exception_blocks): Likewise.
6586 (add_line_counts): Likewise.
6587 (output_intermediate_file): Likewise.
6588 (process_file): Likewise.
6589 (generate_results): Likewise.
6590 (release_structures): Likewise.
6591 (read_graph_file): Likewise.
6592 (read_count_file): Likewise.
6593 (accumulate_line_counts): Likewise.
6594 (output_lines): Likewise.
6596 2017-11-13 Martin Liska <mliska@suse.cz>
6598 * gcov.c (function_info::function_info): Remove num_counts
6599 and add vector<gcov_type>.
6600 (function_info::~function_info): Use the vector.
6601 (process_file): Likewise.
6602 (read_graph_file): Likewise.
6603 (read_count_file): Likewise.
6604 (solve_flow_graph): Likewise.
6606 2017-11-13 Martin Liska <mliska@suse.cz>
6608 * gcov.c (function_info::is_artificial): New function.
6609 (process_file): Erase all artificial early.
6610 (generate_results): Skip as all artificial are already
6613 2017-11-13 Martin Liska <mliska@suse.cz>
6615 * gcov.c (read_graph_file): Store to global vector of functions.
6616 (read_count_file): Iterate the vector.
6617 (process_file): Likewise.
6618 (generate_results): Likewise.
6619 (release_structures): Likewise.
6621 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6623 PR tree-optimization/82954
6624 * gimple-ssa-store-merging.c
6625 (imm_store_chain_info::coalesce_immediate_stores): If
6626 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6628 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6630 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6631 Upddate call to ENDIAN_LANE_N.
6632 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6633 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6634 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6635 and use aarch64_endian_lane_rtx.
6636 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6638 2017-11-12 Tom de Vries <tom@codesourcery.com>
6640 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6642 2017-11-12 Tom de Vries <tom@codesourcery.com>
6644 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6646 2017-11-12 Tom de Vries <tom@codesourcery.com>
6648 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6650 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6651 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6652 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6654 2017-11-11 Martin Sebor <msebor@redhat.com>
6657 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6660 * prefic.c (translate_name): Replace strncpy with memcpy to
6661 avoid -Wstringop-truncation.
6663 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6665 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6667 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6669 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6671 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6673 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6676 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6678 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6679 enable generating XXBRH if the value is in a vector register.
6680 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6681 value is in a vector register.
6682 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6683 register to register bswap64's instead of doing the GPR sequence
6684 used on previous machines.
6685 (bswapdi2_xxbrd): New insn.
6686 (bswapdi2_reg): Disallow on ISA 3.0.
6687 (register to register bswap64 splitter): Do not split the insn on
6688 ISA 3.0 systems that use XXBRD.
6690 2017-11-10 Martin Sebor <msebor@redhat.com>
6693 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6695 (find_subframework_file): Same.
6697 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6699 * auto-profile.c (afdo_indirect_call): Drop frequency.
6700 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6701 (cgraph_node::create_edge): Drop frequency argument.
6702 (cgraph_node::create_indirect_edge): Drop frequency argument.
6703 (cgraph_edge::make_speculative): Drop frequency arguments.
6704 (cgraph_edge::resolve_speculation): Do not update frequencies
6705 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6706 (cgraph_node::dump): Check consistency in IPA mode.
6707 (cgraph_edge::maybe_hot_p): Use IPA counter.
6708 (cgraph_edge::verify_count_and_frequency): Rename to ...
6709 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6710 (cgraph_node::verify_node): Update.
6711 * cgraph.h (struct cgraph_edge): Drop frequency.
6712 (cgraph_edge::frequency): New function.
6713 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6715 (cgraph_edge::rebuild_edges): Likewise.
6716 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6717 (duplicate_thunk_for_node): Do not pass frequency.
6718 (cgraph_node::create_clone): Scale only counts.
6719 (cgraph_node::create_virtual_clone): Do not pass frequency.
6720 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6721 (cgraph_node::create_version_clone): Do not pass frequency.
6722 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6723 (cgraph_node::expand_thunk): Do not pass frequency.
6724 (cgraph_node::create_wrapper): Do not pass frequency.
6725 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6727 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6728 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6729 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6730 (gather_caller_stats): Use frequency function.
6731 (ipcp_cloning_candidate_p): Use frequency function.
6732 (ipcp_propagate_stage): Use frequency function.
6733 (get_info_about_necessary_edges): Use frequency function.
6734 (update_profiling_info): Update only IPA profile.
6735 (update_specialized_profile): Use frequency functoin.
6736 (perhaps_add_new_callers): Update only IPA profile.
6737 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6738 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6739 (dump_ipa_call_summary): Use frequency function.
6740 (estimate_edge_size_and_time): Use frequency function.
6741 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6742 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6743 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6744 (update_noncloned_counts): ... ths one; scale counts only.
6745 (clone_inlined_nodes): Do not scale frequency.
6746 (inline_call): Do not pass frequency.
6747 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6748 (compute_inlined_call_time): Use IPA profile.
6749 (want_inline_small_function_p): Use IPA profile.
6750 (want_inline_self_recursive_call_p): Use IPA profile.
6751 (edge_badness): Use IPA profile.
6752 (lookup_recursive_calls): Use IPA profile.
6753 (recursive_inlining): Do not pass frequency.
6754 (resolve_noninline_speculation): Do not update frequency.
6755 (inline_small_functions): Collect max of IPA profile.
6756 (dump_overall_stats): Dump IPA porfile.
6757 (dump_inline_stats): Dump IPA porfile.
6758 (ipa_inline): Collect IPA stats.
6759 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6760 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6761 (ipa_propagate_frequency): Use frequency function.
6762 (ipa_profile): Cleanup.
6763 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6764 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6765 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6766 (input_node): Do not stream frequency.
6767 (input_edge): Do not stream frequency.
6768 (merge_profile_summaries): Scale only IPA profiles.
6769 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6770 * predict.c (drop_profile): Do not recompute frequency.
6771 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6772 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6773 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6774 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6775 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6776 * tree-inline.c (copy_bb): Do not scale frequency.
6777 (expand_call_inline): Do not scale frequency.
6778 (tree_function_versioning): Do not scale frequency.
6779 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6781 2017-11-10 Julia Koval <julia.koval@intel.com>
6783 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6784 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6785 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6786 (_mm256_maskz_gf2p8affine_epi64_epi8)
6787 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6788 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6789 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6790 (__builtin_ia32_vgf2p8affineqb_v32qi)
6791 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6792 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6794 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6798 (arm_option_override): Refactor.
6799 (arm_option_reconfigure_globals): New.
6800 (arm_options_perform_arch_sanity_checks): New.
6801 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6803 (arm_options_perform_arch_sanity_checks): Likewise
6805 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6807 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6808 (power9-qpmul): New.
6809 * rs6000/rs6000.md ("type" attr): Add qmul.
6810 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6811 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6812 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6814 2017-11-10 Martin Sebor <msebor@redhat.com>
6817 * builtins.c (compute_objsize): Handle arrays that
6818 compute_builtin_object_size likes to fail for. Make extern.
6819 * builtins.h (compute_objsize): Declare.
6820 (check_strncpy_sizes): New function.
6821 (expand_builtin_strncpy): Call check_strncpy_sizes.
6822 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6823 -Wstringop-truncation.
6824 (gimple_fold_builtin_strncat): Same.
6825 * gimple.c (gimple_build_call_from_tree): Set call location.
6826 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6827 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6828 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6829 (handle_builtin_strlen): Use strlen_to_stridx.
6830 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6832 Use strlen_to_stridx.
6833 (pass_strlen::execute): Release strlen_to_stridx.
6834 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6835 (-Wstringop-truncation): Document new option.
6837 2017-11-10 Martin Liska <mliska@suse.cz>
6839 PR gcov-profile/82702
6840 * gcov.c (main): Handle intermediate files in a different
6842 (get_gcov_intermediate_filename): New function.
6843 (output_gcov_file): Remove support of intermediate files.
6844 (generate_results): Allocate intermediate file.
6845 (release_structures): Clean-up properly fn_end.
6846 (output_intermediate_file): Start iterating with line 1.
6848 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6850 PR tree-optimization/82929
6851 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6852 ops_swapped_p non-static data member.
6853 (store_immediate_info::store_immediate_info): Clear it.
6854 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6855 ops set ops_swapped_p.
6856 (count_multiple_uses): Handle ops_swapped_p.
6858 2017-11-10 Martin Liska <mliska@suse.cz>
6860 * coverage.c (coverage_init): Stream information about
6861 support of has_unexecuted_blocks.
6862 * doc/gcov.texi: Document that.
6863 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
6864 * gcov.c (read_graph_file): Likewise.
6865 (output_line_beginning): Fix a small issue with
6868 2017-11-10 Bin Cheng <bin.cheng@arm.com>
6870 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
6871 reference of trivial component.
6873 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6876 * gimple-ssa-store-merging.c
6877 (pass_store_merging::terminate_all_aliasing_chains): For
6878 gimple_store_p stmts also call refs_output_dependent_p.
6880 PR rtl-optimization/82913
6881 * compare-elim.c (try_merge_compare): Punt if def_insn is not
6884 2017-11-09 Jeff Law <law@redhat.com>
6886 * vr-values.h: New file with vr_values class.
6887 * tree-vrp.c: Include vr-values.h
6888 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
6889 data objects into the vr_values class.
6890 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
6891 (get_value_range): Make it a member function within vr_values class.
6892 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
6893 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
6894 (op_with_constant_singleton_value_range): Likewise.
6895 (extract_range_for_var_from_comparison_expr): Likewise.
6896 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
6897 (extract_range_from_binary_expr): Likewise.
6898 (extract_range_from_unary_expr): Likewise.
6899 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
6900 (check_for_binary_op_overflow, extract_range_basic): Likewise.
6901 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
6902 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
6903 (compare_name_with_value, compare_names): Likewise.
6904 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
6905 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
6906 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
6907 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
6908 (extract_range_from_phi_node): Likewise.
6909 (simplify_truth_ops_using_ranges): Likewise.
6910 (simplify_div_or_mod_using_ranges): Likewise.
6911 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
6912 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
6913 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
6914 (simplify_float_conversion_using_ranges): Likewise.
6915 (simplify_internal_call_using_ranges): Likewise.
6916 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
6917 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
6918 poitner via x_vr_values for calls into gimple folder.
6919 (vrp_initialize_lattice): Make this the vr_values ctor.
6920 (vrp_free_lattice): Make this the vr_values dtor.
6921 (set_vr_value): New function.
6922 (class vrp_prop): Add vr_values data member. Add various member
6923 functions as well as member functions that delegate to vr_values.
6924 (check_array_ref): Make a member function within vrp_prop class.
6925 (search_for_addr_array, vrp_initialize): Likewise.
6926 (vrp_finalize): Likewise. Revamp to avoid direct access to
6927 vr_value, values_propagated, etc.
6928 (check_array_bounds): Extract vrp_prop class instance pointer from
6929 walk info structure. Use it to call member functions.
6930 (check_all_array_refs): Make a member function within vrp_prop class.
6931 Smuggle class instance pointer via walk info structure.
6932 (x_vr_values): New local static.
6933 (vrp_valueize): Use x_vr_values to get class instance.
6934 (vr_valueize_1): Likewise.
6935 (class vrp_folder): Add vr_values data member. Add various member
6936 functions as well as member functions that delegate to vr_values.
6937 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
6938 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
6939 class instance from vr_values. Use it to call member functions.
6940 (vrp_dom_walker): Add vr_values data member.
6941 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
6942 instance via x_vr_values.
6943 (identify_jump_threads): Accept vr_values as argument. Store
6944 it into the walker structure.
6945 (evrp_dom_walker): Add vr_values class data member. Add various
6947 (evrp_dom_walker::try_find_new_range): Use vr_values data
6948 member to access the memory allocator.
6949 (evrp_dom_walker::before_dom_children): Store vr_values class
6950 instance into the vrp_folder class.
6951 (evrp_dom_walker::push_value_range): Rework to avoid direct
6952 access to num_vr_values and vr_value.
6953 (evrp_dom_walker::pop_value_range): Likewise.
6954 (execute_early_vrp): Remove call to vrp_initialize_lattice.
6955 Use vr_values to get to dump_all_value_ranges member function.
6956 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
6957 and simplify_cond_using_ranges_2 via vrp_prop class instance.
6958 Pass vr_values class instance down to identify_jump_threads.
6959 Remove call to vrp_free_lattice.
6960 (debug_all_value_ranges): Remove.
6962 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
6963 (vrp_folder): Likewise.
6965 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
6966 Get it from the existing bitmap instead.
6967 (vrp_intersect_ranges_1): Likewise.
6969 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6971 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6973 (store_immediate_info::store_immediate_info): Add bitnotp argument,
6974 set bit_not_p to it.
6975 (imm_store_chain_info::coalesce_immediate_stores): Break group
6976 if bit_not_p is different.
6977 (count_multiple_uses, split_group,
6978 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
6979 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
6980 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
6981 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
6982 to store_immediate_info ctor.
6984 2017-11-09 Jim Wilson <jimw@sifive.com>
6986 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
6987 (scan_prog_file): Likewise.
6989 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6991 * bb-reorder.c (max_entry_frequency): Remove.
6992 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
6993 connect_traces, push_to_next_round_p): Remove prototypes.
6994 (find_traces_1_round): Use counts only.
6995 (push_to_next_round_p): Likewise.
6996 (find_traces): Likewise.
6997 (rotate_loop): Likewise.
6998 (find_traces_1_round): Likewise.
6999 (connect_traces): Likewise.
7000 (edge_order): Likewise.
7002 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7004 * config/arm/arm.c (output_return_instruction): Add comments to
7005 indicate requirement for cmse_nonsecure_entry return to account
7006 for the size of clearing instruction output here.
7007 (thumb_exit): Likewise.
7008 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7009 return in hardfloat mode.
7011 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7013 * config/rs6000/rs6000.c (machine_function): Add a bool,
7014 "toc_is_wrapped_separately".
7015 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7016 if it wasn't explicitly set or unset, we are optimizing for speed, and
7017 doing separate shrink-wrapping.
7018 (rs6000_get_separate_components): Enable the TOC component if
7019 saving the TOC register in the prologue.
7020 (rs6000_components_for_bb): Handle the TOC component.
7021 (rs6000_emit_prologue_components): Store the TOC register where needed.
7022 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7023 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7025 2017-11-09 Martin Jambor <mjambor@suse.cz>
7027 * ipa-param-manipulation.c: New file.
7028 * ipa-param-manipulation.h: Likewise.
7029 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7030 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7031 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7032 (ipa_parm_adjustment): Likewise.
7033 (ipa_parm_adjustment_vec): Likewise.
7034 (ipa_get_vector_of_formal_parms): Moved declaration to
7035 ipa-param-manipulation.h.
7036 (ipa_get_vector_of_formal_parm_types): Likewise.
7037 (ipa_modify_formal_parameters): Likewise.
7038 (ipa_modify_call_arguments): Likewise.
7039 (ipa_combine_adjustments): Likewise.
7040 (ipa_dump_param_adjustments): Likewise.
7041 (ipa_modify_expr): Likewise.
7042 (ipa_get_adjustment_candidate): Likewise.
7043 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7044 ipa-param-manipulation.c.
7045 (ipa_get_vector_of_formal_parm_types): Likewise.
7046 (ipa_modify_formal_parameters): Likewise.
7047 (ipa_modify_call_arguments): Likewise.
7048 (ipa_modify_expr): Likewise.
7049 (get_ssa_base_param): Likewise.
7050 (ipa_get_adjustment_candidate): Likewise.
7051 (index_in_adjustments_multiple_times_p): Likewise.
7052 (ipa_combine_adjustments): Likewise.
7053 (ipa_dump_param_adjustments): Likewise.
7054 * tree-sra.c: Also include ipa-param-manipulation.h
7055 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7058 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7059 Alan Hayward <alan.hayward@arm.com>
7060 David Sherwood <david.sherwood@arm.com>
7062 * doc/sourcebuild.texi (vect_masked_store): Document.
7064 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7065 Alan Hayward <alan.hayward@arm.com>
7066 David Sherwood <david.sherwood@arm.com>
7068 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7070 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7071 Alan Hayward <alan.hayward@arm.com>
7072 David Sherwood <david.sherwood@arm.com>
7074 * doc/sourcebuild.texi (vect_variable_length): Document.
7076 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7077 Alan Hayward <alan.hayward@arm.com>
7078 David Sherwood <david.sherwood@arm.com>
7080 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7082 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7083 Alan Hayward <alan.hayward@arm.com>
7084 David Sherwood <david.sherwood@arm.com>
7086 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7088 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7089 Alan Hayward <alan.hayward@arm.com>
7090 David Sherwood <david.sherwood@arm.com>
7092 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7093 previously undocumented selectors.
7094 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7096 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7098 * doc/rtl.texi (const_vector): Say that elements can be
7099 const_wide_ints too.
7100 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7101 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7102 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7103 * optabs.c (expand_vector_broadcast): Likewise.
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 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7110 scaled index even if the unscaled address was invalid.
7111 Don't increase the complexity of using a scale in that case.
7113 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7114 Alan Hayward <alan.hayward@arm.com>
7115 David Sherwood <david.sherwood@arm.com>
7117 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7118 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7119 UNITS_PER_WORD bytes.
7120 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7121 based on REGMODE_NATURAL_SIZE of the inner mode.
7122 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7123 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7124 * expmed.c (lowpart_bit_field_p): Divide the value up into
7125 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7126 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7127 whether something is likely to occupy more than one register.
7129 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7132 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7133 profile_count::adjust_for_ipa_scaling.
7134 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7135 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7137 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7139 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7141 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7142 (split_group): Add total_orig and total_new arguments, estimate the
7143 number of statements related to the store group without store merging
7144 and with store merging.
7145 (imm_store_chain_info::output_merged_store): Adjust split_group
7146 callers, punt if estimated number of statements with store merging
7147 is not smaller than estimated number of statements without it.
7149 (handled_load): Remove has_single_use checks.
7150 (pass_store_merging::process_store): Likewise.
7152 2017-11-09 Richard Biener <rguenther@suse.de>
7154 PR tree-optimization/82902
7155 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7157 2017-11-09 Martin Liska <mliska@suse.cz>
7160 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7163 2017-11-09 Martin Liska <mliska@suse.cz>
7165 PR tree-optimization/82669
7166 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7168 2017-11-09 Martin Liska <mliska@suse.cz>
7170 PR gcov-profile/48463
7171 * coverage.c (coverage_begin_function): Output also end locus
7172 of a function and information whether the function is
7174 * gcov-dump.c (tag_function): Parse and print the information.
7175 * gcov.c (INCLUDE_MAP): Add include.
7176 (INCLUDE_SET): Likewise.
7177 (struct line_info): Move earlier in the source file because
7178 of vector<line_info> in function_info structure.
7179 (line_info::line_info): Likewise.
7180 (line_info::has_block): Likewise.
7181 (struct source_info): Add new member index.
7182 (source_info::get_functions_at_location): New function.
7183 (function_info::group_line_p): New function.
7184 (output_intermediate_line): New function.
7185 (output_intermediate_file): Use the mentioned function.
7186 (struct function_start): New.
7187 (struct function_start_pair_hash): Likewise.
7188 (process_file): Add code that identifies group functions.
7189 Assign lines either to global or function scope.
7190 (generate_results): Skip artificial functions.
7191 (find_source): Assign index for each source file.
7192 (read_graph_file): Read new flag artificial and end_line.
7193 (add_line_counts): Assign it either to global of function scope.
7194 (accumulate_line_counts): Isolate core of the function to
7195 accumulate_line_info and call it for both function and global
7197 (accumulate_line_info): New function.
7198 (output_line_beginning): Fix GNU coding style.
7199 (print_source_line): New function.
7200 (output_line_details): Likewise.
7201 (output_function_details): Likewise.
7202 (output_lines): Iterate both source (global) scope and function
7204 (struct function_line_start_cmp): New class.
7205 * doc/gcov.texi: Reflect changes in documentation.
7207 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7210 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7211 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7212 and similarly for not instead of neg.
7214 2017-11-08 Andi Kleen <ak@linux.intel.com>
7216 * config/i386/i386.opt: Add -mforce-indirect-call.
7217 * config/i386/predicates.md: Check for flag_force_indirect_call.
7218 * doc/invoke.texi: Document -mforce-indirect-call
7220 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7222 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7224 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7225 config/riscv/riscv.c (predict.h): New include.
7226 (riscv_slow_unaligned_access_p): No longer static.
7227 (riscv_block_move_straight): Add require.
7228 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7230 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7233 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7234 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7235 nonimmediate_operand predicate for operand 1 instead of
7238 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7240 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7242 * config/aarch64/constraints.md (Uml): New constraint.
7243 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7246 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7248 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7249 of two vec_duplicates into a vec_concat.
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 vec_concat.
7255 * config/aarch64/constraints.md (Utq): New constraint.
7256 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7259 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7261 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7262 Simplify vec_merge of vec_duplicate and const_vector.
7263 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7265 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7266 mode iterator. Update predicate on operand 1 to
7267 handle non-const_vec constants. Delete constraints.
7268 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7270 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7272 PR tree-optimization/78821
7273 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7275 (store_operand_info::store_operand_info): Initialize it to false.
7276 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7277 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7278 store in the group, and if chain_info is non-NULL, to ignore altogether
7280 (compatible_load_p): Fail if bit_not_p does not match.
7281 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7282 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7283 (pass_store_merging::process_store): Adjust
7284 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7285 call terminate_all_aliasing_chains newly when adding a store into
7286 a chain with non-NULL chain_info.
7288 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7290 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7292 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7294 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7296 (aarch64_layout_frame): Initialise emit_frame_chain.
7297 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7298 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7300 2017-11-08 Martin Liska <mliska@suse.cz>
7302 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7303 of gimple_call_internal_p.
7305 2017-11-07 Tom de Vries <tom@codesourcery.com>
7307 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7309 2017-11-07 Tom de Vries <tom@codesourcery.com>
7311 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7314 2017-11-08 Martin Liska <mliska@suse.cz>
7317 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7319 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7321 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7323 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7326 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7327 (*cmp<mode>_ccz_1): New insn with $k alternative.
7330 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7331 RTX_COMM_COMPARE as commutative as well.
7332 (ix86_binary_operator_ok): Formatting fix.
7333 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7334 *<code><mode>3<mask_name><round_saeonly_name>,
7335 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7336 *<s>mul<mode>3_highpart<mask_name>,
7337 *vec_widen_umult_even_v16si<mask_name>,
7338 *vec_widen_umult_even_v8si<mask_name>,
7339 *vec_widen_umult_even_v4si<mask_name>,
7340 *vec_widen_smult_even_v16si<mask_name>,
7341 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7342 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7343 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7344 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7345 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7346 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7347 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7348 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7349 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7350 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7351 ix86_binary_operator_ok. Formatting fixes.
7352 (*<plusminus_insn><mode>3<mask_name><round_name>,
7353 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7356 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7358 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7359 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7360 eq and ne if TARGET_ISEL.
7361 (cmp): New code_iterator.
7362 (UNS, UNSU_, UNSIK): New code_attrs.
7363 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7364 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7366 ("eq<mode>3"): ... this.
7367 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7369 ("ne<mode>3"): ... this.
7371 2017-11-07 Julia Koval <julia.koval@intel.com>
7374 * common/config/i386/i386-common.c
7375 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7376 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7377 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7378 * config/i386/i386.opt: Ditto.
7379 * config/i386/i386.c (ix86_target_string): Ditto.
7380 (ix86_option_override_internal): Ditto.
7381 (ix86_init_mpx_builtins): Move MPX to args2.
7382 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7383 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7384 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7385 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7386 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7387 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7388 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7390 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7393 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7394 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7395 (zero-extendsidi peephole2): Remove peephole.
7397 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7400 * stor-layout.c: Include attribs.h.
7401 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7402 explicit lookup of "aligned" attribute.
7404 2017-11-07 Andrew Waterman <andrew@sifive.com>
7406 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7407 (riscv_expand_block_move): Likewise.
7408 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7410 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7411 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7412 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7413 (riscv_adjust_block_mem): Likewise.
7414 (riscv_block_move_loop): Likewise.
7415 (riscv_expand_block_move): Likewise.
7416 * config/riscv/riscv.md (movmemsi): New pattern.
7418 2017-11-07 Michael Clark <michaeljclark@mac.com>
7420 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7421 (MUSL_DYNAMIC_LINKER): Likewise.
7423 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7425 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7426 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7429 2017-11-07 Richard Biener <rguenther@suse.de>
7431 * match.pd: Fix build.
7433 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7434 Jackson Woodruff <jackson.woodruff@arm.com>
7436 PR tree-optimization/71026
7437 * match.pd: Canonicalize negate in division.
7439 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7442 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7444 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7446 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7447 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7449 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7451 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7452 non-scalar integral types.
7453 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7454 (-(A-B), -(~A)): New transformations.
7456 2017-11-07 Tom de Vries <tom@codesourcery.com>
7458 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7459 semicolon after "do {} while (0)".
7460 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7461 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7462 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7463 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7464 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7466 2017-11-07 Tom de Vries <tom@codesourcery.com>
7468 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7469 after "do {} while (0)".
7470 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7471 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7472 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7473 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7474 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7476 2017-11-07 Tom de Vries <tom@codesourcery.com>
7479 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7481 (arm_rtx_costs_internal): Add missing semicolon after
7482 HANDLE_NARROW_SHIFT_ARITH call.
7484 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7486 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7487 disable isel if it was not set explicitly.
7489 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7491 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7492 (add_type_attribute) likewise.
7494 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7496 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7497 of struct ix86_frame.
7498 (ix86_initial_elimination_offset): Likewise.
7499 (ix86_expand_split_stack_prologue): Likewise.
7501 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7503 * tree-vrp.h (enum value_range_type): Update stale comment.
7505 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7506 Alan Hayward <alan.hayward@arm.com>
7507 David Sherwood <david.sherwood@arm.com>
7509 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7510 (aarch64_expand_vec_perm_const): Take the number of units too.
7511 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7512 (aarch64_expand_vec_perm_const): Likewise.
7513 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7514 (vec_perm<mode>): Update accordingly.
7516 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7517 Alan Hayward <alan.hayward@arm.com>
7518 David Sherwood <david.sherwood@arm.com>
7520 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7521 Take the number of units too.
7522 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7523 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7524 but check for a vector mode before rather than after the call.
7525 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7526 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7527 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7528 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7529 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7530 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7531 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7532 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7533 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7534 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7535 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7536 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7537 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7538 (aarch64_sqdmull2_n<mode>): Update accordingly.
7540 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7541 Alan Hayward <alan.hayward@arm.com>
7542 David Sherwood <david.sherwood@arm.com>
7544 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7545 the number of units too.
7546 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7547 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7548 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7549 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7550 (vec_store_lanesxi<mode>): Update accordingly.
7552 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7553 Alan Hayward <alan.hayward@arm.com>
7554 David Sherwood <david.sherwood@arm.com>
7556 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7557 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7558 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7559 of units rather than the mode.
7560 * config/aarch64/iterators.md (nunits): New mode attribute.
7561 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7562 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7563 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7564 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7565 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7566 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7567 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7568 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7569 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7570 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7571 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7572 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7573 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7574 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7575 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7576 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7577 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7578 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7579 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7580 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7581 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7582 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7583 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7584 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7585 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7586 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7587 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7588 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7589 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7590 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7591 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7592 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7593 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7594 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7595 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7596 (aarch64_simd_vec_setv2di): Likewise.
7598 2017-11-06 Carl Love <cel@us.ibm.com>
7600 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7602 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7603 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7605 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7607 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7608 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7609 Add power 8 macro expansions.
7610 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7611 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7612 power 8 instructions. (VSX_XXBR): Add iterator.
7614 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7616 * config/arm/arm.md (predicable_short_it): Change default to "no",
7617 improve documentation, remove uses that are identical to the default.
7618 (enabled_for_depr_it): Rename to enabled_for_short_it.
7619 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7620 Remove default uses.
7621 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7622 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7623 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7624 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7626 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7629 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7630 float128 helper macros, which are no longer used after deleting
7631 the old 'q' built-in functions, and moving the round to odd
7632 built-in functions to being special built-in functions.
7633 (BU_FLOAT128_2): Likewise.
7634 (BU_FLOAT128_1_HW): Likewise.
7635 (BU_FLOAT128_2_HW): Likewise.
7636 (BU_FLOAT128_3_HW): Likewise.
7637 (FABSQ): Delete old 'q' built-in functions.
7638 (COPYSIGNQ): Likewise.
7639 (SQRTF128_ODD): Move round to odd built-in functions to be
7640 special built-in functions, so that we can handle
7641 -mabi=ieeelongdouble.
7642 (TRUNCF128_ODD): Likewise.
7643 (ADDF128_ODD): Likewise.
7644 (SUBF128_ODD): Likewise.
7645 (MULF128_ODD): Likewise.
7646 (DIVF128_ODD): Likewise.
7647 (FMAF128_ODD): Likewise.
7648 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7649 built-in names to 'f128'.
7650 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7651 old 'q' built-in functions, as the machine independent code for
7652 'f128' built-in functions handles this.
7653 (rs6000_expand_builtin): Add expansion for float128 round to odd
7654 functions, keying off on -mabi=ieeelongdouble of whether to use
7655 the KFmode or TFmode variant.
7656 (rs6000_init_builtins): Initialize the _Float128 round to odd
7658 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7659 _Float128 'q' built-in functions are now mapped into the new
7660 'f128' built-in functions.
7662 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7664 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7665 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7667 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7670 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7671 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7672 * toplev.c: Include "ipa-fnsummary.h".
7673 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7675 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7677 PR tree-optimization/82838
7678 * gimple-ssa-store-merging.c
7679 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7680 on a separate gimple_seq which is then appended to seq.
7682 2017-11-06 Jeff Law <law@redhat.com>
7685 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7686 (get_probe_interval): New functions.
7687 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7688 (ix86_adjust_stack_and_probe): Likewise.
7689 (output_adjust_stack_and_probe): Likewise.
7690 (ix86_emit_probe_stack_range): Likewise.
7691 (ix86_expand_prologue): Likewise.
7693 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7695 PR tree-optimization/82816
7696 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7697 if the modes of the two types are the same.
7698 (convert_plusminus_to_widen): Likewise.
7700 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7702 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7704 (usadv16qi): New define_expand.
7705 (usadv8hi): New define_expand.
7707 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7710 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7712 (inline_transform): Likewise.
7713 * predict.c (counts_to_freqs): Remove useless conditional.
7714 * profile-count.h (profile_count::apply_scale): Move sanity check.
7715 * tree-inline.c (copy_bb): Always scale.
7716 (copy_cfg_body): Likewise.
7718 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7721 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7723 (*cmp_ite0): Add enabled_for_depr_it attribute.
7724 (*cmp_ite1): Likewise.
7726 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7728 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7731 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7733 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7734 return the value by pointer.
7735 (extract_range_from_multiplicative_op_1): Update accordingly.
7736 Return as soon as an operation fails.
7738 2017-11-05 Tom de Vries <tom@codesourcery.com>
7741 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7742 (DEF_SANITIZER_BUILTIN): ... here.
7743 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7744 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7746 2017-11-05 Tom de Vries <tom@codesourcery.com>
7749 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7751 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7752 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7753 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7755 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7756 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7757 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7759 2017-11-05 Tom de Vries <tom@codesourcery.com>
7762 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7765 2017-11-04 Michael Clark <michaeljclark@mac.com>
7767 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7768 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7770 (*addsi3_extended): Likewise.
7771 (*addsi3_extended2): Likewise.
7772 (<optab>si3): Likewise.
7773 (<optab>di3): Likewise.
7774 (<optab><mode>3): Likewise.
7775 (<*optabe>si3_internal): Likewise.
7776 (zero_extendqi<SUPERQI:mode>2): Likewise.
7777 (*add<mode>hi3): Likewise.
7778 (*xor<mode>hi3): Likewise.
7779 (<optab>di3): Likewise.
7780 (*<optab>si3_extend): Likewise.
7781 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7782 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7783 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7785 2017-11-04 Andrew Waterman <andrew@sifive.com>
7787 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7788 TARGET_STRICT_ALIGN based upon -mtune argument.
7790 2017-11-04 Andrew Waterman <andrew@sifive.com>
7792 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7794 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7796 * config/i386/i386.c (choose_basereg): Use optional scratch
7797 register and add assertion.
7798 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7799 needed, and don't allocate stack.
7800 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7801 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7802 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7804 2017-11-03 Jeff Law <law@redhat.com>
7806 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7807 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7808 to probe at the start of a noreturn function.
7810 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7812 PR tree-optimization/78821
7813 * gimple-ssa-store-merging.c: Update the file comment.
7814 (MAX_STORE_ALIAS_CHECKS): Define.
7815 (struct store_operand_info): New type.
7816 (store_operand_info::store_operand_info): New constructor.
7817 (struct store_immediate_info): Add rhs_code and ops data members.
7818 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7819 and op1r arguments to the ctor, initialize corresponding data members.
7820 (struct merged_store_group): Add load_align_base and load_align
7822 (merged_store_group::merged_store_group): Initialize them.
7823 (merged_store_group::do_merge): Update them.
7824 (merged_store_group::apply_stores): Pick the constant for
7825 encode_tree_to_bitpos from one of the two operands, or skip
7826 encode_tree_to_bitpos if neither operand is a constant.
7827 (class pass_store_merging): Add process_store method decl. Remove
7828 bool argument from terminate_all_aliasing_chains method decl.
7829 (pass_store_merging::terminate_all_aliasing_chains): Remove
7830 var_offset_p argument and corresponding handling.
7831 (stmts_may_clobber_ref_p): New function.
7832 (compatible_load_p): New function.
7833 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7834 if there is overlap and rhs_code is not INTEGER_CST. For
7835 non-overlapping stores terminate group if rhs is not mergeable.
7836 (get_alias_type_for_stmts): Change first argument from
7837 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7838 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7839 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7841 (get_location_for_stmts): Change first argument from
7842 auto_vec<gimple *> & to vec<gimple *> &.
7843 (struct split_store): Remove orig_stmts data member, add orig_stores.
7844 (split_store::split_store): Create orig_stores rather than orig_stmts.
7845 (find_constituent_stmts): Renamed to ...
7846 (find_constituent_stores): ... this. Change second argument from
7847 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7848 to info structures rather than the statements.
7849 (split_group): Rename ALLOW_UNALIGNED argument to
7850 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7851 it. Adjust find_constituent_stores caller.
7852 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7853 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7854 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7855 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7856 (mem_valid_for_store_merging): New function.
7857 (handled_load): New function.
7858 (pass_store_merging::process_store): New method.
7859 (pass_store_merging::execute): Use process_store method. Adjust
7860 terminate_all_aliasing_chains caller.
7862 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7864 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
7865 Return true for more constants, symbols and label references.
7866 (aarch64_valid_floating_const): Remove unused function.
7868 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7871 * config/aarch64/aarch64.c (aarch64_layout_frame):
7872 Undo forcing of LR at bottom of frame.
7874 2017-11-03 Jeff Law <law@redhat.com>
7877 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
7878 for int_registers_saved.
7880 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
7881 extracted from tree-ssa-dom.c.
7882 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
7883 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
7884 (record_equivalences_from_incoming_edge): Add additional argument
7885 to single_pred_edge_ignoring_loop_edges call.
7886 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
7887 (uncprop_dom_walker::before_dom_children): Add additional argument
7888 to single_pred_edge_ignoring_loop_edges call.
7889 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
7890 single_pred_edge_ignoring_loop_edges rather than open coding.
7891 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
7893 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
7895 * match.pd (-(-A)): Rewrite.
7897 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7899 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
7900 (rs6000_emit_int_cmove): New declaration.
7901 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
7902 (rs6000_emit_sISEL): Delete.
7903 (rs6000_emit_int_cmove): Make non-static.
7904 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
7905 instead of rs6000_emit_sISEL.
7907 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
7909 * asan.c (create_cond_insert_point): Maintain profile.
7910 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
7912 * basic-block.h (struct basic_block_def): Remove frequency.
7913 (EDGE_FREQUENCY): Use to_frequency
7914 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
7916 (find_traces): Update to use to_frequency.
7917 (find_traces_1_round): Likewise; use only IPA counts.
7918 (bb_to_key): Likewise.
7919 (connect_traces): Use IPA counts only.
7920 (copy_bb_p): Update to use to_frequency.
7921 (fix_up_crossing_landing_pad): Likewise.
7922 (sanitize_hot_paths): Likewise.
7923 * bt-load.c (basic_block_freq): Likewise.
7924 * cfg.c (init_flow): Set count_max to uninitialized.
7925 (check_bb_profile): Remove frequencies; check counts.
7926 (dump_bb_info): Do not dump frequencies.
7927 (update_bb_profile_for_threading): Update counts only.
7928 (scale_bbs_frequencies_int): Likewise.
7929 (MAX_SAFE_MULTIPLIER): Remove.
7930 (scale_bbs_frequencies_gcov_type): Update counts only.
7931 (scale_bbs_frequencies_profile_count): Update counts only.
7932 (scale_bbs_frequencies): Update counts only.
7933 * cfg.h (struct control_flow_graph): Add count-max.
7934 (update_bb_profile_for_threading): Update prototype.
7935 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
7936 (find_many_sub_basic_blocks): Likewise.
7937 * cfgcleanup.c (try_forward_edges): Likewise.
7938 (try_crossjump_to_edge): Likewise.
7939 * cfgexpand.c (expand_gimple_cond): Likewise.
7940 (expand_gimple_tailcall): Likewise.
7941 (construct_init_block): Likewise.
7942 (construct_exit_block): Likewise.
7943 * cfghooks.c (verify_flow_info): Check consistency of counts.
7944 (dump_bb_for_graph): Do not dump frequencies.
7945 (split_block_1): Do not update frequencies.
7946 (split_edge): Do not update frequencies.
7947 (make_forwarder_block): Do not update frequencies.
7948 (duplicate_block): Do not update frequencies.
7949 (account_profile_record): Do not update frequencies.
7950 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
7951 for global heuristics.
7952 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
7953 (expected_loop_iterations_unbounded): Use counts only.
7954 * cfgloopmanip.c (scale_loop_profile): Simplify.
7955 (create_empty_loop_on_edge): Simplify
7957 (duplicate_loop_to_header_edge): Simplify
7958 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
7959 (update_br_prob_note): Take care of removing note when profile
7961 (relink_block_chain): Do not dump frequency.
7962 (rtl_account_profile_record): Use to_frequency.
7963 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
7964 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
7965 (cgraph_update_edges_for_call_stmt_node): Likewise.
7966 (cgraph_edge::verify_count_and_frequency): Update.
7967 (cgraph_node::verify_node): Temporarily disable frequency verification.
7968 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
7969 to_cgraph_frequency.
7970 (cgraph_edge::rebuild_edges): Convert to ipa counts.
7971 * cgraphunit.c (init_lowered_empty_function): Do not initialize
7973 (cgraph_node::expand_thunk): Update profile.
7974 * except.c (dw2_build_landing_pads): Do not update frequency.
7975 * final.c (compute_alignments): Use to_frequency.
7976 (dump_basic_block_info): Do not dump frequency.
7977 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
7978 (dump_gimple_bb_header): Do not dump frequency.
7979 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
7981 * gimple-streamer-in.c (input_bb): Do not stream frequency.
7982 * gimple-streamer-out.c (output_bb): Do not stream frequency.
7983 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
7984 (init_before_recovery): Do not update frequency.
7985 (sched_create_recovery_edges): Do not update frequency.
7986 * hsa-gen.c (convert_switch_statements): Do not update frequency.
7987 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
7988 (ipa_cp_c_finalize): Set max_count to uninitialized.
7989 * ipa-fnsummary.c (get_minimal_bb): Use counts.
7990 (param_change_prob): Use counts.
7991 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
7993 * ipa-split.c (consider_split): Use to_frequency.
7994 (split_function): Use to_frequency.
7995 * ira-build.c (loop_compare_func): Likewise.
7996 (mark_loops_for_removal): Likewise.
7997 (mark_all_loops_for_removal): Likewise.
7998 * loop-doloop.c (doloop_modify): Do not update frequency.
7999 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8001 * lto-streamer-in.c (input_function): Update count_max.
8002 * omp-expand.c (expand_omp_taskreg): Update count_max.
8003 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8004 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8005 (maybe_hot_count_p): Use ipa counts only.
8006 (maybe_hot_bb_p): Simplify.
8007 (maybe_hot_edge_p): Simplify.
8008 (probably_never_executed): Do not take frequency argument.
8009 (probably_never_executed_bb_p): Do not pass frequency.
8010 (probably_never_executed_edge_p): Likewise.
8011 (combine_predictions_for_bb): Check that profile is nonzero.
8012 (propagate_freq): Do not set frequency.
8013 (drop_profile): Simplify.
8014 (counts_to_freqs): Simplify.
8015 (expensive_function_p): Use to_frequency.
8016 (propagate_unlikely_bbs_forward): Simplify.
8017 (determine_unlikely_bbs): Simplify.
8018 (estimate_bb_frequencies): Add hack to silence graphite issues.
8019 (compute_function_frequency): Use ipa counts.
8020 (pass_profile::execute): Update.
8021 (rebuild_frequencies): Use counts only.
8022 (force_edge_cold): Use counts only.
8023 * profile-count.c (profile_count::dump): Dump new count types.
8024 (profile_count::differs_from_p): Check compatiblity.
8025 (profile_count::to_frequency): New function.
8026 (profile_count::to_cgraph_frequency): New function.
8027 * profile-count.h (struct function): Declare.
8028 (enum profile_quality): Add profile_guessed_local and
8029 profile_guessed_global0.
8030 (class profile_proability): Decrease number of bits to 29;
8031 update from_reg_br_prob_note and to_reg_br_prob_note.
8032 (class profile_count: Update comment; decrease number of bits
8033 to 61. Check compatibility.
8034 (profile_count::compatible_p): New private member function.
8035 (profile_count::ipa_p): New member function.
8036 (profile_count::operator<): Handle global zero correctly.
8037 (profile_count::operator>): Handle global zero correctly.
8038 (profile_count::operator<=): Handle global zero correctly.
8039 (profile_count::operator>=): Handle global zero correctly.
8040 (profile_count::nonzero_p): New member function.
8041 (profile_count::force_nonzero): New member function.
8042 (profile_count::max): New member function.
8043 (profile_count::apply_scale): Handle IPA scalling.
8044 (profile_count::guessed_local): New member function.
8045 (profile_count::global0): New member function.
8046 (profile_count::ipa): New member function.
8047 (profile_count::to_frequency): Declare.
8048 (profile_count::to_cgraph_frequency): Declare.
8049 * profile.c (OVERLAP_BASE): Delete.
8050 (compute_frequency_overlap): Delete.
8051 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8052 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8053 * sched-ebb.c (rank): Use counts only.
8054 * shrink-wrap.c (handle_simple_exit): Use counts only.
8055 (try_shrink_wrapping): Use counts only.
8056 (place_prologue_for_one_component): Use counts only.
8057 * tracer.c (find_best_predecessor): Use to_frequency.
8058 (find_trace): Use to_frequency.
8059 (tail_duplicate): Use to_frequency.
8060 * trans-mem.c (expand_transaction): Do not update frequency.
8061 * tree-call-cdce.c: Do not update frequency.
8062 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8063 (gimple_merge_blocks): Likewise.
8064 (gimple_split_edge): Likewise.
8065 (gimple_duplicate_sese_region): Likewise.
8066 (gimple_duplicate_sese_tail): Likewise.
8067 (move_sese_region_to_fn): Likewise.
8068 (gimple_account_profile_record): Likewise.
8069 (insert_cond_bb): Likewise.
8070 * tree-complex.c (expand_complex_div_wide): Likewise.
8071 * tree-eh.c (lower_resx): Update profile.
8072 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8074 (initialize_cfun): Do not initialize frequencies
8075 (freqs_to_counts): Delete.
8076 (copy_cfg_body): Ignore count parameter.
8077 (copy_body): Update.
8078 (expand_call_inline): Update count_max.
8079 (optimize_inline_calls): Update count_max.
8080 (tree_function_versioning): Update count_max.
8081 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8082 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8084 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8085 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8086 (try_peel_loop): Likewise.
8087 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8089 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8090 (tree_transform_and_unroll_loop): Do not use frequencies
8091 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8092 Use reliable prediction only.
8093 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8094 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8095 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8096 probability scaling.
8097 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8098 not update frequency
8099 (any_remaining_duplicated_blocks): Likewise.
8100 (update_profile): Likewise.
8101 (estimated_freqs_path): Delete.
8102 (freqs_to_counts_path): Delete.
8103 (clear_counts_path): Delete.
8104 (ssa_fix_duplicate_block_edges): Likewise.
8105 (duplicate_thread_path): Likewise.
8106 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8107 * tree-tailcall.c (decrease_profile): Do not update frequency.
8108 (eliminate_tail_call): Likewise.
8109 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8110 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8111 (optimize_mask_stores): Likewise.
8112 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8113 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8114 (ubsan_expand_ptr_ifn): Update profile.
8115 * value-prof.c (gimple_ic): Simplify.
8116 * value-prof.h (gimple_ic): Update prototype.
8117 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8118 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8120 (want_inline_self_recursive_call_p): Likewise.
8121 (resolve_noninline_speculation): Only cummulate defined counts.
8122 (inline_small_functions): Use nonzero_p.
8123 (ipa_inline): Do not access freed node.
8125 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8127 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8128 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8130 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8132 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8133 non-legitimate address.
8135 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8137 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8138 (*lt0_<mode>di, *lt0_<mode>si): New.
8140 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8142 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8143 TARGET_PAIRED_FLOAT.
8145 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8146 Jim Wilson <jim.wilson@linaro.org>
8148 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8149 * config/aarch64/aarch64-tune.md: Regenerated.
8150 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8151 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8153 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8155 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8156 (arc_expand_prologue): Restore blink for millicode.
8157 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8159 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8162 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8163 gen_vec_duplicate after forcing the scalar into a register.
8165 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8167 * combine (try_combine): Print the insns input to try_combine to the
8170 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8173 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8174 Remove second argument from aarch64_process_target_attr call.
8175 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8177 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8179 (aarch64_handle_attr_arch): Remove second argument.
8180 (aarch64_handle_attr_cpu): Ditto.
8181 (aarch64_handle_attr_tune): Ditto.
8182 (aarch64_handle_attr_isa_flags): Ditto.
8183 (aarch64_process_one_target_attr): Ditto.
8184 (aarch64_process_target_attr): Ditto.
8185 (aarch64_option_valid_attribute_p): Remove second argument.
8186 on aarch64_process_target_attr call.
8188 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8190 * diagnostic.c: Include "selftest-diagnostic.h".
8191 (selftest::assert_location_text): New function.
8192 (selftest::test_diagnostic_get_location_text): New function.
8193 (selftest::diagnostic_c_tests): Call it.
8195 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8197 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8198 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8199 (class selftest::test_diagnostic_context): Move to...
8200 * selftest-diagnostic.c: New file.
8201 * selftest-diagnostic.h: New file.
8203 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8205 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8206 offset range for FT32B.
8207 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8208 * config/ft32/ft32.md: Add TARGET_NOPM.
8209 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8210 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8212 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8214 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8216 2017-11-02 Jeff Law <law@redhat.com>
8218 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8219 virtual keyword on FINAL OVERRIDE members.
8221 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8222 virtuals together. Add virtual destructor.
8223 (substitute_and_fold_engine): Similarly.
8225 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8227 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8229 2017-11-02 Richard Biener <rguenther@suse.de>
8231 PR tree-optimization/82795
8232 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8234 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8236 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8238 (gcc_GAS_CHECK_FEATURE): Remove.
8239 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8240 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8241 * configure: Regenerate.
8243 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8245 * config/arc/arc.c (hwloop_optimize): Account for empty
8248 2017-11-02 Richard Biener <rguenther@suse.de>
8251 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8252 Truncate ARRAY_REF index and element size.
8254 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8256 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8258 2017-11-01 Jeff Law <law@redhat.com>
8260 * tree-ssa-ccp.c (ccp_folder): New class derived from
8261 substitute_and_fold_engine.
8262 (ccp_folder::get_value): New member function.
8263 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8264 (ccp_fold_stmt): Remove prototype.
8265 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8266 * tree-ssa-copy.c (copy_folder): New class derived from
8267 substitute_and_fold_engine.
8268 (copy_folder::get_value): Renamed from get_value.
8269 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8270 * tree-vrp.c (vrp_folder): New class derived from
8271 substitute_and_fold_engine.
8272 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8273 (vrp_folder::get_value): New member function.
8274 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8275 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8276 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8277 provide a class interface to folder/substitute routines.
8278 (ssa_prop_fold_stmt_fn): Remove typedef.
8279 (ssa_prop_get_value_fn): Likewise.
8280 (subsitute_and_fold): Remove prototype.
8281 (replace_uses_in): Likewise.
8282 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8283 Renamed from replace_uses_in. Call the virtual member function
8284 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8285 (substitute_and_fold_dom_walker): Remove initialization of
8286 data member entries for calbacks. Add substitute_and_fold_engine
8287 member and initialize it.
8288 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8289 member functions for get_value, replace_phi_args_in c
8290 replace_uses_in, and fold_stmt calls.
8291 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8292 substitute_and_fold. Remove assert. Update ctor call.
8294 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8295 (ssa_prop_visit_phi_fn): Likewise.
8296 (class ssa_propagation_engine): New class to provide an interface
8298 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8300 (ssa_prop_visit_phi): Likewise.
8301 (ssa_propagation_engine::simulate_stmt): Moved into class.
8302 Call visit_phi/visit_stmt from the class rather than via
8303 file scoped static variables.
8304 (ssa_propagation_engine::simulate_block): Moved into class.
8305 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8306 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8307 set file scoped statics for the visit_stmt/visit_phi callbacks.
8308 * tree-complex.c (complex_propagate): New class derived from
8309 ssa_propagation_engine.
8310 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8311 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8312 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8314 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8315 ssa_propagation_engine.
8316 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8317 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8318 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8319 * tree-ssa-copy.c (copy_prop): New class derived from
8320 ssa_propagation_engine.
8321 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8322 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8323 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8324 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8325 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8326 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8327 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8329 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8331 PR rtl-optimization/82778
8332 PR rtl-optimization/82597
8333 * compare-elim.c (struct comparison): Add in_a_setter field.
8334 (find_comparison_dom_walker::before_dom_children): Remove killed
8335 bitmap and df_simulate_find_defs call, instead walk the defs.
8336 Compute last_setter and initialize in_a_setter. Merge definitions
8337 with first initialization for a few variables.
8338 (try_validate_parallel): Use insn_invalid_p instead of
8339 recog_memoized. Return insn rather than just the pattern.
8340 (try_merge_compare): Fix up comment. Don't uselessly test if
8341 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8343 (execute_compare_elim_after_reload): Remove df_chain_add_problem
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/aarch64.c (aarch64_rtx_costs): Use
8351 aarch64_hard_regno_nregs to get the number of registers
8354 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8355 Alan Hayward <alan.hayward@arm.com>
8356 David Sherwood <david.sherwood@arm.com>
8358 * config/aarch64/constraints.md (Upl): Rename to...
8360 * config/aarch64/aarch64.md
8361 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
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_add_constant_internal)
8369 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8372 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8373 Alan Hayward <alan.hayward@arm.com>
8374 David Sherwood <david.sherwood@arm.com>
8376 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8377 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8378 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8380 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8381 of a single element.
8382 * config/aarch64/iterators.md: Add a comment above the permute
8383 unspecs to say that they are generated directly by
8384 aarch64_expand_vec_perm_const.
8385 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8387 2017-11-01 Nathan Sidwell <nathan@acm.org>
8389 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8391 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8393 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8394 and medany code models, and describe what they do.
8396 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8398 Revert accidental duplicate:
8400 * combine.c (can_change_dest_mode): Reject changes in
8401 REGMODE_NATURAL_SIZE.
8403 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8405 PR rtl-optimization/64682
8406 PR rtl-optimization/69567
8407 PR rtl-optimization/69737
8408 PR rtl-optimization/82683
8409 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8410 register mentioned in the note, drop the note, unless it came from I3,
8411 in which case it should go to I3 again.
8413 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8415 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8416 and return false if not.
8417 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8419 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8421 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8422 range is known to be empty.
8424 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8425 Alan Hayward <alan.hayward@arm.com>
8426 David Sherwood <david.sherwood@arm.com>
8428 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8429 and CONST_VECTOR_NUNITS instead of computing the number of units from
8430 the byte sizes of the vector and element.
8431 (simplify_binary_operation_1): Likewise.
8432 (simplify_const_binary_operation): Likewise.
8433 (simplify_ternary_operation): Likewise.
8435 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8436 Alan Hayward <alan.hayward@arm.com>
8437 David Sherwood <david.sherwood@arm.com>
8439 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8440 (int_mem_offset): ...this new function.
8441 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8442 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8443 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8446 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8447 Alan Hayward <alan.hayward@arm.com>
8448 David Sherwood <david.sherwood@arm.com>
8450 * lower-subreg.c (interesting_mode_p): New function.
8451 (compute_costs, find_decomposable_subregs, decompose_register)
8452 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8453 (resolve_clobber, dump_choices): Use it.
8455 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8456 Alan Hayward <alan.hayward@arm.com>
8457 David Sherwood <david.sherwood@arm.com>
8459 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8461 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8462 Alan Hayward <alan.hayward@arm.com>
8463 David Sherwood <david.sherwood@arm.com>
8465 * alias.c (find_base_value, find_base_term): Only process integer
8466 truncations. Check the precision rather than the size.
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 * machmode.h (is_narrower_int_mode): New function
8473 * optabs.c (expand_float, expand_fix): Use it.
8474 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8476 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8477 Alan Hayward <alan.hayward@arm.com>
8478 David Sherwood <david.sherwood@arm.com>
8480 * rtl.h (narrower_subreg_mode): New function.
8481 * ira-color.c (update_costs_from_allocno): Use it.
8483 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8484 Alan Hayward <alan.hayward@arm.com>
8485 David Sherwood <david.sherwood@arm.com>
8487 * optabs-query.h (convert_optab_p): New function, split out from...
8488 (convert_optab_handler): ...here.
8489 (widening_optab_handler): Delete.
8490 (find_widening_optab_handler): Remove permit_non_widening parameter.
8491 (find_widening_optab_handler_and_mode): Likewise. Provide an
8492 override that operates on mode class wrappers.
8493 * optabs-query.c (widening_optab_handler): Delete.
8494 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8495 parameter. Assert that the two modes are the same class and that
8496 the "from" mode is narrower than the "to" mode. Use
8497 convert_optab_handler instead of widening_optab_handler.
8498 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8499 instead of widening_optab_handler.
8500 * expr.c (expand_expr_real_2): Update calls to
8501 find_widening_optab_handler.
8502 * optabs.c (expand_widen_pattern_expr): Likewise.
8503 (expand_binop_directly): Take the insn_code as a parameter.
8504 (expand_binop): Only call find_widening_optab_handler for
8505 conversion optabs; use optab_handler otherwise. Update calls
8506 to find_widening_optab_handler and expand_binop_directly.
8507 Use convert_optab_handler instead of widening_optab_handler.
8508 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8509 find_widening_optab_handler and use scalar_mode rather than
8511 (convert_plusminus_to_widen): Likewise.
8513 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8514 Alan Hayward <alan.hayward@arm.com>
8515 David Sherwood <david.sherwood@arm.com>
8517 * machmode.h (fixed_size_mode): New class.
8518 * rtl.h (get_pool_mode): Return fixed_size_mode.
8519 * gengtype.c (main): Add fixed_size_mode.
8520 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8521 (get_raw_arg_mode): Likewise.
8522 * doc/tm.texi: Regenerate.
8523 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8524 * targhooks.c (default_get_reg_raw_mode): Likewise.
8525 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8526 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8527 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8528 (msp430_get_raw_result_mode): Likewise.
8529 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8530 * dbxout.c (dbxout_parms): Require fixed-size modes.
8531 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8532 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8533 * omp-low.c (lower_oacc_reductions): Likewise.
8534 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8535 (simplify_subreg): Update accordingly.
8536 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8537 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8538 that aren't fixed-size.
8539 (get_pool_mode): Return a fixed_size_mode.
8540 (output_constant_pool_2): Take a fixed_size_mode.
8542 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8543 Alan Hayward <alan.hayward@arm.com>
8544 David Sherwood <david.sherwood@arm.com>
8546 * doc/rtl.texi (vec_series): Document.
8547 (const): Say that the operand can be a vec_series.
8548 * rtl.def (VEC_SERIES): New rtx code.
8549 * rtl.h (const_vec_series_p_1): Declare.
8550 (const_vec_series_p): New function.
8551 * emit-rtl.h (gen_const_vec_series): Declare.
8552 (gen_vec_series): Likewise.
8553 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8554 (gen_vec_series): Likewise.
8555 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8556 * simplify-rtx.c (simplify_unary_operation): Handle negations
8558 (simplify_binary_operation_series): New function.
8559 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8560 (test_vector_ops_series): New function.
8561 (test_vector_ops): Call it.
8562 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8563 gen_const_vec_series.
8564 (altivec_lvsr): Likewise.
8565 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8567 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8568 Alan Hayward <alan.hayward@arm.com>
8569 David Sherwood <david.sherwood@arm.com>
8571 * doc/rtl.texi (const): Update description of address constants.
8572 Say that vector constants are allowed too.
8573 * common.md (E, F): Use CONSTANT_P instead of checking for
8575 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8576 checking for CONST_VECTOR.
8577 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8579 * expr.c (expand_expr_real_2): Check for vector modes instead
8580 of checking for CONST_VECTOR.
8581 * rtl.h (const_vec_p): New function.
8582 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8583 (unwrap_const_vec_duplicate): Handle them here too.
8585 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8586 David Malcolm <dmalcolm@redhat.com>
8587 Alan Hayward <alan.hayward@arm.com>
8588 David Sherwood <david.sherwood@arm.com>
8590 * rtl.h (vec_duplicate_p): New function.
8591 * selftest-rtl.c (assert_rtx_eq_at): New function.
8592 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8593 (assert_rtx_eq_at): Declare.
8594 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8595 * selftest-run-tests.c (selftest::run_tests): Call it.
8596 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8597 (simplify_unary_operation_1): Recursively handle vector duplicates.
8598 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8600 (simplify_subreg): Handle subregs of vector duplicates.
8601 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8602 (selftest::simplify_rtx_c_tests): New functions.
8604 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8605 Alan Hayward <alan.hayward@arm.com>
8606 David Sherwood <david.sherwood@arm.com>
8608 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8609 (gen_vec_duplicate): Likewise.
8610 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8612 (gen_const_vector): ...here.
8613 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8614 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8615 whose elements are all equal.
8616 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8617 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8618 (simplify_relational_operation): Likewise.
8619 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8621 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8622 (aarch64_expand_vector_init): Likewise.
8623 * config/arm/arm.c (neon_vdup_constant): Likewise.
8624 (neon_expand_vector_init): Likewise.
8625 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8626 (arm_block_set_unaligned_vect): Likewise.
8627 (arm_block_set_aligned_vect): Likewise.
8628 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8629 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8630 (expand_vec_perm_even_odd_pack): Likewise.
8631 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8632 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8633 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8634 gen_const_vec_duplicate.
8635 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8636 * config/mips/mips.c (mips_gen_const_int_vector): Use
8637 gen_const_vec_duplicate.
8638 (mips_expand_vector_init): Use CONST0_RTX.
8639 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8640 (define_split): Use gen_const_vec_duplicate.
8641 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8642 (define_split): Use gen_const_vec_duplicate.
8643 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8644 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8645 * config/spu/spu.c (spu_const): Likewise.
8647 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8648 Alan Hayward <alan.hayward@arm.com>
8649 David Sherwood <david.sherwood@arm.com>
8651 * combine.c (can_change_dest_mode): Reject changes in
8652 REGMODE_NATURAL_SIZE.
8654 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8656 * configure.ac (--enable-libssp): New.
8657 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8658 trying to determine target-specific default. Adjust indentation.
8659 * configure: Regenerated.
8660 * doc/install.texi (Configuration): Expand --disable-libssp
8663 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8665 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8668 2017-10-31 Martin Jambor <mjambor@suse.cz>
8671 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8673 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8675 * auto-profile.c (autofdo_source_profile::read): Use
8676 UNKNOWN_LOCATION rather than 0.
8677 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8678 (warning_at): ...this overload.
8679 (warning_at_rich_loc_n): Rename to...
8680 (warning_n): ...this overload.
8681 (error_at_rich_loc): Rename to...
8682 (error_at): ...this overload.
8683 (pedwarn_at_rich_loc): Rename to...
8684 (pedwarn): ...this overload.
8685 (permerror_at_rich_loc): Rename to...
8686 (permerror): ...this overload.
8687 (inform_at_rich_loc): Rename to...
8688 (inform): ...this overload.
8689 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8690 (diagnostic_n_impl_richloc): Rename to...
8691 (diagnostic_n_impl): ...this rich_location *-based decl.
8692 (inform_at_rich_loc): Rename to...
8693 (inform): ...this, and add an assertion.
8694 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8695 (warning_at_rich_loc): Rename to...
8696 (warning_at): ...this, and add an assertion.
8697 (warning_at_rich_loc_n): Rename to...
8698 (warning_n): ...this, and add an assertion.
8699 (warning_n): Update location_t-based implementation for removal of
8700 location_t-based diagnostic_n_impl.
8701 (pedwarn_at_rich_loc): Rename to...
8702 (pedwarn): ...this, and add an assertion.
8703 (permerror_at_rich_loc): Rename to...
8704 (permerror): ...this, and add an assertion.
8705 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8706 (error_at_rich_loc): Rename to...
8707 (error_at): ...this, and add an assertion.
8708 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8709 (driver::do_spec_on_infiles): Likewise.
8710 * substring-locations.c (format_warning_va): Update for renaming
8711 of inform_at_rich_loc.
8713 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8715 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8716 _Float<N>X built-in functions so that the variant without the
8717 "__builtin_" prefix is only enabled for the GNU C and Objective C
8718 languages when they are in non-strict ANSI/ISO mode.
8719 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8720 * target.def (floatn_builtin_p): Add a target hook to control
8721 whether _Float<N> and _Float<N>X built-in functions without the
8722 "__builtin_" prefix are enabled, and return true for C and
8723 Objective C in the default hook. Include langhooks.h in
8725 * targhooks.h (default_floatn_builtin_p): Likewise.
8726 * targhooks.c (default_floatn_builtin_p): Likewise.
8727 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8728 floatn_builtin_p target hook.
8729 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8731 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8732 Eric Botcazou <ebotcazou@adacore.com>
8734 PR rtl-optimization/81803
8735 * lra-constraints.c (curr_insn_transform): Also reload the whole
8736 register for a strict subreg no wider than a word if this is for
8737 a WORD_REGISTER_OPERATIONS target.
8739 2017-10-31 Jason Merrill <jason@redhat.com>
8741 * gdbinit.in: Skip over inlines from timevar.h.
8743 2017-10-31 Martin Liska <mliska@suse.cz>
8745 * doc/gcov.texi: Document new option.
8746 * gcov.c (print_usage): Likewise print it.
8747 (process_args): Support the argument.
8748 (format_count): New function.
8749 (format_gcov): Use the function.
8751 2017-10-31 Martin Liska <mliska@suse.cz>
8753 * gcov.c (struct name_map): do not use typedef.
8754 Define operator== and operator<.
8755 (name_search): Remove.
8756 (name_sort): Remove.
8757 (main): Do not allocate names.
8758 (process_file): Add vertical space.
8759 (generate_results): Use std::find.
8760 (release_structures): Do not release memory.
8761 (find_source): Use std::find.
8763 2017-10-31 Martin Liska <mliska@suse.cz>
8765 * gcov.c (struct line_info): Remove it's typedef.
8766 (line_info::line_info): Add proper ctor.
8767 (line_info::has_block): Do not use a typedef.
8768 (struct source_info): Do not use typedef.
8769 (circuit): Likewise.
8770 (get_cycles_count): Likewise.
8771 (output_intermediate_file): Iterate via vector iterator.
8772 (add_line_counts): Use std::vector methods.
8773 (accumulate_line_counts): Likewise.
8774 (output_lines): Likewise.
8776 2017-10-31 Martin Liska <mliska@suse.cz>
8778 * gcov.c (struct source_info): Remove typedef.
8779 (source_info::source_info): Add proper ctor.
8780 (accumulate_line_counts): Use struct, not it's typedef.
8781 (output_gcov_file): Likewise.
8782 (output_lines): Likewise.
8783 (main): Do not allocate an array.
8784 (output_intermediate_file): Use size of vector container.
8785 (process_file): Resize the vector.
8786 (generate_results): Do not preallocate, use newly added vector
8788 (release_structures): Do not release sources.
8789 (find_source): Use vector methods.
8790 (add_line_counts): Do not use typedef.
8792 2017-10-31 Martin Liska <mliska@suse.cz>
8794 * doc/gcov.texi: Document that.
8795 * gcov.c (add_line_counts): Mark lines with a non-executed
8797 (output_line_beginning): Handle such lines.
8798 (output_lines): Pass new argument.
8799 (output_intermediate_file): Print it in intermediate format.
8801 2017-10-31 Martin Liska <mliska@suse.cz>
8803 * color-macros.h: New file.
8804 * diagnostic-color.c: Factor out color related to macros to
8806 * doc/gcov.texi: Document -k option.
8807 * gcov.c (INCLUDE_STRING): Include string.h.
8808 (print_usage): Add -k option.
8809 (process_args): Parse it.
8810 (pad_count_string): New function.
8811 (output_line_beginning): Likewise.
8812 (DEFAULT_LINE_START): New macro.
8813 (output_lines): Support color output.
8815 2017-10-31 Martin Liska <mliska@suse.cz>
8817 PR gcov-profile/82633
8818 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8819 their interaction with GCOV infrastructure.
8820 * configure.ac: Add -fkeep-{inline,static}-functions to
8822 * configure: Regenerate.
8824 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8827 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8829 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8832 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8833 into a register if it does not fit into an immediate offset field.
8835 2017-10-31 Olivier Hainque <hainque@adacore.com>
8837 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8839 2017-10-31 Julia Koval <julia.koval@intel.com>
8841 * config.gcc: Add gfniintrin.h.
8842 * config/i386/gfniintrin.h: New.
8843 * config/i386/i386-builtin-types.def
8844 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8845 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8846 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8847 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8848 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8849 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8850 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8851 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8852 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8853 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8854 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8855 * config/i386/immintrin.h: Include gfniintrin.h.
8856 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8858 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
8860 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
8862 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8864 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
8865 (arm_ashldi3_1bit): Remove pattern.
8866 (ashrdi3): Remove shift by 1 expansion.
8867 (arm_ashrdi3_1bit): Remove pattern.
8868 (lshrdi3): Remove shift by 1 expansion.
8869 (arm_lshrdi3_1bit): Remove pattern.
8870 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
8871 cost of ashldi3 by 1.
8872 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
8873 (<shift>di3_neon): Likewise.
8875 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8877 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
8878 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
8879 and (*aarch64_simd_mov<VQ:mode>).
8880 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
8881 pattern alternative.
8882 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
8883 attributes to match pattern alternative.
8885 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
8887 * config.gcc (powerpc*-*-*): Add emmintrin.h.
8888 * config/rs6000/emmintrin.h: New file.
8889 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
8891 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8893 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
8894 * (movdi_vfp_cortexa8): Remove pattern.
8896 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8898 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
8900 (Specific, alpha*-dec-osf5.1): Remove.
8901 (Specific, mips-sgi-irix5): Remove.
8902 (Specific, mips-sgi-irix6): Remove.
8904 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8907 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
8908 arguments to clear_bit_region_be.
8910 2017-10-30 Jim Wilson <wilson@tuliptree.org>
8912 * gimplify.c: Include memmodel.h.
8914 2017-10-30 Martin Jambor <mjambor@suse.cz>
8916 * omp-grid.c (grid_attempt_target_gridification): Also insert a
8917 condition whether loop should be executed at all.
8919 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
8921 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
8922 gimple folding of vec_madd() intrinsics.
8923 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
8924 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
8925 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
8927 2017-10-30 Richard Biener <rguenther@suse.de>
8929 PR tree-optimization/82762
8931 2017-10-23 Richard Biener <rguenther@suse.de>
8933 PR tree-optimization/82129
8935 2017-08-01 Richard Biener <rguenther@suse.de>
8937 PR tree-optimization/81181
8938 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8939 (compute_antic): ... end of iteration here.
8941 2017-10-30 Joseph Myers <joseph@codesourcery.com>
8943 * doc/invoke.texi (C Dialect Options): Document -std=c17,
8944 -std=iso9899:2017 and -std=gnu17.
8945 * doc/standards.texi (C Language): Document C17 support.
8946 * doc/cpp.texi (Overview): Mention -std=c17.
8947 (Standard Predefined Macros): Document C11 and C17 values of
8948 __STDC_VERSION__. Do not refer to C99 support as incomplete.
8949 * doc/extend.texi (Inline): Do not list individual options for
8950 standards newer than C99.
8951 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
8953 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
8956 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
8958 * asan.c (asan_finish_file): Align asan globals array by shadow
8961 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8964 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
8965 (struct store_immediate_info): Add bitregion_start and bitregion_end
8967 (store_immediate_info::store_immediate_info): Add brs and bre
8968 arguments and initialize bitregion_{start,end} from those.
8969 (struct merged_store_group): Add bitregion_start, bitregion_end,
8970 align_base and mask fields. Drop unnecessary struct keyword from
8971 struct store_immediate_info. Add do_merge method.
8972 (clear_bit_region_be): Use memset instead of loop storing zeros.
8973 (merged_store_group::do_merge): New method.
8974 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
8975 stores as long as the surrounding bitregions have no gaps.
8976 (merged_store_group::merge_overlapping): Use do_merge.
8977 (merged_store_group::apply_stores): Test that bitregion_{start,end}
8978 is byte aligned, rather than requiring that start and width are
8979 byte aligned. Drop unnecessary struct keyword from
8980 struct store_immediate_info. Allocate and populate also mask array.
8981 Make start of the arrays relative to bitregion_start rather than
8982 start and size them according to bitregion_{end,start} difference.
8983 (struct imm_store_chain_info): Drop unnecessary struct keyword from
8984 struct store_immediate_info.
8985 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
8986 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
8987 struct keyword from struct store_immediate_info.
8988 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
8989 between stores as long as the surrounding bitregions have no gaps.
8991 (struct split_store): Add orig non-static data member.
8992 (split_store::split_store): Initialize orig to false.
8993 (find_constituent_stmts): Return store_immediate_info *, non-NULL
8994 if there is exactly a single original stmt. Change stmts argument
8995 to pointer from reference, if NULL, don't push anything to it. Add
8996 first argument, use it to optimize skipping over orig stmts that
8997 are known to be before bitpos already. Simplify.
8998 (split_group): Return unsigned int count how many stores are or
8999 would be needed rather than a bool. Add allow_unaligned argument.
9000 Change split_stores argument from reference to pointer, if NULL,
9001 only do a dry run computing how many stores would be produced.
9002 Rewritten algorithm to use both alignment and misalign if
9003 !allow_unaligned and handle bitfield stores with gaps.
9004 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9005 from bitregion_start instead of start. Compute allow_unaligned
9006 here, if true, do 2 split_group dry runs to compute which one
9007 produces fewer stores and prefer aligned if equal. Punt if
9008 new count is bigger or equal than original before emitting any
9009 statements, rather than during that. Remove no longer needed
9010 new_ssa_names tracking. Replace num_stmts with
9011 split_stores.length (). Use 32-bit stack allocated entries
9012 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9013 if possible. Handle bitfields with gaps.
9014 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9015 Compute bitregion_{start,end} for the stores and construct
9016 store_immediate_info with that. Formatting fixes.
9018 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9021 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9022 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9024 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9026 * gimplify.c: Include tm_p.h.
9028 * common.opt (gcoff): Re-add as ignored option.
9029 (gcoff1, gcoff2, gcoff3): Likewise.
9031 * Makefile.in (OBJS): Delete sdbout.o.
9032 (GTFILES): Delete $(srcdir)/sdbout.c.
9033 * debug.h: Delete sdb_debug_hooks.
9034 * final.c: Delete sdbout.h include.
9035 (final_scan_insn): Delete SDB_DEBUG check.
9036 (rest_of_clean_state): Likewise.
9037 * output.h: Delete sdb_begin_function_line.
9040 * toplev.c: Delete sdbout.h include.
9041 (process_options): Delete SDB_DEBUG check.
9042 * tree-core.h (tree_type_common): Delete pointer field of
9044 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9045 TYPE_SYMTAB_POINTER.
9046 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9047 (TYPE_SYMTAB_IS_POINTER): Delete.
9048 (TYPE_SYMTAB_IS_DIE): Renumber.
9049 * xcoffout.c: Refer to former sdbout.c file.
9050 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9052 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9053 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9054 (Debugging Options): Delete -gcoff.
9055 (-gstabs): Delete SDB reference.
9057 (-gcoff@var{level}): Delete.
9058 * doc/passes.texi (Debugging information output): Delete SDB and
9059 sdbout.c references.
9060 * doc/tm.texi: Regenerate.
9061 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9062 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9064 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9065 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9066 (SDB_DEBUGGING_INFO): Delete.
9067 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9068 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9069 * target.def (output_source_filename): Delete COFF reference.
9071 * common.opt (gcoff): Delete.
9072 (gxcoff+): Update Negative chain.
9073 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9075 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9076 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9077 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9079 (expand_function_start): Change sdb reference to past tense.
9080 (expand_function_end): Change sdb reference to past tense.
9081 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9082 * opts.c (debug_type_names): Delete coff entry.
9083 (common_handle_option): Delete OPT_gcoff case.
9084 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9086 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9087 * config/cris/cris.h: Delete SDB reference in comment.
9088 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9089 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9090 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9091 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9093 (ix86_expand_prologue): Likewise.
9094 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9095 * config/ia64/ia64.h: Likewise.
9096 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9097 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9099 * config/mmix/mmix.h: Likewise.
9100 * config/nds32/nds32.c: Likewise.
9101 * config/stormy/storym16.h: Likewise.
9102 * config/visium/visium.h: Likewise.
9103 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9105 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9107 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9108 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9109 FRAME_POINTER_REGNUM point at high end of local var area.
9111 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9113 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9114 Move comment around. Do not reset best_edge for a copiable
9115 destination if the copy would cause a partition change.
9116 (better_edge_p): Remove redundant check.
9118 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9120 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9122 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9124 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9125 for math functions that have _Float<N> and _Float<N>X variants.
9126 (mathfn_built_in_2): Add support for math functions that have
9127 _Float<N> and _Float<N>X variants.
9128 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9129 (expand_builtin_mathfn_ternary): Add support for fma with
9130 _Float<N> and _Float<N>X variants.
9131 (expand_builtin): Likewise.
9132 (fold_builtin_3): Likewise.
9133 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9134 create math function _Float<N> and _Float<N>X variants as external
9136 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9137 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9138 the __builtin_ prefix and if not strict ansi, without the prefix.
9139 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9140 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9141 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9142 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9143 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9144 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9145 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9146 function signatures for fma _Float<N> and _Float<N>X variants.
9147 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9148 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9149 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9150 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9151 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9152 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9153 * gencfn-macros.c (print_case_cfn): Add support for math functions
9154 that have _Float<N> and _Float<N>X variants.
9155 (print_define_operator_list): Likewise.
9156 (fltfn_suffixes): Likewise.
9158 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9159 for math functions that have _Float<N> and _Float<N>X variants.
9160 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9161 and _Float<N>X variants.
9162 (COPYSIGN): Likewise.
9165 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9166 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9168 (integer_valued_read_call_p): Likewise.
9169 * fold-const-call.c (fold_const_call_ss): Likewise.
9170 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9171 _Float<N> and _Float<N>X variants.
9172 (fold_const_call_ssss): Add support for fma _Float<N> and
9173 _Float<N>X variants.
9174 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9175 support for copysign and fma _Float<N> and _Float<N>X variants.
9176 (backprop::process_builtin_call_use): Likewise.
9177 * tree-call-cdce.c (can_test_argument_range); Add support for
9178 sqrt _Float<N> and _Float<N>X variants.
9179 (edom_only_function): Likewise.
9180 (get_no_error_domain): Likewise.
9181 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9182 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9183 copysign _Float<N> and _Float<N>X variants.
9184 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9185 handled by machine independent code.
9186 (FMAF128): Likewise.
9187 * doc/cpp.texi (Common Predefined Macros): Document defining
9188 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9189 fma _Float<N> and _Float<N>X variants.
9191 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9194 * config/i386/i386-modes.def (CCFPU): Remove definition.
9195 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9196 (ix86_cc_modes_compatible): Ditto.
9197 (ix86_expand_carry_flag_compare): Ditto.
9198 (ix86_expand_int_movcc): Ditto.
9199 (ix86_expand_int_addcc): Ditto.
9200 (ix86_reverse_condition): Ditto.
9201 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9202 Return true/false for unordered/ordered fp comparisons.
9203 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9204 (ix86_prepare_fp_compare_args): Update for rename.
9205 (ix86_expand_fp_compare): Update for rename. Generate unordered
9206 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9207 (ix86_expand_sse_compare_and_jump): Ditto.
9208 * config/i386/predicates.md (fcmov_comparison_operator):
9209 Remove CCFPU mode handling.
9210 (ix86_comparison_operator): Ditto.
9211 (ix86_carry_flag_operator): Ditto.
9212 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9213 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9214 (*cmpu<mode>_cc_i387): Ditto.
9215 (FPCMP): Remove mode iterator.
9216 (unord): Remove mode attribute.
9217 (unord_subst): New define_subst transformation
9218 (unord): New define_subst attribute.
9220 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9221 (*cmpi<unord>xf_i387): Ditto.
9222 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9223 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9224 using unord_subst transformation.
9225 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9226 (round_saeonly): Also handle CCFP mode.
9227 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9228 Remove UNSPEC_SAHF unspec handling.
9230 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9232 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9234 2017-10-27 Jeff Law <law@redhat.com>
9236 * gimple-ssa-sprintf.c: Include domwalk.h.
9237 (class sprintf_dom_walker): New class, derived from dom_walker.
9238 (sprintf_dom_walker::before_dom_children): New function.
9239 (struct call_info): Moved into sprintf_dom_walker class
9240 (compute_formath_length, handle_gimple_call): Likewise.
9241 (sprintf_length::execute): Call the dominator walker rather
9242 than walking the statements.
9244 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9245 gimple statement locations.
9246 (check_array_bounds): Corresponding changes. Get the statement's
9247 location directly from wi->stmt.
9249 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9252 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9254 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9256 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9257 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9259 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9262 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9263 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9264 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9265 maybe_get_pool_constant.
9266 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9269 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9271 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9272 2.26 caveat. Update gas and gld versions.
9273 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9276 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9278 * cgraph.h (set_malloc_flag): Declare.
9279 * cgraph.c (set_malloc_flag_1): New function.
9280 (set_malloc_flag): Likewise.
9281 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9282 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9284 (read_ipa_call_summary): Add support for reading is_return_callee.
9285 (write_ipa_call_summary): Stream is_return_callee.
9286 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9287 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9288 ipa-prop.h, ipa-fnsummary.h.
9289 (pure_const_names): Change to static.
9290 (malloc_state_e): Define.
9291 (malloc_state_names): Define.
9292 (funct_state_d): Add field malloc_state.
9293 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9294 (check_retval_uses): New function.
9295 (malloc_candidate_p): Likewise.
9296 (analyze_function): Add support for malloc attribute.
9297 (pure_const_write_summary): Stream malloc_state.
9298 (pure_const_read_summary): Add support for reading malloc_state.
9299 (dump_malloc_lattice): New function.
9300 (propagate_malloc): New function.
9301 (warn_function_malloc): New function.
9302 (ipa_pure_const::execute): Call propagate_malloc and
9303 ipa_free_fn_summary.
9304 (pass_local_pure_const::execute): Add support for malloc attribute.
9305 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9306 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9308 2017-10-27 Martin Liska <mliska@suse.cz>
9310 PR gcov-profile/82457
9311 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9312 for fork-like functions to be properly instrumented.
9314 2017-10-27 Richard Biener <rguenther@suse.de>
9317 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9318 info when we redirected EH.
9320 2017-10-26 Michael Collison <michael.collison@arm.com>
9322 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9324 (<optab>_trunchf<GPI:mode>2: New pattern.
9325 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9326 * config/aarch64/iterators.md (wv): New mode attribute.
9327 (vf, VF): New mode attributes.
9328 (vgp, VGP): New mode attributes.
9329 (s): Update attribute with SImode and DImode prefixes.
9331 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9333 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9334 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9335 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9336 (nios2_option_overide): Initialize it. Don't allow R0-relative
9337 addressing with PIC.
9338 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9339 (nios2_symbolic_constant_p): Likewise.
9340 (nios2_legitimate_address_p): Likewise.
9341 (nios2_r0rel_section_name_p): New.
9342 (nios2_symbol_ref_in_r0rel_data_p): New.
9343 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9344 (r0rel_constant_p): New.
9345 (nios2_print_operand_address): Handle r0rel_constant_p.
9346 (nios2_cdx_narrow_form_p): Likewise.
9347 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9348 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9349 (Nios II Options): Document -mr0rel-sec.
9351 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9353 * config/nios2/nios2.c: Include xregex.h.
9354 (nios2_gprel_sec_regex): New.
9355 (nios2_option_overide): Initialize it. Don't allow GP-relative
9356 addressing with PIC.
9357 (nios2_small_section_name_p): Check for regex match.
9358 * config/nios2/nios2.opt (mgprel-sec=): New option.
9359 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9360 (Nios II Options): Document -mgprel-sec.
9362 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9364 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9366 2017-10-26 Tom de Vries <tom@codesourcery.com>
9368 PR tree-optimization/82707
9369 * gimple.c (gimple_copy): Fix unsharing of
9370 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9372 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9374 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9375 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9376 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9377 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9378 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9379 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9380 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9381 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9382 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9383 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9384 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9385 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9386 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9387 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9388 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9389 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9390 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9392 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9394 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9396 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9397 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9398 warning to rs6000.c. Remove the Undocumented flag, since it has
9400 (-mabi=ibmlongdouble): Likewise.
9401 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9402 already set, set the default format for long double.
9403 (rs6000_debug_reg_global): Print whether long double is IBM or
9405 (rs6000_option_override_internal): Rework setting long double
9406 format. Only warn if the user is changing the long double default
9407 and they did not use -Wno-psabi.
9408 * doc/invoke.texi (PowerPC options): Update the documentation for
9409 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9411 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9412 Alan Hayward <alan.hayward@arm.com>
9413 David Sherwood <david.sherwood@arm.com>
9415 * rtl.h (wider_subreg_mode): New function.
9416 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9417 rather than an unsigned int *.
9418 * ira-color.c (regno_max_ref_width): Replace with...
9419 (regno_max_ref_mode): ...this new variable.
9420 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9421 Use wider_subreg_mode.
9422 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9423 rather than an unsigned int *.
9424 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9425 (process_alt_operands): Likewise.
9426 (invariant_p): Likewise.
9427 * lra-spills.c (assign_mem_slot): Likewise.
9428 (add_pseudo_to_slot): Likewise.
9429 * lra.c (collect_non_operand_hard_regs): Likewise.
9430 (add_regs_to_insn_regno_info): Likewise.
9431 * reload1.c (regno_max_ref_width): Replace with...
9432 (regno_max_ref_mode): ...this new variable.
9433 (reload): Update accordingly. Update call to
9434 ira_sort_regnos_for_alter_reg.
9435 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9436 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9437 (scan_paradoxical_subregs): Likewise.
9439 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9441 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9442 (aarch64_frame): Add emit_frame_chain boolean.
9443 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9444 Move eh_return case to aarch64_layout_frame.
9445 (aarch64_layout_frame): Initialize emit_frame_chain.
9446 (aarch64_expand_prologue): Use emit_frame_chain.
9448 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9450 * config/aarch64/aarch64.c (aarch64_layout_frame):
9451 Ensure LR is always stored at the bottom of the callee-saves.
9452 Remove rarely used frame layout which saves callee-saves at top of
9453 frame, so the store of LR can be used as a valid probe in all cases.
9455 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9457 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9458 Improve unaligned TImode/TFmode base/offset split.
9460 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9461 Alan Hayward <alan.hayward@arm.com>
9462 David Sherwood <david.sherwood@arm.com>
9464 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9465 * combine.c (find_single_use_1): Likewise.
9466 (expand_field_assignment): Likewise.
9467 (move_deaths): Likewise.
9468 * lra-constraints.c (simplify_operand_subreg): Likewise.
9469 (curr_insn_transform): Likewise.
9470 * lra.c (collect_non_operand_hard_regs): Likewise.
9471 (add_regs_to_insn_regno_info): Likewise.
9472 * rtlanal.c (reg_referenced_p): Likewise.
9473 (covers_regno_no_parallel_p): Likewise.
9475 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9477 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9478 Don't print any bits outside the precision of the value.
9479 * wide-int.cc (test_printing): Add some new tests.
9481 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9483 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9484 supports -xbrace_comment option.
9485 * configure: Regenerate.
9486 * config.in: Regenerate.
9487 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9488 (ASM_CPU_SPEC): Use it.
9490 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9492 * target.def (static_rtx_alignment): New hook.
9493 * targhooks.h (default_static_rtx_alignment): Declare.
9494 * targhooks.c (default_static_rtx_alignment): New function.
9495 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9496 * doc/tm.texi: Regenerate.
9497 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9498 instead of targetm.constant_alignment. Remove call to
9500 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9501 (cris_preferred_mininum_alignment): New function, split out from...
9502 (cris_constant_alignment): ...here.
9503 (cris_static_rtx_alignment): New function.
9504 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9506 (ix86_constant_alignment): ...here.
9507 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9508 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9509 (mmix_static_rtx_alignment): New function.
9510 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9511 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9513 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9516 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9517 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9519 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9521 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9522 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9523 mark as a sign-extending load.
9524 (local_pic_load_u): Define.
9526 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9529 * fold-const.c (operand_equal_for_comparison_p): Also return true
9530 if ARG0 is a simple variant of ARG1 with narrower precision.
9531 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9533 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9535 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9537 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9538 scatter_static, scatter_per_elt.
9539 * x86-tune-costs.h: Add new cost entries.
9541 2017-10-25 Richard Biener <rguenther@suse.de>
9543 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9544 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9545 class pass_fre): Move to ...
9546 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9547 class pass_fre): ... here and adjust for statistics.
9549 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9552 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9553 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9554 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9555 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9558 2017-10-25 Richard Biener <rguenther@suse.de>
9560 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9561 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9562 eliminate_push_avail, eliminate_insert): Move inside...
9563 (class eliminate_dom_walker): ... this class in preparation
9565 (fini_eliminate): Remove by merging with ...
9566 (eliminate): ... this function. Adjust for class changes.
9567 (pass_pre::execute): Remove fini_eliminate call.
9568 (pass_fre::execute): Likewise.
9570 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9573 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9574 (VPERMI2, VPERMI2I): New mode iterators.
9575 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9576 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9578 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9579 mode iterator. Remove 3 old define_insn patterns.
9580 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9581 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9582 VPERMI2 mode iterator, remove the other two expanders.
9583 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9584 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9585 instructions, remove the other two patterns.
9586 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9587 mode iterator, remove the other two patterns.
9588 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9589 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9590 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9591 and adjust argument order accordingly.
9592 (ix86_expand_vec_perm): Adjust caller.
9593 (expand_vec_perm_1): Likewise.
9594 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9595 (expand_vec_perm_vpermt2_vpshub2): ... this.
9596 (ix86_expand_vec_perm_const_1): Adjust caller.
9597 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9600 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9601 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9602 (vec_shl_<mode>): Remove unused expander.
9603 (avx512bw_<shift_insn><mode>3): New define_insn.
9604 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9605 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9607 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9610 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9613 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9615 PR rtl-optimization/82396
9616 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9617 (autopref_multipass_init): Simplify initialization.
9618 (autopref_rank_data): Simplify sort order.
9619 * gcc/sched-int.h (autopref_multipass_data_): Remove
9620 multi_mem_insn_p, min_offset and max_offset.
9622 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9625 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9626 Check special value of flag_omit_frame_pointer.
9627 (aarch64_can_eliminate): Likewise.
9628 (aarch64_override_options_after_change_1): Simplify handling of
9629 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9631 2017-10-24 Richard Biener <rguenther@suse.de>
9633 PR tree-optimization/82697
9634 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9635 zero for conditional load and unconditional store.
9637 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9639 * doc/install.texi: Document bootstrap-cet.
9641 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9644 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9645 ENDBR instruction at function entrance if function is only
9648 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9651 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9652 patterns to better describe from which operation the CF is computed.
9653 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9654 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9655 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9656 is 0, use _0 suffixed expanders instead of emitting a comparison
9659 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9661 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9662 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9664 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9667 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9668 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9669 * loop-iv.c (iv_get_reaching_def): Likewise.
9670 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9671 variable is promoted and the partition contains undefined values.
9673 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9675 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9677 (nios2_address_cost): Define.
9678 (nios2_legitimize_address): Recognize (exp + constant) directly.
9679 (TARGET_ADDRESS_COST): Define.
9681 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9683 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9684 (nios2_symbolic_memory_operand_p): Declare.
9685 (nios2_split_large_constant): Declare.
9686 (nios2_split_symbolic_memory_operand): Declare.
9687 * config/nios2/nios2.c: Adjust includes.
9688 (nios2_symbolic_constant_allowed): New.
9689 (nios2_symbolic_constant_p): New.
9690 (nios2_plus_symbolic_constant_p): New.
9691 (nios2_valid_addr_expr_p): Recognize addresses involving
9693 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9694 (nios2_symbolic_memory_operand_p): New.
9695 (nios2_large_constant_p): New.
9696 (nios2_split_large_constant): New.
9697 (nios2_split_plus_large_constant): New.
9698 (nios2_split_symbolic_memory_operand): New.
9699 (nios2_legitimize_address): Code refactoring. Handle addresses
9700 involving symbolic constants.
9701 (nios2_emit_move_sequence): Likewise.
9702 (nios2_print_operand): Improve error output.
9703 (nios2_print_operand_address): Handle LO_SUM.
9704 (nios2_cdx_narrow_form_p): Likewise.
9705 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9706 operands involving symbolic constants.
9707 (movhi_internal, movsi_internal): Likewise.
9708 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9709 (extendhisi2, extendqi<mode>2): Likewise.
9711 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9713 * tree-pass.h (PROP_rtl_split_insns): Define.
9714 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9716 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9718 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9720 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9723 * target.def (const_not_ok_for_debug_p): Default to
9724 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9725 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9726 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9727 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9728 which targetm.const_not_ok_for_debug_p returned true.
9729 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9731 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9733 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9734 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9735 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9737 (ix86_const_not_ok_for_debug_p): New function.
9738 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9739 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9741 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9744 * system.h: Conditionally include "unique-ptr.h" if
9745 INCLUDE_UNIQUE_PTR is defined.
9746 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9747 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9749 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9751 * config/rl78/rl78.md: New define_expand "subdi3".
9753 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9756 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9757 DF_REF_INSN if DF_REF_INSN_INFO is false.
9759 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9761 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9762 xmm_move instead of sse_move.
9763 (sse_store_index): New function.
9764 (ix86_register_move_cost): Be more sensible about mismatch stall;
9765 model AVX moves correctly; make difference between sse->integer and
9767 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9768 moves; make difference between SSE and AVX.
9769 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9770 and zmm_move. Increase size of sse load and store tables;
9771 add unaligned load and store tables; add ssemmx_to_integer.
9772 * x86-tune-costs.h: Update all entries according to real
9773 move latencies from Agner Fog's manual and chip documentation.
9775 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9778 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9779 * config/i386/constraints.md (Wf): New constraint.
9780 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9781 (cmp<dwi>_doubleword): Removed.
9782 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9783 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9784 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9785 expand with cmp<dwi>_doubleword. For LTU and GEU use
9786 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9788 * common.opt (gcolumn-info): Enable by default.
9789 * doc/invoke.texi (gcolumn-info): Document new default.
9791 2017-10-23 Richard Biener <rguenther@suse.de>
9793 PR tree-optimization/82672
9794 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9795 Fold the stmt if we propagated into it.
9797 2017-10-23 Richard Biener <rguenther@suse.de>
9799 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9800 (bitmap_remove_expr_from_set): ... this. All callers call this
9801 for non-constant values.
9802 (bitmap_set_subtract): Rename to...
9803 (bitmap_set_subtract_expressions): ... this. Adjust and
9805 (bitmap_set_contains_value): Remove superfluous check.
9806 (bitmap_set_replace_value): Inline into single caller ...
9807 (bitmap_value_replace_in_set): ... here and simplify.
9808 (dependent_clean): Merge into ...
9809 (clean): ... this using an overload. Adjust.
9810 (prune_clobbered_mems): Adjust.
9811 (compute_antic_aux): Likewise.
9812 (compute_partial_antic_aux): Likewise.
9814 2017-10-23 Richard Biener <rguenther@suse.de>
9816 PR tree-optimization/82129
9818 2017-08-01 Richard Biener <rguenther@suse.de>
9820 PR tree-optimization/81181
9821 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9822 (compute_antic): ... end of iteration here.
9824 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9826 * target.def (starting_frame_offset): New hook.
9827 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9828 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9829 * doc/tm.texi.in: Regenerate.
9830 * hooks.h (hook_hwi_void_0): Declare.
9831 * hooks.c (hook_hwi_void_0): New function.
9832 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9833 STARTING_FRAME_OFFSET.
9834 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9835 * reload1.c (reload): Likewise.
9836 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9837 instead of STARTING_FRAME_OFFSET.
9838 * function.c (try_fit_stack_local): Likewise.
9839 (assign_stack_local_1): Likewise
9840 (instantiate_virtual_regs): Likewise.
9841 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9842 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9843 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9844 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9845 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9846 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9847 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9848 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9849 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9850 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9851 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9852 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9853 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9854 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9855 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9856 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9857 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9858 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
9859 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
9860 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
9861 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
9862 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
9863 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
9864 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
9865 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
9866 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
9867 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
9868 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
9869 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
9870 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
9871 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
9872 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
9873 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
9874 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
9875 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
9876 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
9877 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
9878 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
9879 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
9880 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
9881 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
9882 * config/avr/avr.c (avr_starting_frame_offset): Make static and
9883 return a HOST_WIDE_INT.
9884 (avr_builtin_setjmp_frame_value): Use it instead of
9885 STARTING_FRAME_OFFSET.
9886 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9887 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
9888 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
9890 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9891 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
9892 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
9893 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9894 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
9895 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
9896 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9897 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
9898 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
9899 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9900 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
9901 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
9903 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9904 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
9905 * config/mips/mips.c (mips_compute_frame_info): Refer to
9906 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9907 (mips_starting_frame_offset): New function.
9908 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9909 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
9910 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
9911 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
9912 and return a HOST_WIDE_INT.
9913 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9914 (mmix_initial_elimination_offset): Refer to
9915 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9916 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
9917 * config/pa/pa.c (pa_starting_frame_offset): New function.
9918 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
9919 (pa_expand_prologue): Likewise.
9920 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9921 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
9922 !FRAME_GROWS_DOWNWARD handling to...
9923 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9924 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
9925 !FRAME_GROWS_DOWNWARD handling to...
9926 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9927 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
9928 !FRAME_GROWS_DOWNWARD handling to...
9929 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9930 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
9932 (rs6000_starting_frame_offset): New function.
9933 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
9934 !FRAME_GROWS_DOWNWARD handling to...
9935 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9936 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
9937 !FRAME_GROWS_DOWNWARD handling to...
9938 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9939 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
9940 !FRAME_GROWS_DOWNWARD handling to...
9941 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9942 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
9943 (rs6000_starting_frame_offset): New function.
9944 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
9945 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
9946 * config/vax/vax.c (vax_starting_frame_offset): New function.
9947 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
9948 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9949 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
9950 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
9951 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9952 * system.h (STARTING_FRAME_OFFSET): Poison.
9954 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9956 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9957 SCALAR_TYPE_MODE instead of TYPE_MODE.
9959 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9960 Alan Hayward <alan.hayward@arm.com>
9961 David Sherwood <david.sherwood@arm.com>
9963 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
9965 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9966 Alan Hayward <alan.hayward@arm.com>
9967 David Sherwood <david.sherwood@arm.com>
9969 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
9971 2017-10-23 Richard Biener <rguenther@suse.de>
9973 PR tree-optimization/82129
9974 * tree-ssa-pre.c (bitmap_set_and): Remove.
9975 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
9976 canonicalizing expressions in the set to those with lowest
9977 ID rather than taking that from the first edge.
9979 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9981 * combine.c (rtx_equal_for_field_assignment_p): Use
9982 byte_lowpart_offset.
9984 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9985 Alan Hayward <alan.hayward@arm.com>
9986 David Sherwood <david.sherwood@arm.com>
9988 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
9989 to a SUBREG_PROMOTED_VAR.
9991 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9992 Alan Hayward <alan.hayward@arm.com>
9993 David Sherwood <david.sherwood@arm.com>
9995 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
9996 (expand_debug_source_expr): Likewise.
9997 * combine.c (combine_simplify_rtx): Likewise.
9998 * cse.c (fold_rtx): Likewise.
9999 * optabs.c (expand_float): Likewise.
10000 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10001 (simplify_binary_operation_1): Likewise.
10003 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10004 Alan Hayward <alan.hayward@arm.com>
10005 David Sherwood <david.sherwood@arm.com>
10007 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10008 (record_promoted_value): Likewise.
10009 * expr.c (expand_expr_real_2): Likewise.
10010 * ree.c (update_reg_equal_equiv_notes): Likewise.
10011 (combine_set_extension): Likewise.
10012 * rtlanal.c (low_bitmask_len): Likewise.
10013 * simplify-rtx.c (neg_const_int): Likewise.
10014 (simplify_binary_operation_1): Likewise.
10016 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10017 Alan Hayward <alan.hayward@arm.com>
10018 David Sherwood <david.sherwood@arm.com>
10020 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10021 * regcprop.c (maybe_mode_change): Likewise.
10022 * reload1.c (alter_reg): Likewise.
10024 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10026 * inchash.h (inchash::hash::add_wide_int): New function.
10027 * lto-streamer-out.c (hash_tree): Use it.
10029 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10031 * inchash.h (inchash::hash::add_wide_int): Rename to...
10032 (inchash::hash::add_hwi): ...this.
10033 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10034 (polymorphic_call_target_hasher::hash): Likewise.
10035 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10036 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10037 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10038 * lto-streamer-out.c (hash_tree): Likewise.
10039 * optc-save-gen.awk: Likewise.
10040 * tree.c (add_expr): Likewise.
10042 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10045 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10046 for ordered inequality comparisons even with TARGET_IEEE_FP.
10048 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10051 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10052 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10053 Expand with cmp<dwi>_doubleword.
10055 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10057 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10058 List CET intrinsics.
10059 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10060 specific to -fcf-protection option.
10062 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10064 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10065 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10066 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10067 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10068 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10069 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10070 (extra_objs): Add cet.o for Linux/x86 targets.
10071 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10072 * config/i386/cet.c: New file.
10073 * config/i386/cetintrin.h: Likewise.
10074 * config/i386/t-cet: Likewise.
10075 * config/i386/cpuid.h (bit_SHSTK): New.
10076 (bit_IBT): Likewise.
10077 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10078 pass IBT and SHSTK bits.
10079 * config/i386/i386-builtin-types.def
10080 (VOID_FTYPE_UNSIGNED_PVOID): New.
10081 (VOID_FTYPE_UINT64_PVOID): Likewise.
10082 * config/i386/i386-builtin.def: Add CET intrinsics.
10083 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10084 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10085 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10086 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10088 * config/i386/i386.c (rest_of_insert_endbranch): New.
10089 (pass_data_insert_endbranch): Likewise.
10090 (pass_insert_endbranch): Likewise.
10091 (make_pass_insert_endbranch): Likewise.
10092 (ix86_notrack_prefixed_insn_p): Likewise.
10093 (ix86_target_string): Add -mibt, -mshstk flags.
10094 (ix86_option_override_internal): Add flag_cf_protection
10096 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10097 (ix86_print_operand): Add 'notrack' prefix output.
10098 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10099 (ix86_expand_builtin): Expand CET intrinsics.
10100 (x86_output_mi_thunk): Add 'endbranch' instruction.
10101 * config/i386/i386.h (TARGET_IBT): New.
10102 (TARGET_IBT_P): Likewise.
10103 (TARGET_SHSTK): Likewise.
10104 (TARGET_SHSTK_P): Likewise.
10105 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10106 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10107 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10108 (builtin_setjmp_setup): New pattern.
10109 (builtin_longjmp): Likewise.
10110 (rdssp<mode>): Likewise.
10111 (incssp<mode>): Likewise.
10112 (saveprevssp): Likewise.
10113 (rstorssp): Likewise.
10114 (wrss<mode>): Likewise.
10115 (wruss<mode>): Likewise.
10116 (setssbsy): Likewise.
10117 (clrssbsy): Likewise.
10118 (nop_endbr): Likewise.
10119 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10121 * config/i386/immintrin.h: Include <cetintrin.h>.
10122 * config/i386/linux-common.h
10123 (file_end_indicate_exec_stack_and_cet): New prototype.
10124 (TARGET_ASM_FILE_END): New.
10126 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10128 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10129 latencies instead of having separate table; make difference between
10130 integer and float costs.
10131 * i386.h (processor_costs): Remove scalar_stmt_cost,
10132 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10133 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10135 * x86-tune-costs.h: Remove entries which has been removed in
10136 procesor_costs from all tables; make cond_taken_branch_cost
10137 and cond_not_taken_branch_cost COST_N_INSNS based.
10139 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10141 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10143 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10145 * config/i386/i386.md (isa): Remove fma_avx512f.
10146 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10147 <avx512>_fmadd_<mode>_mask3<round_name>,
10148 <avx512>_fmsub_<mode>_mask<round_name>,
10149 <avx512>_fmsub_<mode>_mask3<round_name>,
10150 <avx512>_fnmadd_<mode>_mask<round_name>,
10151 <avx512>_fnmadd_<mode>_mask3<round_name>,
10152 <avx512>_fnmsub_<mode>_mask<round_name>,
10153 <avx512>_fnmsub_<mode>_mask3<round_name>,
10154 <avx512>_fmaddsub_<mode>_mask<round_name>,
10155 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10156 <avx512>_fmsubadd_<mode>_mask<round_name>,
10157 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10158 (*vec_widen_umult_even_v16si<mask_name>,
10159 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10160 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10162 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10164 * extend.texi: Add 'nocf_check' documentation.
10165 * gimple.texi: Add second parameter to
10166 gimple_build_call_from_tree.
10167 * invoke.texi: Add -fcf-protection documentation.
10168 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10170 2017-10-20 Richard Biener <rguenther@suse.de>
10172 PR tree-optimization/82473
10173 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10174 the largest input type.
10176 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10178 * c-attribs.c (handle_nocf_check_attribute): New function.
10179 (c_common_attribute_table): Add 'nocf_check' handling.
10180 * gimple-parser.c: Add second argument NULL to
10181 gimple_build_call_from_tree.
10182 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10183 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10185 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10186 * common.opt: Add fcf-protection flag.
10187 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10188 * flag-types.h: Add enum cf_protection_level.
10189 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10190 Add 'nocf_check' attribute propagation to gimple call.
10191 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10192 (gimple_build_call_from_tree): Update prototype.
10193 (gimple_call_nocf_check_p): New function.
10194 (gimple_call_set_nocf_check): Likewise.
10195 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10196 * ipa-icf.c: Add nocf_check attribute in statement hash.
10197 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10198 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10199 * toplev.c (process_options): Add flag_cf_protection handling.
10201 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10203 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10205 2017-10-20 Richard Biener <rguenther@suse.de>
10207 PR tree-optimization/82603
10208 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10209 remove false predicated stores.
10211 2017-10-20 Richard Biener <rguenther@suse.de>
10213 * graphite-isl-ast-to-gimple.c
10214 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10215 Remove return value and simplify, dump copied stmt after lhs
10217 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10218 Reduce dump verbosity.
10219 (gsi_insert_earliest): Likewise.
10220 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10221 * graphite.c (print_global_statistics): Adjust dumping.
10222 (print_graphite_scop_statistics): Likewise.
10223 (print_graphite_statistics): Do not dump loops here.
10224 (graphite_transform_loops): But here.
10226 2017-10-20 Nicolas Roche <roche@adacore.com>
10228 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10229 * configure: Regenerate.
10231 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10234 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10235 functions when optimizing replace GIMPLE_RETURN stmts with
10236 calls to __builtin_unreachable ().
10239 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10240 for -fsanitize=thread link of executables.
10241 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10242 link of executables.
10245 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10246 New mode iterators.
10247 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10248 define_insns for logical vector shifts to use VI248_AVX512BW
10249 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10250 condition, useless isa and prefix attributes. Change the first
10251 2 of these define_insns to ...
10252 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10253 define_insn for avx512vl.
10254 (<shift_insn><mode>3): ... and this, new define_insn without
10255 masking for non-avx512vl.
10258 * config/i386/sse.md (*andnot<mode>3,
10259 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10260 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10261 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10262 not applied use empty suffix even for TARGET_AVX512VL.
10263 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10264 is applied, supply evex,evex or evex,evex,evex instead of just
10267 2017-10-20 Julia Koval <julia.koval@intel.com>
10269 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10270 (OPTION_MASK_ISA_GFNI_UNSET): New.
10271 (ix86_handle_option): Handle OPT_mgfni.
10272 * config/i386/cpuid.h (bit_GFNI): New.
10273 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10274 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10275 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10276 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10277 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10278 * config/i386/i386.opt: Add mgfni.
10280 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10282 * config/msp430/msp430.c (msp430_option_override): Disable
10283 -fdelete-null-pointer-checks.
10284 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10286 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10288 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10289 of x87 and SSE instructions.
10291 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10293 * asan.c (create_cond_insert_point): Do not update edge count.
10294 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10295 (afdo_propagate_circuit): Likewise.
10296 (afdo_calculate_branch_prob): Likewise.
10297 (afdo_annotate_cfg): Likewise.
10298 * basic-block.h (struct edge_def): Remove count.
10299 (edge_def::count): New accessor.
10300 * bb-reorder.c (rotate_loop): Update.
10301 (find_traces_1_round): Update.
10302 (connect_traces): Update.
10303 (sanitize_hot_paths): Update.
10304 * cfg.c (unchecked_make_edge): Update.
10305 (make_single_succ_edge): Update.
10306 (check_bb_profile): Update.
10307 (dump_edge_info): Update.
10308 (update_bb_profile_for_threading): Update.
10309 (scale_bbs_frequencies_int): Update.
10310 (scale_bbs_frequencies_gcov_type): Update.
10311 (scale_bbs_frequencies_profile_count): Update.
10312 (scale_bbs_frequencies): Update.
10313 * cfganal.c (connect_infinite_loops_to_exit): Update.
10314 * cfgbuild.c (compute_outgoing_frequencies): Update.
10315 (find_many_sub_basic_blocks): Update.
10316 * cfgcleanup.c (try_forward_edges): Update.
10317 (try_crossjump_to_edge): Update
10318 * cfgexpand.c (expand_gimple_cond): Update
10319 (expand_gimple_tailcall): Update
10320 (construct_exit_block): Update
10321 * cfghooks.c (verify_flow_info): Update
10322 (redirect_edge_succ_nodup): Update
10323 (split_edge): Update
10324 (make_forwarder_block): Update
10325 (duplicate_block): Update
10326 (account_profile_record): Update
10327 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10328 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10329 * cfgloopmanip.c (scale_loop_profile): Update.
10331 (lv_adjust_loop_entry_edge): Update.
10332 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10333 (force_nonfallthru_and_redirect): Update.
10334 (purge_dead_edges): Update.
10335 (rtl_flow_call_edges_add): Update.
10336 * cgraphunit.c (init_lowered_empty_function): Update.
10337 (cgraph_node::expand_thunk): Update.
10338 * gimple-pretty-print.c (dump_probability): Update.
10339 (dump_edge_probability): Update.
10340 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10341 * haifa-sched.c (sched_create_recovery_edges): Update.
10342 * hsa-gen.c (convert_switch_statements): Update.
10343 * ifcvt.c (dead_or_predicable): Update.
10344 * ipa-inline-transform.c (inline_transform): Update.
10345 * ipa-split.c (split_function): Update.
10346 * ipa-utils.c (ipa_merge_profiles): Update.
10347 * loop-doloop.c (add_test): Update.
10348 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10349 * lto-streamer-in.c (input_cfg): Update.
10350 (input_function): Update.
10351 * lto-streamer-out.c (output_cfg): Update.
10352 * modulo-sched.c (sms_schedule): Update.
10353 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10354 * predict.c (maybe_hot_edge_p): Update.
10355 (unlikely_executed_edge_p): Update.
10356 (probably_never_executed_edge_p): Update.
10357 (dump_prediction): Update.
10358 (drop_profile): Update.
10359 (propagate_unlikely_bbs_forward): Update.
10360 (determine_unlikely_bbs): Update.
10361 (force_edge_cold): Update.
10362 * profile.c (compute_branch_probabilities): Update.
10363 * reg-stack.c (better_edge): Update.
10364 * shrink-wrap.c (handle_simple_exit): Update.
10365 * tracer.c (better_p): Update.
10366 * trans-mem.c (expand_transaction): Update.
10367 (split_bb_make_tm_edge): Update.
10368 * tree-call-cdce.c: Update.
10369 * tree-cfg.c (gimple_find_sub_bbs): Update.
10370 (gimple_split_edge): Update.
10371 (gimple_duplicate_sese_region): Update.
10372 (gimple_duplicate_sese_tail): Update.
10373 (gimple_flow_call_edges_add): Update.
10374 (insert_cond_bb): Update.
10375 (execute_fixup_cfg): Update.
10376 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10377 * tree-complex.c (expand_complex_div_wide): Update.
10378 * tree-eh.c (lower_resx): Update.
10379 (unsplit_eh): Update.
10380 (cleanup_empty_eh_move_lp): Update.
10381 * tree-inline.c (copy_edges_for_bb): Update.
10382 (freqs_to_counts): Update.
10383 (copy_cfg_body): Update.
10384 * tree-ssa-dce.c (remove_dead_stmt): Update.
10385 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10386 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10387 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10388 (unloop_loops): Update.
10389 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10390 * tree-ssa-loop-split.c (connect_loops): Update.
10391 (split_loop): Update.
10392 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10393 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10394 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10395 * tree-ssa-reassoc.c (branch_fixup): Update.
10396 * tree-ssa-tail-merge.c (replace_block_by): Update.
10397 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10398 (compute_path_counts): Update.
10399 (update_profile): Update.
10400 (recompute_probabilities): Update.
10401 (update_joiner_offpath_counts): Update.
10402 (estimated_freqs_path): Update.
10403 (freqs_to_counts_path): Update.
10404 (clear_counts_path): Update.
10405 (ssa_fix_duplicate_block_edges): Update.
10406 (duplicate_thread_path): Update.
10407 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10408 (case_bit_test_cmp): Update.
10409 (collect_switch_conv_info): Update.
10410 (gen_inbound_check): Update.
10411 (do_jump_if_equal): Update.
10412 (emit_cmp_and_jump_insns): Update.
10413 * tree-tailcall.c (decrease_profile): Update.
10414 (eliminate_tail_call): Update.
10415 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10416 (vect_do_peeling): Update.
10417 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10418 * ubsan.c (ubsan_expand_null_ifn): Update.
10419 (ubsan_expand_ptr_ifn): Update.
10420 * value-prof.c (gimple_divmod_fixed_value): Update.
10421 (gimple_mod_pow2): Update.
10422 (gimple_mod_subtract): Update.
10423 (gimple_ic): Update.
10424 (gimple_stringop_fixed_value): Update.
10426 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10429 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10431 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10433 PR rtl-optimization/82395
10434 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10435 based on non_spilled_static_chain_regno_p.
10437 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10439 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10440 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10441 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10443 2017-10-19 Martin Sebor <msebor@redhat.com>
10445 PR tree-optimization/82596
10446 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10448 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10450 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10451 (get_mem_refs_of_builtin_call): Likewise.
10452 * builtins.c (expand_builtin_apply): Adjust call to
10453 allocate_dynamic_stack_space.
10454 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10455 the third argument to allocate_dynamic_stack_space, otherwise -1.
10456 (expand_builtin): Deal with all alloca variants.
10457 (is_inexpensive_builtin): Likewise.
10458 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10459 * calls.c (special_function_p): Deal with all alloca variants.
10460 (initialize_argument_information): Adjust call to
10461 allocate_dynamic_stack_space.
10462 (expand_call): Likewise.
10463 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10464 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10465 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10466 use it for the stack usage computation.
10467 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10468 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10469 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10470 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10471 (in_loop_p): Remove first argument and useless check.
10472 (pass_walloca::execute): Remove useless test and adjust call to above.
10473 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10474 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10475 (gimplify_call_expr): Deal with all alloca variants.
10476 * hsa-gen.c (gen_hsa_alloca): Likewise.
10477 (gen_hsa_insns_for_call): Likewise.
10478 * ipa-pure-const.c (special_builtin_state): Likewise.
10479 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10480 * tree-object-size.c (alloc_object_size): Likewise.
10481 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10482 (call_may_clobber_ref_p_1): Likewise.
10483 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10484 (ccp_fold_stmt): Likewise.
10485 (optimize_stack_restore): Likewise.
10486 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10487 (mark_all_reaching_defs_necessary_1): Likewise.
10488 (propagate_necessity): Likewise.
10489 (eliminate_unnecessary_stmts): Likewise.
10490 * tree.c (build_common_builtin_nodes): Build
10491 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10492 (build_alloca_call_expr): New function.
10493 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10494 (CASE_BUILT_IN_ALLOCA): Likewise.
10495 (build_alloca_call_expr): Declare.
10496 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10498 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10501 * dwarf2out.c (new_die_raw): New static inline function.
10502 (new_die): Use it to create the DIE.
10503 (add_AT_external_die_ref): Likewise.
10504 (clone_die): Likewise.
10505 (clone_as_declaration): Likewise.
10506 (dwarf2out_vms_debug_main_pointer): Likewise.
10507 (base_type_die): Likewise. Remove early return for corner cases.
10508 Do not call add_pubtype on the DIE here.
10509 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10510 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10511 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10512 native order exists for base types, attach the DIE manually and call
10513 add_pubtype on it. Do not equate a reverse order DIE to the type.
10515 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10517 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10518 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10520 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10522 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10524 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10525 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10526 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10527 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10528 * opts.c (sanitizer_opts): Add builtin.
10529 * ubsan.c (instrument_builtin): New function.
10530 (pass_ubsan::execute): Call it.
10531 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10532 * doc/invoke.texi: Document -fsanitize=builtin.
10534 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10535 builtins, store max (log2 (align), 0) into uchar field instead of
10536 align into uptr field.
10537 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10538 store uchar 0 field instead of uptr 0 field.
10539 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10540 instead of passing one address of struct with 2 locations pass
10541 two addresses of structs with 1 location each.
10542 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10543 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10544 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10545 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10546 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10547 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10548 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10549 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10551 2017-10-19 Martin Liska <mliska@suse.cz>
10554 * file-find.c (remove_prefix): Remove.
10555 * file-find.h (remove_prefix): Likewise.
10556 * gcc-ar.c: Remove smartness of lookup.
10558 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10560 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10561 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10562 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10564 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10567 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10568 (setcc + and to xor + setcc): New peephole2.
10570 2017-10-19 Tom de Vries <tom@codesourcery.com>
10572 * doc/sourcebuild.texi (Test Directives, Variants of
10573 dg-require-support): Add dg-require-stack-size.
10575 2017-10-19 Martin Liska <mliska@suse.cz>
10578 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10579 that have a large alignment.
10580 (gimplify_target_expr): Likewise.
10582 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10584 PR rtl-optimization/82602
10585 * ira.c (rtx_moveable_p): Return false for volatile asm.
10587 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10590 * config/i386/i386-modes.def (CCGZ): New CC mode.
10591 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10592 * config/i386/predicates.md (ix86_comparison_operator):
10594 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10595 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10596 with double-word subtraction.
10597 (put_condition_code): Handle CCGZmode.
10599 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10601 * wide-int.cc (debug (const wide_int &)): New.
10602 (debug (const wide_int *)): New.
10603 (debug (const widest_int &)): New.
10604 (debug (const widest_int *)): New.
10606 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10608 PR middle-end/82556
10609 * lra-constraints.c (curr_insn_transform): Use non-input operand
10610 instead of output one for matched reload.
10612 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10614 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10615 (tree-ssa-loop-ivopts.h): New header file.
10616 (struct builtin_info): New fields.
10617 (classify_builtin_1): Compute and record base and offset parts for
10618 memset builtin partition by calling strip_offset.
10619 (offset_cmp, fuse_memset_builtins): New functions.
10620 (finalize_partitions): Fuse adjacent memset partitions by calling
10622 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10623 Expose the interface.
10624 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10626 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10628 PR tree-optimization/82574
10629 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10630 that data reference must be executed exactly once per iteration
10631 against the outermost loop in nest.
10632 (classify_partition): Update call to above function.
10634 2017-10-18 Richard Biener <rguenther@suse.de>
10636 PR tree-optimization/82591
10637 * graphite.c (graphite_transform_loops): Move code gen message
10639 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10640 Here. Handle scop_to_isl_ast failing.
10641 (scop_to_isl_ast): Limit the number of ISL operations.
10643 2017-10-18 Richard Biener <rguenther@suse.de>
10645 * graphite-isl-ast-to-gimple.c
10646 (translate_isl_ast_to_gimple::set_rename): Simplify.
10647 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10648 (graphite_copy_stmts_from_block): ... here.
10649 (copy_bb_and_scalar_dependences): Simplify.
10650 (add_parameters_to_ivs_params): Canonicalize.
10651 (generate_entry_out_of_ssa_copies): Simplify.
10652 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10653 by passing in ISL dimension.
10654 (parameter_index_in_region_1): Rename to ...
10655 (parameter_index_in_region): ... this.
10656 (extract_affine): Adjust assert, pass down parameter index.
10657 (add_param_constraints): Use range-info when available.
10658 (build_scop_context): Adjust.
10659 * sese.c (new_sese_info): Adjust.
10660 (free_sese_info): Likewise.
10661 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10662 Remove unused typedefs.
10663 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10665 2017-10-18 Martin Liska <mliska@suse.cz>
10667 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10669 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10671 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10672 (s390_sched_init): Do not reset s390_sched_state if we entered the
10673 current basic block via a fallthru edge and all others are unlikely.
10675 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10677 * config/s390/s390.c (NUM_SIDES): New variable.
10678 (LONGRUNNING_THRESHOLD): New variable.
10679 (LATENCY_FACTOR): New variable.
10680 (s390_sched_score): Decrease score for long-running instructions on
10682 (s390_sched_variable_issue): Perform bookkeeping for long-running
10685 2017-10-18 Richard Biener <rguenther@suse.de>
10687 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10688 Simplify with removal of the parameter rename map.
10689 (set_rename): Likewise.
10690 (should_copy_to_new_region): Likewise.
10691 (graphite_copy_stmts_from_block): Likewise.
10692 (copy_bb_and_scalar_dependences): Remove initialization of
10693 unused copied_bb_map.
10694 (copy_def): Remove.
10695 (copy_internal_parameters): Likewise.
10696 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10697 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10698 Use INTEGRAL_TYPE_P.
10699 (parameter_index_in_region_1): Rename to ...
10700 (assign_parameter_index_in_region): ... this. Assert we have
10701 a parameter we handle.
10702 (scan_tree_for_params): Adjust.
10703 * sese.h (parameter_rename_map_t): Remove.
10704 (struct sese_info_t): Remove unused parameter_rename_map and
10705 copied_bb_map members.
10706 * sese.c (new_sese_info): Adjust.
10707 (free_sese_info): Likewise.
10709 2017-10-18 Martin Liska <mliska@suse.cz>
10712 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10713 on an abnormal edge.
10715 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10717 * doc/invoke.texi (ffunction-sections and fdata-sections):
10720 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10722 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10723 the use statement can throw internally.
10725 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10727 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10728 any RTX present on the RHS of a SET.
10729 * compare-elim.c (try_eliminate_compare): Restore comment.
10731 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10733 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10734 may be also called on tcc_type nodes.
10735 * langhooks.c (lhd_tree_size): Likewise.
10737 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10739 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10740 format_warning_at_substring.
10741 (maybe_warn): Convert source_range * param to a location_t. Pass
10742 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10743 (format_directive): Remove code to extract source_ranges and
10744 source_range * in favor of just a location_t.
10745 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10747 * substring-locations.c (format_warning_va): Convert
10748 source_range * param to a location_t.
10749 (format_warning_at_substring): Likewise.
10750 * substring-locations.h (format_warning_va): Likewise.
10751 (format_warning_at_substring): Likewise.
10753 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10755 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10757 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10758 and scatter/gather ops.
10760 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10761 vec_gather_load and vec_scatter_store.
10762 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10763 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10764 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10765 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10766 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10767 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10769 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10771 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10772 Detect FTST instruction and handle its register pops. Only pop
10773 second operand if can_pop_second_op is true.
10774 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10775 set can_pop_second_op to false in the compare_for_stack_reg call.
10777 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10778 output_fp_compare for stack register operands.
10779 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10780 instructions here. Do not emit stack register pops here. Assert
10781 that FCOMPP pops next to top stack register. Rewrite function.
10783 2017-10-17 Nathan Sidwell <nathan@acm.org>
10785 PR middle-end/82577
10786 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10787 use DECL_ASSEMBLER_NAME_RAW.
10789 PR middle-end/82546
10790 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10793 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10794 Wilco Dijkstra <wilco.dijkstra@arm.com>
10796 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10797 converstion to Pmode from the buf_addr.
10799 2017-10-17 Richard Biener <rguenther@suse.de>
10801 * graphite-dependences.c (scop_get_reads_and_writes): Change
10802 output parameters to references.
10804 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10806 PR 71026/tree-optimization
10807 * fold-const.c (distribute_real_division): Removed.
10808 (fold_binary_loc): Remove calls to distribute_real_divison.
10810 2017-10-17 Richard Biener <rguenther@suse.de>
10812 * graphite-scop-detection.c
10813 (scop_detection::stmt_has_simple_data_refs_p): Always use
10814 the full nest as region.
10815 (try_generate_gimple_bb): Likewise.
10816 * sese.c (scalar_evolution_in_region): Simplify now that
10817 SCEV can handle instantiation in regions.
10818 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10819 in the non-loop part of a function if requested.
10821 2017-10-17 Richard Biener <rguenther@suse.de>
10823 PR tree-optimization/82563
10824 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10826 (graphite_regenerate_ast_isl): Call it.
10827 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10829 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10831 PR tree-optimization/82549
10832 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10833 Formatting fixes. Instead of calling make_bit_field_ref with negative
10836 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10838 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10839 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10840 _mm_maskz_reduce_ss): New.
10841 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10842 __builtin_ia32_reducess_mask): Ditto..
10843 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10844 * config/i386/sse.md (reduces<mode>): Renamed to ...
10845 (reduces<mode><mask_scalar_name>): ... this.
10846 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10848 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10849 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10851 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10853 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10854 * selftest-run-tests.c (selftest::run_tests): Call
10855 selftest::unique_ptr_tests_cc_tests.
10856 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10857 * unique-ptr-tests.cc: New file.
10859 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
10862 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10864 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10865 (make_hard_regno_born, make_hard_regno_dead): Update
10866 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
10868 2017-10-16 Jeff Law <law@redhat.com>
10870 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10872 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
10874 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
10876 2017-10-16 Olivier Hainque <hainque@adacore.com>
10878 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
10879 with_cpu if we were configured for an e500v2 target cpu name.
10881 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
10883 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
10884 * doc/invoke.texi: Document +nodsp as a valid extension for
10887 2017-10-16 Martin Liska <mliska@suse.cz>
10889 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
10890 (test_set_range): Likewise.
10891 (test_range_functions): Rename to ...
10892 (test_bit_in_range): ... this.
10893 (sbitmap_c_tests): Add new test.
10895 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10897 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
10899 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
10900 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
10902 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10904 * config/aarch64/aarch64-builtins.c
10905 (aarch64_types_quadopu_lane_qualifiers): New.
10906 (TYPES_QUADOPU_LANE): New.
10907 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
10908 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
10909 (aarch64_<sur>dot_laneq<vsi2qi>): New.
10910 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
10911 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
10912 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
10913 (Vdottype, DOTPROD): New.
10914 (sur): Add SDOT and UDOT.
10916 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10918 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
10919 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
10920 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
10921 Add TARGET_DOTPROD.
10922 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
10923 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
10924 Enable TARGET_DOTPROD.
10925 (cortex-a75.cortex-a55): Likewise.
10926 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
10928 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10930 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
10931 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
10933 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
10935 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
10936 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
10937 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
10938 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
10939 * config/arm/types.md (neon_dot, neon_dot_q): New.
10940 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
10942 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10944 * config/arm/arm.h (TARGET_DOTPROD): New.
10945 * config/arm/arm.c (arm_arch_dotprod): New.
10946 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
10947 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
10948 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
10949 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
10950 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
10951 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
10952 * doc/invoke.texi (armv8.2-a): Document dotprod
10954 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10956 * i386.c (ix86_vec_cost): New function.
10957 (ix86_rtx_costs): Handle vector operations better.
10958 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
10959 * x86-tune-costs.h: Add new costs to all tables.
10961 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10963 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
10965 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
10966 divsd, sqrtss and sqrtsd
10967 * x86-tune-costs.h: Add new entries to all costs.
10968 (znver1_cost): Fix to match real instruction latencies.
10970 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10971 Michael Collison <michael.collison@arm.com>
10973 * compare-elim.c: Include emit-rtl.h.
10974 (can_merge_compare_into_arith): New function.
10975 (try_validate_parallel): Likewise.
10976 (try_merge_compare): Likewise.
10977 (try_eliminate_compare): Call the above when no previous clobber
10979 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
10982 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10984 PR middle-end/62263
10985 PR middle-end/82498
10986 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
10987 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
10989 PR middle-end/62263
10990 PR middle-end/82498
10991 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
10992 to be any operand_equal_p operands. For & (B - 1) require
10993 B to be power of 2. Recognize
10994 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
10996 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10999 * optabs.c (expand_memory_blockage): Fix call of
11000 targetm.have_memory_blockage.
11002 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11005 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11006 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11007 objects to extra_objs instead of overwriting it.
11009 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11011 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11012 Use any_fp_register_operand as operand[3] predicate. Simplify
11013 equality test for operands[2] and operands[4] memory location.
11014 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11015 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11016 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11017 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11018 any_fp_register_operand as operand[1] predicate. Simplify
11019 equality test for operands[0] and operands[3] memory location.
11020 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11021 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11022 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11024 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11026 * target-insns.def: Add memory_blockage.
11027 * optabs.c (expand_memory_blockage): New function.
11028 (expand_asm_memory_barrier): Rename ...
11029 (expand_asm_memory_blockage): ... to this.
11030 (expand_mem_thread_fence): Call expand_memory_blockage
11031 instead of expand_asm_memory_barrier.
11032 (expand_mem_singnal_fence): Ditto.
11033 (expand_atomic_load): Ditto.
11034 (expand_atomic_store): Ditto.
11035 * doc/md.texi (Standard Pattern Names For Generation):
11036 Document memory_blockage instruction pattern.
11038 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11040 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11041 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11042 * config/rl78/rl78.md: New define_expand "adddi3".
11044 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11046 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11049 2017-10-13 Jeff Law <law@redhat.com>
11051 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11053 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11056 * internal-fn.c (expand_mul_overflow): If both operands have
11057 the same highpart of -1 or 0 and the topmost bit of lowpart
11058 is different, overflow is if res <= 0 rather than res < 0.
11060 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11062 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11063 TARGET_P9_VECTOR code for unaligned_load case.
11065 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11067 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11069 2017-10-13 Nathan Sidwell <nathan@acm.org>
11071 * tree-core.h (tree_contains_struct): Make bool.
11072 * tree.c (tree_contains_struct): Likewise.
11073 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11074 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11075 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11076 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11077 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11079 2017-10-13 Richard Biener <rguenther@suse.de>
11081 * graphite-isl-ast-to-gimple.c
11082 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11083 parameters and dominance check.
11084 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11085 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11086 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11087 Do not update SSA form here or do intermediate IL verification.
11088 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11089 (graphite_initialize): Remove check on the number of loops in
11090 the function and inline into graphite_transform_loops.
11091 (graphite_finalize): Inline into graphite_transform_loops.
11092 (graphite_transform_loops): Perform SSA update and IL verification
11094 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11096 2017-10-13 Richard Biener <rguenther@suse.de>
11098 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11099 graphite_expression_type_precision): Avoid global constructor
11101 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11102 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11103 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11104 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11105 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11106 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11108 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11111 * config/i386/i386.h (ix86_red_zone_size): New.
11112 * config/i386/i386.md (push peephole2s): Replace
11113 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11115 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11116 Alan Hayward <alan.hayward@arm.com>
11117 David Sherwood <david.sherwood@arm.com>
11119 * combine.c (can_change_dest_mode): Reject changes in
11120 REGMODE_NATURAL_SIZE.
11122 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11123 Alan Hayward <alan.hayward@arm.com>
11124 David Sherwood <david.sherwood@arm.com>
11126 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11127 (expand_debug_source_expr): Likewise.
11128 * combine.c (combine_simplify_rtx): Likewise.
11129 * cse.c (fold_rtx): Likewise.
11130 * fwprop.c (canonicalize_address): Likewise.
11131 * targhooks.c (default_shift_truncation_mask): Likewise.
11133 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11134 Alan Hayward <alan.hayward@arm.com>
11135 David Sherwood <david.sherwood@arm.com>
11137 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11138 (widened_mode): Likewise.
11139 (expand_unop): Likewise.
11140 * ree.c (transform_ifelse): Likewise.
11141 (merge_def_and_ext): Likewise.
11142 (combine_reaching_defs): Likewise.
11143 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11145 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11146 Alan Hayward <alan.hayward@arm.com>
11147 David Sherwood <david.sherwood@arm.com>
11149 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11150 * combine.c (gen_lowpart_for_combine): Likewise.
11151 * dwarf2out.c (rtl_for_decl_location): Likewise.
11152 * final.c (alter_subreg): Likewise.
11153 * rtlhooks.c (gen_lowpart_general): Likewise.
11154 (gen_lowpart_if_possible): Likewise.
11156 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11157 Alan Hayward <alan.hayward@arm.com>
11158 David Sherwood <david.sherwood@arm.com>
11160 * calls.c (expand_call): Use subreg_lowpart_offset.
11161 * cse.c (cse_insn): Likewise.
11162 * regcprop.c (copy_value): Likewise.
11163 (copyprop_hardreg_forward_1): Likewise.
11165 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11168 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11169 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11170 =Q constraints to +Q and into insn condition add check
11171 that operands[0] and operands[1] are equal.
11172 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11173 =Q constraints to +Q and into insn condition add check
11174 that operands[0] is equal to either operands[1] or operands[2].
11177 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11178 instead of handling MINUS_EXPR twice (once for each argument),
11179 canonicalize operand order and handle just once, use rtype where
11180 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11183 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11184 any values of __C while still being pattern recognizable as a simple
11185 rotate instruction.
11187 2017-10-13 Richard Biener <rguenther@suse.de>
11189 PR tree-optimization/82451
11191 2017-10-02 Richard Biener <rguenther@suse.de>
11193 PR tree-optimization/82355
11194 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11195 a mapping for the enclosing loop but avoid generating one for
11196 the loop tree root.
11197 (copy_bb_and_scalar_dependences): Remove premature codegen
11198 error on PHIs in blocks duplicated into multiple places.
11199 * graphite-scop-detection.c
11200 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11201 in the region use it as loop and nest to analyze the DR in.
11202 (try_generate_gimple_bb): Likewise.
11203 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11204 (add_loop_constraints): For blocks in a loop not in the region
11205 create a dimension with a single iteration.
11206 * sese.h (gbb_loop_at_index): Remove assert.
11208 * cfgloop.c (loop_preheader_edge): For the loop tree root
11209 return the single successor of the entry block.
11210 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11211 Reset the SCEV hashtable and niters.
11212 * graphite-scop-detection.c
11213 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11214 assert that we only have POLYNOMIAL_CHREC that vary in loops
11215 contained in the region.
11216 (scop_detection::graphite_can_represent_expr): Adjust.
11217 (scop_detection::stmt_has_simple_data_refs_p): For loops
11218 not in the region set loop to NULL. The nest is now the
11219 entry edge to the region.
11220 (try_generate_gimple_bb): Likewise.
11221 * sese.c (scalar_evolution_in_region): Adjust for
11222 instantiate_scev change.
11223 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11224 Make nest parameter the edge into the region.
11225 (create_data_ref): Likewise.
11226 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11227 entry edge into a region and adjust instantiate_scev calls.
11228 (create_data_ref): Likewise.
11229 (graphite_find_data_references_in_stmt): Likewise.
11230 (find_data_references_in_stmt): Pass the loop preheader edge
11231 from the nest argument.
11232 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11233 parameter the edge into the region.
11234 (instantiate_parameters): Use the loop preheader edge as entry.
11235 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11237 (get_instantiated_value_entry): Make instantiate_below parameter
11238 the edge into the region.
11239 (instantiate_scev_name): Likewise. Adjust dominance checks,
11240 when we cannot use loop-based instantiation instantiate by
11241 walking use-def chains.
11242 (instantiate_scev_poly): Adjust.
11243 (instantiate_scev_binary): Likewise.
11244 (instantiate_scev_convert): Likewise.
11245 (instantiate_scev_not): Likewise.
11246 (instantiate_array_ref): Remove.
11247 (instantiate_scev_3): Likewise.
11248 (instantiate_scev_2): Likewise.
11249 (instantiate_scev_1): Likewise.
11250 (instantiate_scev_r): Do not blindly handle N-operand trees.
11251 Do not instantiate array-refs. Handle all constants and invariants.
11252 (instantiate_scev): Make instantiate_below parameter
11253 the edge into the region.
11254 (resolve_mixers): Use the loop preheader edge for the region
11255 parameter to instantiate_scev_r.
11256 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11258 2017-10-13 Richard Biener <rguenther@suse.de>
11260 PR tree-optimization/82525
11261 * graphite-isl-ast-to-gimple.c
11262 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11264 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11265 Fail code generation when we cannot represent the isl integer.
11266 (binary_op_to_tree): Elide modulo operations that are no-ops
11267 in the type we code generate. Remove now superfluous code
11270 2017-10-13 Richard Biener <rguenther@suse.de>
11272 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11273 (scop_detection::harmful_loop_in_region): Remove premature
11274 IV type restriction.
11275 (scop_detection::graphite_can_represent_scev): We can handle
11276 pointer IVs just fine.
11278 2017-10-13 Alan Modra <amodra@gmail.com>
11280 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11281 "Clobbers and Scratch Registers". Add paragraph on
11282 alternative to clobbers for scratch registers and OpenBLAS
11285 2017-10-13 Alan Modra <amodra@gmail.com>
11287 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11288 example of a memory input for a string of known length. Move
11289 commentary out of table. Add a number of new examples
11290 covering array memory inputs.
11292 2017-10-12 Martin Liska <mliska@suse.cz>
11294 PR tree-optimization/82493
11295 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11296 (test_range_functions): New function.
11297 (sbitmap_c_tests): Likewise.
11298 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11299 * selftest.h (sbitmap_c_tests): New function.
11301 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11303 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11305 * config/rs6000/amo.h: Fix spacing issue.
11307 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11310 * config/i386/i386.md (*ashl<mode>3_mask_1,
11311 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11312 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11315 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11317 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11318 (profile_probability): Set max_probability
11319 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11321 * profile-count.c (profile_probability::differs_from_p): Do not
11322 rely on max_probaiblity == 10000
11324 2017-10-12 Jeff Law <law@redhat.com>
11326 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11329 2017-10-12 Martin Sebor <msebor@redhat.com>
11333 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11334 (handle_alias_pairs): Call it.
11335 * common.opt (-Wattribute-alias): New option.
11336 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11337 * doc/invoke.texi (-Wattribute-alias): Document.
11339 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11342 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11344 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11346 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11347 (make_hard_regno_born, make_hard_regno_dead): Update
11348 bb_killed_pseudos and bb_gen_pseudos.
11350 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11352 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11354 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11356 * config/alpha/alpha.c (alpha_split_conditional_move):
11357 Use std::swap instead of manually swapping.
11358 (alpha_stdarg_optimize_hook): Ditto.
11359 (alpha_canonicalize_comparison): Ditto.
11361 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11363 * tree-loop-distribution.c (struct builtin_info): New struct.
11364 (struct partition): Refactor fields into struct builtin_info.
11365 (partition_free): Free struct builtin_info.
11366 (build_size_arg_loc, build_addr_arg_loc): Delete.
11367 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11368 information from struct builtin_info.
11369 (find_single_drs): New function refactored from classify_partition.
11370 Also moved builtin validity checks to this function.
11371 (compute_access_range, alloc_builtin): New functions.
11372 (classify_builtin_st, classify_builtin_ldst): New functions.
11373 (classify_partition): Refactor code into functions find_single_drs,
11374 classify_builtin_st and classify_builtin_ldst.
11375 (distribute_loop): Don't do runtime alias check when distributing
11377 (find_seed_stmts_for_distribution): New function.
11378 (pass_loop_distribution::execute): Refactor code finding seed
11379 stmts into above function. Support distribution for the innermost
11380 two-level loop nest. Adjust dump information.
11382 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11384 * tree-loop-distribution.c: Adjust the general comment.
11385 (NUM_PARTITION_THRESHOLD): New macro.
11386 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11387 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11388 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11389 in call to build_partition_graph.
11390 (finalize_partitions): New parameter. Make loop distribution more
11391 conservative by fusing more partitions.
11392 (distribute_loop): Don't do runtime alias check in case of loop nest
11394 (find_seed_stmts_for_distribution): New function.
11395 (prepare_perfect_loop_nest): New function.
11396 (pass_loop_distribution::execute): Refactor code finding seed stmts
11397 and loop nest into above functions. Support loop nest distribution.
11398 Adjust dump information accordingly.
11400 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11402 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11403 and set PTYPE_SEQUENTIAL for merged partition.
11405 2017-10-12 Richard Biener <rguenther@suse.de>
11407 PR tree-optimization/69728
11409 2017-09-19 Richard Biener <rguenther@suse.de>
11411 PR tree-optimization/69728
11412 * graphite-sese-to-poly.c (schedule_error): New global.
11413 (add_loop_schedule): Handle empty domain by failing the
11415 (build_original_schedule): Handle schedule_error.
11417 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11418 domain by returning an unchanged schedule.
11420 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11422 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11423 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11425 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11427 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11430 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11433 * expr.c (store_field): Don't optimize away bitsize == 0 store
11434 from CALL_EXPR with addressable return type.
11436 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11438 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11439 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11440 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11441 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11442 TARGET_ISEL instead of TARGET_ISEL<sel>.
11444 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11446 * config/rs6000/rs6000.c
11447 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11449 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11451 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11453 (reg_or_cint_operand): Fix comment.
11454 (reg_or_zero_operand): New predicate.
11455 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11456 * config/rs6000/rs6000.c (output_isel): Delete.
11457 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11458 instead of reg_or_cint_operand. Output instruction directly (not via
11460 (isel_unsigned_<mode>): Ditto.
11461 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11462 gpc_reg_operand. Add an instruction alternative for this. Output
11463 instruction directly.
11464 (*isel_reversed_unsigned_<mode>): Ditto.
11466 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11468 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11469 (TARGET_CANONICALIZE_COMPARISON): Define.
11471 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11474 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11475 Check whether the dest is REG before adding REG_EQUIV note.
11477 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11480 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11482 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11483 (make_hard_regno_born, make_hard_regno_dead): Update
11484 bb_killed_pseudos and bb_gen_pseudos.
11486 2017-10-11 Nathan Sidwell <nathan@acm.org>
11488 * incpath.h (enum incpath_kind): Name enum, prefix values.
11489 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11490 * incpath.c (heads, tails): Use INC_MAX.
11491 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11492 (merge_include_chains, split_quote_chain,
11493 register_include_chains): Update incpath_kind names.
11494 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11495 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11497 (add_framework_path, darwin_register_objc_includes): Likewise.
11498 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11500 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11502 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11503 Do not use float_operator operator predicate.
11504 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11505 * config/i386/predicates.md (float_operator): Remove predicate.
11507 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11509 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11510 (*jccxf_i387): Ditto.
11511 (*jcc<mode>_i387): Ditto.
11512 (*jccu<mode>_i387): Ditto.
11513 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11514 (*jcc_*_i387 splitters): Remove.
11515 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11516 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11517 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11520 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11522 * profile-count.h (slow_safe_scale_64bit): New function.
11523 (safe_scale_64bit): New inline.
11524 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11525 * profile-count.c: Include wide-int.h
11526 (slow_safe_scale_64bit): New.
11528 2017-10-11 Nathan Sidwell <nathan@acm.org>
11530 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11531 HAS_DECL_ASSEMBLER_NAME_P.
11532 * gimple-expr.c (gimple_decl_printable_name: Check
11533 HAS_DECL_ASSEMBLER_NAME_P too.
11534 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11535 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11536 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11537 * passes.c (rest_of_decl_compilation): Check
11538 HAS_DECL_ASSEMBLER_NAME_P too.
11539 * recog.c (verify_changes): Likewise.
11540 * tree-pretty-print.c (dump_decl_name): Likewise.
11541 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11543 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11544 (SET_DECL_ASSEMBLER_NAME): Use it.
11545 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11546 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11547 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11549 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11551 * config.gcc (i386, x86_64): Add extra objects.
11552 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11553 (ix86_min_insn_size): Declare.
11554 (ix86_issue_rate): Declare.
11555 (ix86_adjust_cost): Declare.
11556 (ia32_multipass_dfa_lookahead): Declare.
11557 (ix86_macro_fusion_p): Declare.
11558 (ix86_macro_fusion_pair_p): Declare.
11559 (ix86_bd_has_dispatch): Declare.
11560 (ix86_bd_do_dispatch): Declare.
11561 (ix86_core2i7_init_hooks): Declare.
11562 (ix86_atom_sched_reorder): Declare.
11563 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11564 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11565 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11566 (rip_relative_addr_p): Rename to ...
11567 (ix86_rip_relative_addr_p): ... this one; export.
11568 (memory_address_length): Update.
11569 (ix86_issue_rate): Move to x86-tune-sched.c.
11570 (ix86_flags_dependent): Move to x86-tune-sched.c.
11571 (ix86_agi_dependent): Move to x86-tune-sched.c.
11572 (exact_dependency_1): Move to x86-tune-sched.c.
11573 (exact_store_load_dependency): Move to x86-tune-sched.c.
11574 (ix86_adjust_cost): Move to x86-tune-sched.c.
11575 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11576 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11577 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11578 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11579 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11580 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11581 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11582 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11583 (min_insn_size): Rename to ...
11584 (ix86_min_insn_size): ... this one; export.
11585 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11586 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11587 (core2i7_first_cycle_multipass_backtrack): Move to
11588 x86-tune-sched-core.c.
11589 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11590 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11591 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11592 (ix86_avoid_jump_mispredicts): Update.
11593 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11594 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11595 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11596 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11597 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11598 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11599 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11600 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11601 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11602 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11603 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11604 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11605 (BIG): Move to ix86-tune-sched-bd.c.
11606 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11607 (enum insn_path): Move to ix86-tune-sched-bd.c.
11608 (get_mem_group): Move to ix86-tune-sched-bd.c.
11609 (is_cmp): Move to ix86-tune-sched-bd.c.
11610 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11611 (is_branch): Move to ix86-tune-sched-bd.c.
11612 (is_prefetch): Move to ix86-tune-sched-bd.c.
11613 (init_window): Move to ix86-tune-sched-bd.c.
11614 (allocate_window): Move to ix86-tune-sched-bd.c.
11615 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11616 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11617 (process_end_window): Move to ix86-tune-sched-bd.c.
11618 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11619 (find_constant): Move to ix86-tune-sched-bd.c.
11620 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11621 (has_immediate): Move to ix86-tune-sched-bd.c.
11622 (get_insn_path): Move to ix86-tune-sched-bd.c.
11623 (get_insn_group): Move to ix86-tune-sched-bd.c.
11624 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11625 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11626 (add_insn_window): Move to ix86-tune-sched-bd.c.
11627 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11628 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11629 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11630 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11631 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11632 (do_dispatch): Move to ix86-tune-sched-bd.c.
11633 (has_dispatch): Move to ix86-tune-sched-bd.c.
11634 * config/i386/t-i386: Add new object files.
11635 * config/i386/x86-tune-costs.h: New file.
11636 * config/i386/x86-tune-sched-atom.c: New file.
11637 * config/i386/x86-tune-sched-bd.c: New file.
11638 * config/i386/x86-tune-sched-core.c: New file.
11639 * config/i386/x86-tune-sched.c: New file.
11641 2017-10-11 Liu Hao <lh_mouse@126.com>
11643 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11644 the generic version below instead.
11645 (should_colorize): Recognize Windows consoles as terminals
11647 * pretty-print.c [__MINGW32__] (write_all): New function.
11648 [__MINGW32__] (find_esc_head): Likewise.
11649 [__MINGW32__] (find_esc_terminator): Likewise.
11650 [__MINGW32__] (eat_esc_sequence): Likewise.
11651 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11653 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11656 2017-10-11 Richard Biener <rguenther@suse.de>
11658 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11659 Properly call analyze_scalar_evolution with the loop of the stmt.
11661 2017-10-11 Richard Biener <rguenther@suse.de>
11663 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11664 * tree-core.h (tree_base): Add chrec_var union member.
11665 * tree.h (CHREC_VAR): Remove.
11666 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11667 * tree-chrec.h (build_polynomial_chrec): Adjust.
11668 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11669 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11671 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11673 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11674 * match.pd: ... here.
11675 ((T) X == (T) Y): Relax condition.
11677 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11679 PR tree-optimization/82472
11680 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11682 (break_alias_scc_partitions): Update postorder number.
11684 2017-10-11 Martin Liska <mliska@suse.cz>
11687 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11689 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11691 2017-10-11 Martin Liska <mliska@suse.cz>
11693 * print-rtl.c (print_insn): Move declaration of idbuf
11694 to same scope as name.
11696 2017-10-11 Martin Liska <mliska@suse.cz>
11701 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11702 volatile arguments.
11704 2017-10-11 Martin Liska <mliska@suse.cz>
11707 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11708 volatile arguments.
11710 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11712 * config.gcc (default_gnu_indirect_function): Default to yes for
11713 arm*-*-linux* with glibc.
11715 2017-10-11 Richard Biener <rguenther@suse.de>
11717 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11718 default-defs and types we do not want to analyze.
11719 (interpret_loop_phi): Replace unreachable code with an assert.
11720 (compute_scalar_evolution_in_loop): Remove and inline ...
11721 (analyze_scalar_evolution_1): ... here, replacing condition with
11722 what makes the intent clearer. Remove handling of cases
11723 get_scalar_evolution now handles.
11725 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11727 PR rtl-optimization/81434
11728 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11729 comment for main loop. In sched_group_found if, also add checks for
11730 pass and min_cost_group.
11732 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11734 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11735 (rs6000_insn_cost): New function.
11736 * config/rs6000/rs6000.md (cost): New attribute.
11738 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11739 H.J. Lu <hongjiu.lu@intel.com>
11743 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11744 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11745 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11746 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11747 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11748 OPTION_MASK_ISA_MMX and some other bit set require both
11749 mmx and the ISAs without the mmx bit.
11750 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11751 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11752 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11753 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11754 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11755 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11756 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11757 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11758 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11759 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11760 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11761 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11762 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11763 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11764 Add OPTION_MASK_ISA_MMX.
11766 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11768 * config.gcc (armv7*-*-freebsd*): New target.
11769 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11771 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11773 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11774 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11777 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11779 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11780 a template parameter.
11781 (WIDE_INT_REF_FOR): Update accordingly.
11782 * tree.h (wi::int_traits <const_tree>): Delete.
11783 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11784 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11785 (wi::tree_to_wide_ref): New typedef.
11786 (wi::to_wide): New function.
11787 * calls.c (get_size_range): Use wi::to_wide when operating on
11788 trees as wide_ints.
11789 * cgraph.c (cgraph_node::create_thunk): Likewise.
11790 * config/i386/i386.c (ix86_data_alignment): Likewise.
11791 (ix86_local_alignment): Likewise.
11792 * dbxout.c (stabstr_O): Likewise.
11793 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11794 * expr.c (const_vector_from_tree): Likewise.
11795 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11796 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11797 (fold_negate_expr_1, int_const_binop_1, const_binop)
11798 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11799 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11800 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11801 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11802 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11803 (fold_not_const, round_up_loc): Likewise.
11804 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11805 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11806 (alloca_call_type): Likewise.
11807 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11808 * godump.c (go_output_typedef): Likewise.
11809 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11810 * internal-fn.c (get_min_precision): Likewise.
11811 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11812 * ipa-polymorphic-call.c
11813 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11814 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11815 (ipa_modify_call_arguments): Likewise.
11816 * match.pd: Likewise.
11817 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11818 * print-tree.c (print_node_brief, print_node): Likewise.
11819 * stmt.c (expand_case): Likewise.
11820 * stor-layout.c (layout_type): Likewise.
11821 * tree-affine.c (tree_to_aff_combination): Likewise.
11822 * tree-cfg.c (group_case_labels_stmt): Likewise.
11823 * tree-data-ref.c (dr_analyze_indices): Likewise.
11824 (prune_runtime_alias_test_list): Likewise.
11825 * tree-dump.c (dequeue_and_dump): Likewise.
11826 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11827 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11828 * tree-pretty-print.c (dump_generic_node): Likewise.
11829 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11830 (simple_iv_with_niters): Likewise.
11831 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11832 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11833 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11834 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11835 (refine_value_range_using_guard, number_of_iterations_ne_max)
11836 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11837 (get_cst_init_from_scev, record_nonwrapping_iv)
11838 (scev_var_range_cant_overflow): Likewise.
11839 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11840 * tree-ssa-pre.c (compute_avail): Likewise.
11841 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11842 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11843 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11844 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11845 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11846 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11847 (dump_case_nodes, try_switch_expansion): Likewise.
11848 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11849 (vect_do_peeling): Likewise.
11850 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11851 * tree-vect-stmts.c (vectorizable_load): Likewise.
11852 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11853 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11854 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11855 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11856 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11857 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11858 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
11859 (evrp_dom_walker::before_dom_children): Likewise.
11860 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
11861 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
11862 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
11863 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
11864 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
11865 (get_range_pos_neg): Likewise.
11866 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
11867 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
11868 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
11869 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
11870 * config/avr/avr.c (avr_fold_builtin): Likewise.
11871 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
11872 * config/msp430/msp430.c (msp430_attr): Likewise.
11873 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
11874 * config/powerpcspe/powerpcspe-c.c
11875 (altivec_resolve_overloaded_builtin): Likewise.
11876 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
11877 (rs6000_expand_ternop_builtin): Likewise.
11878 * config/rs6000/rs6000-c.c
11879 (altivec_resolve_overloaded_builtin): Likewise.
11880 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
11881 (rs6000_expand_ternop_builtin): Likewise.
11882 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
11884 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11886 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
11887 when copying loop nest with only one inner loop.
11889 2017-10-10 Richard Biener <rguenther@suse.de>
11891 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
11892 blocks if SCEV is active.
11893 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
11895 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
11896 (scev_initialize): Assert we are not yet initialized.
11898 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11900 * tree-loop-distribution.c (generate_loops_for_partition): Remove
11901 inner loop's exit stmt by making it always exit the loop, otherwise
11902 we would generate an infinite empty loop.
11904 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11906 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
11907 renaming variables in new preheader if it's deleted.
11909 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11911 * tree-loop-distribution.c (struct partition): Remove unused field
11912 loops of the structure.
11913 (partition_alloc, partition_free): Ditto.
11914 (build_rdg_partition_for_vertex): Ditto.
11916 2017-10-09 Jeff Law <law@redhat.com>
11918 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
11919 return type to match prototype and documentation.
11921 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11923 * config/rs6000/rs6000.c (processor_costs): Move to ...
11924 * config/rs6000/rs6000.h: ... here.
11925 (rs6000_cost): Declare.
11927 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
11929 * except.c (setjmp_fn): New global variable.
11930 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
11931 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
11932 if DONT_USE_BUILTIN_SETJMP is defined.
11934 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11936 * target.def (insn_cost): New hook.
11937 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
11938 * doc/tm.texi: Regenerate.
11939 * rtlanal.c (insn_cost): Use the new hook.
11941 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11943 * combine.c (combine_validate_cost): Compute the new insn_cost,
11944 not just pattern_cost.
11945 (try_combine): Adjust comment.
11947 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11949 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
11951 * combine.c (uid_insn_cost): Adjust comment.
11952 (combine_validate_cost): Adjust comment. Use pattern_cost instead
11954 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
11955 * dse.c (find_shift_sequence): Ditto.
11956 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
11957 (bb_valid_for_noce_process_p): Use pattern_cost.
11958 * rtl.h (insn_rtx_cost): Delete.
11959 (pattern_cost): New prototype.
11960 (insn_cost): New prototype.
11961 * rtlanal.c (insn_rtx_cost): Rename to...
11962 (pattern_cost): ... this.
11965 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
11967 * config/i386/i386.md (*jcc_2): Remove insn pattern.
11968 (*jcc<mode>_0_r_i387): Ditto.
11969 (*jccxf_r_i387): Ditto.
11970 (*jcc<mode>_r_i387): Ditto.
11971 (*jccu<mode>_r_i387): Ditto.
11972 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
11973 (*jcc): Rename from *jcc_1.
11975 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11977 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
11978 deferred rescans after the lvx/stvx recombination pre-pass.
11980 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
11982 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
11983 memory operation instruction support.
11984 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
11985 (rs6000-ibm-aix[789]*): Likewise.
11986 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
11987 Document new functions.
11989 2017-10-09 Richard Biener <rguenther@suse.de>
11991 PR tree-optimization/82397
11992 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
11993 equality only for semantically equal trees.
11995 2017-10-09 Richard Biener <rguenther@suse.de>
11997 PR tree-optimization/82449
11998 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
11999 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12000 allow constant addresses.
12001 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12004 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12006 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12009 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12012 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12015 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12018 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12020 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12023 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12024 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12026 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12028 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12029 (WI_BINARY_PREDICATE_RESULT): Likewise.
12030 (wi::binary_traits::operator_result): New type.
12031 (wi::binary_traits::predicate_result): Likewise.
12032 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12033 (generic_wide_int::operator==, generic_wide_int::operator!=)
12034 (generic_wide_int::operator&, generic_wide_int::and_not)
12035 (generic_wide_int::operator|, generic_wide_int::or_not)
12036 (generic_wide_int::operator^, generic_wide_int::operator+
12037 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12039 (operator~, unary operator-, operator==, operator!=, operator&)
12040 (operator|, operator^, operator+, binary operator-, operator*): New
12042 * expr.c (get_inner_reference): Use wi::bit_and_not.
12043 * fold-const.c (fold_binary_loc): Likewise.
12044 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12045 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12046 (bit_value_binop): Likewise.
12047 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12048 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12049 (extract_range_from_binary_expr_1): Likewise.
12050 (masked_increment): Likewise.
12051 (simplify_bit_ops_using_ranges): Likewise.
12053 2017-10-09 Martin Jambor <mjambor@suse.cz>
12056 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12057 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12058 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12060 (hsa_fixup_mov_insn_type): New function.
12061 (hsa_op_with_type::get_in_type): Use it.
12062 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12063 immediates in an assert.
12064 (hsa_op_with_type::extend_int_to_32bit): New method.
12065 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12066 types. Convert to dest type if necessary.
12067 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12068 (reg_for_gimple_ssa): Pass false as min32int to
12069 hsa_type_for_scalar_tree_type.
12070 (gen_hsa_addr): Fixup type when creating addresable temporary.
12071 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12072 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12073 necessary. Call hsa_fixup_mov_insn_type.
12074 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12075 extend operands and convert to dest type if necessary.
12076 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12077 to dest type if necessary.
12078 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12079 if conversion nt necessary and size matches.
12080 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12081 to dest type if necessary.
12082 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12083 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12085 (gen_hsa_clrsb): Likewise.
12086 (gen_hsa_ffs): Likewise.
12087 (gen_hsa_divmod): Extend operands and convert to dest type if
12089 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12091 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12093 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12094 Remove empty default arguments. Use a brace block as output
12096 (conditional return): Ditto.
12098 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12099 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12101 (group_ending_nop): Ditto.
12102 (doloop_end): Ditto.
12103 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12104 (splitters for those): Ditto.
12106 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12108 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12109 a conditional jump (and the compare for it) so that pc_rtx is the
12111 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12112 for the deleted and renamed ctr<mode>_internal[234] patterns.
12113 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12114 Delete second conditional return pattern.
12115 (ctr<mode>_internal2): Delete this second bdnz pattern.
12116 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12117 (ctr<mode>_internal4): Delete this second bdz pattern.
12119 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12121 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12122 (always_initialized_rtx_for_ssa_name_p): New predicate.
12123 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12124 (finish_out_of_ssa): Free new field of SA.
12125 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12126 * tree-ssa-coalesce.c: Include tree-ssa.h.
12127 (get_parm_default_def_partitions): Remove extern keyword.
12128 (get_undefined_value_partitions): New function.
12129 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12130 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12131 uninitialized bits.
12132 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12134 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12136 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12138 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12140 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12141 for targets that preffer 128bit.
12143 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12145 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12147 2017-10-08 Olivier Hainque <hainque@adacore.com>
12149 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12150 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12152 (thumb_set_return_address): Likewise.
12154 2017-10-08 Olivier Hainque <hainque@adacore.com>
12156 * common/config/arm/arm-common.c (arm_except_unwind_info):
12157 Handle DWARF2_UNWIND_INFO.
12159 2017-10-07 Michael Collison <michael.collison@arm.com>
12161 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12164 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12166 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12167 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12168 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12169 defined, force the creation of a new block for a dispatch label.
12171 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12173 * invoke.texi (Wsuggest-attribute=cold): Document.
12174 * common.opt (Wsuggest-attribute=cold): New
12175 * ipa-pure-const.c (warn_function_cold): New function.
12176 * predict.c (compute_function_frequency): Use it.
12177 * predict.h (warn_function_cold): Declare.
12179 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12181 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12184 2017-10-06 Martin Liska <mliska@suse.cz>
12186 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12187 keyword for member functions.
12188 (struct sanopt_tree_couple): New struct.
12189 (struct sanopt_tree_couple_hash): New function.
12190 (struct sanopt_ctx): Add new hash_map.
12191 (has_dominating_ubsan_ptr_check): New function.
12192 (record_ubsan_ptr_check_stmt): Likewise.
12193 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12194 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12195 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12197 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12200 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12201 aarch64_simd_valid_immediate on CONST_VECTORs.
12202 (aarch64_reg_or_bic_imm): Likewise.
12204 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12206 PR rtl-optimization/82396
12207 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12209 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12211 * graphite-dependences.c (scop_get_reads): Move code to...
12212 (scop_get_must_writes): Move code to...
12213 (scop_get_may_writes): Move code to...
12214 (scop_get_reads_and_writes): ... here.
12215 (scop_get_dependences): Call scop_get_reads_and_writes.
12217 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12219 PR tree-optimization/82434
12220 * fold-const.h (can_native_encode_type_p,
12221 can_native_encode_string_p): Remove.
12222 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12223 don't encode anything, just return what would be otherwise returned.
12224 (native_encode_fixed, native_encode_complex, native_encode_vector):
12226 (native_encode_string): Likewise. Inline by hand
12227 can_native_encode_string_p.
12228 (can_native_encode_type_p): Remove.
12229 (can_native_encode_string_p): Remove.
12230 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12231 STRING_CSTs using can_native_encode_string_p, test all
12232 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12233 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12234 argument from native_encode_expr.
12235 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12236 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12239 2017-10-06 Richard Biener <rguenther@suse.de>
12241 PR tree-optimization/82397
12242 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12243 operand_equal_p but rely on data_ref_compare_tree for detecting
12245 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12246 to match up with dr_group_sort_cmp.
12248 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12251 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12253 * config/s390/s390-builtin-types.def: Regenerate.
12255 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12258 * config/s390/s390-builtin-types.def: Regenerate.
12259 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12260 Change flag from B_VXE to B_VX.
12261 (s390_vec_min_dbl): Remove B_VXE flag.
12263 2017-10-06 Richard Biener <rguenther@suse.de>
12265 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12266 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12267 translate_isl_ast_to_gimple::is_valid_rename,
12268 translate_isl_ast_to_gimple::get_rename,
12269 translate_isl_ast_to_gimple::get_def_bb_for_const,
12270 translate_isl_ast_to_gimple::get_new_name,
12271 translate_isl_ast_to_gimple::collect_all_ssa_names,
12272 translate_isl_ast_to_gimple::copy_loop_phi_args,
12273 translate_isl_ast_to_gimple::collect_all_ssa_names,
12274 translate_isl_ast_to_gimple::copy_loop_phi_args,
12275 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12276 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12277 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12278 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12279 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12280 translate_isl_ast_to_gimple::copy_cond_phi_args,
12281 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12282 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12283 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12284 translate_isl_ast_to_gimple::rename_uses,
12285 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12286 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12287 (set_rename_for_each_def): Likewise.
12288 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12289 here. Handle rewriting SCEV analyzable uses here.
12290 (copy_bb_and_scalar_dependences): Generate code for PHI
12292 (graphite_regenerate_ast_isl): Adjust.
12293 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12294 (add_write, add_read): New functions.
12295 (build_cross_bb_scalars_def): Use it and simplify.
12296 (build_cross_bb_scalars_use): Likewise.
12297 (graphite_find_cross_bb_scalar_vars): Inline into...
12298 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12299 simulating out-of-SSA. Compute liveout and add dependencies.
12300 (build_scops): Force an empty entry block.
12301 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12303 (sese_build_liveouts): Declare.
12304 (sese_trivially_empty_bb_p): Likewise.
12305 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12306 compute liveout and debug_liveout.
12307 (sese_bad_liveouts_use): Remove.
12308 (sese_reset_debug_liveouts_bb): Likewise.
12309 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12310 (sese_build_liveouts): Build liveout and debug_liveout and store
12312 (new_sese_info): Adjust.
12313 (free_sese_info): Likewise.
12314 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12315 do not build liveout here.
12316 (move_sese_in_condition): Adjust region entry.
12317 (scev_analyzable_p): Match up with chrec_apply requirements.
12318 (sese_trivially_empty_bb_p): New.
12319 * tree-into-ssa.c (get_reaching_def): Properly support generating
12320 default-defs for incremental rewrite of anonymous names.
12322 2017-10-06 Richard Biener <rguenther@suse.de>
12324 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12325 precision do not perform modulo reduction.
12327 2017-10-06 Richard Biener <rguenther@suse.de>
12329 PR tree-optimization/82436
12330 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12331 conservatively choose the vectorization factor when checking
12332 whether we can perform the required load permutation.
12333 (vect_transform_slp_perm_load): Assert when we may not fail.
12335 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12337 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12338 message for incompatible -msdata=* and -mcall-* options.
12340 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12342 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12343 rate for post-reload scheduling.
12345 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12347 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12349 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12351 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12352 to improve monte carlo in scimark.
12354 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12356 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12357 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12358 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12359 pentium4_cost, nocona_cost): Set reassociation width to 1.
12360 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12361 width to 2 for fp operations and 1 otherwise.
12362 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12364 (atom_cost): Set reassociation width to 2.
12365 (slm_cost, generic_cost): Set fp reassociation width
12366 to 2 and 1 otherwise.
12367 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12368 (core_cost): Set fp reassociation width to 4 and vector to 2.
12369 (ix86_reassociation_width): Rewrite using cost table; special case
12370 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12371 and TARGET_AVX128_OPTIMAL.
12372 * config/i386/i386.h (processor_costs): Add
12373 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12374 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12375 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12376 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12377 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12378 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12380 2017-10-05 Nathan Sidwell <nathan@acm.org>
12382 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12384 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12386 * config/arm/arm.c (arm_test_fpu_data): New.
12387 (arm_run_selftests): Call arm_test_fpu_data.
12389 2017-10-04 Nathan Sidwell <nathan@acm.org>
12391 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12393 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12395 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12397 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12398 check type for aarch64_simd_valid_immediate.
12399 (aarch64_output_simd_mov_immediate): Update prototype.
12400 (aarch64_simd_valid_immediate): Update prototype.
12401 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12402 support for ORR-immediate.
12403 (and<mode>3): modified pattern to add support for BIC-immediate.
12404 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12405 now checks for valid immediate for BIC and ORR based on new enum
12407 (aarch64_output_simd_mov_immediate): Function now used to output
12408 BIC/ORR imm as well based on new enum argument.
12409 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12411 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12412 (aarch64_reg_or_bic_imm): Likewise.
12414 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12416 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12417 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12419 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12423 PR rtl-optimization/82396
12424 * haifa-sched.c (autopref_multipass_init): Simplify
12426 (autopref_rank_data): Simplify sort order.
12427 * sched-int.h (autopref_multipass_data_): Remove
12428 multi_mem_insn_p, min_offset and max_offset.
12430 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12432 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12434 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12436 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12437 vect_doubleint_cvt.
12439 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12441 * doc/sourcebuild.texi: Document vect_long_mult.
12443 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12445 PR tree-optimization/82413
12446 * fold-const.c (build_range_check): Use widest_int when comparing
12447 the maximum ETYPE value with HIGH.
12449 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12451 PR rtl-optimization/82396
12452 * haifa-sched.c (autopref_multipass_init): Simplify
12454 (autopref_rank_data): Simplify sort order.
12455 * sched-int.h (autopref_multipass_data_): Remove
12456 multi_mem_insn_p, min_offset and max_offset.
12458 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12460 PR tree-optimization/82381
12461 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12462 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12465 PR tree-optimization/82374
12466 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12467 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12468 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12469 current_function_decl to the new decl.
12471 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12473 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12474 helper macro for IEEE float128 hardware built-in functions.
12475 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12477 (TRUNCF128_ODD): Likewise.
12478 (ADDF128_ODD): Likewise.
12479 (SUBF128_ODD): Likewise.
12480 (MULF128_ODD): Likewise.
12481 (DIVF128_ODD): Likewise.
12482 (FMAF128_ODD): Likewise.
12483 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12484 UNSPEC_TRUNC_ROUND_TO_ODD.
12485 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12486 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12487 floating point round to odd instructions.
12488 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12489 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12490 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12491 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12492 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12493 (trunc<mode>sf2_hw): Change the truncate with round to odd
12494 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12495 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12496 to odd hardware instructions.
12497 (sub<mode>3_odd): Likewise.
12498 (mul<mode>3_odd): Likewise.
12499 (div<mode>3_odd): Likewise.
12500 (sqrt<mode>2_odd): Likewise.
12501 (fma<mode>4_odd): Likewise.
12502 (fms<mode>4_odd): Likewise.
12503 (nfma<mode>4_odd): Likewise.
12504 (nfms<mode>4_odd): Likewise.
12505 (trunc<mode>df2_odd): Change the truncate with round to odd
12506 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12508 * doc/extend.texi (PowerPC built-in functions): Update documentation
12509 for existing IEEE float128-bit built-in functions. Add built-in
12510 functions that generate the IEEE 128-bit floating point round to
12513 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12515 PR rtl-optimization/77729
12516 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12517 to (X&(C1&~C2))|C2 transformations.
12519 2017-10-03 Martin Jambor <mjambor@suse.cz>
12521 PR tree-optimization/82363
12522 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12523 mismatch, mark lacc written regardless of racc.
12525 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12527 PR tree-optimization/82381
12528 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12529 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12530 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12531 fallthrough into reassoc_stmt_dominates_stmt_p.
12534 * combine.c (combine_instructions): Don't combine in unreachable
12537 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12540 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12541 function to not use the have_cpu variable. Do not set cpu_index,
12542 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12543 or the default cpu.
12544 (rs6000_valid_attribute_p): Remove duplicate initializations of
12545 old_optimize and func_optimize.
12546 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12547 (rs6000_activate_target_options): Make global.
12548 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12551 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12553 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12554 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12555 Set *poffset to 0 and *psize and *pmax_size to -1 if
12556 *poffset + *psize overflows in HOST_WIDE_INT.
12558 PR tree-optimization/82387
12559 PR tree-optimization/82388
12560 PR tree-optimization/82389
12561 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12562 instead of live_bytes non-NULL.
12564 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12567 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12570 2017-10-02 Richard Biener <rguenther@suse.de>
12572 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12573 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12574 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12576 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12578 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12579 requested precision matches the type's.
12580 * calls.c (alloc_max_size): Calculate the new candidate size as
12581 a widest_int and use wi::to_widest when comparing it with the
12582 current candidate size.
12583 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12584 zero rather than integer_zero_node.
12585 * match.pd: Check for a no-op conversion before using wi::add
12586 rather than after. Use tree_to_uhwi when summing small shift
12587 counts into an unsigned int.
12589 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12590 Alan Hayward <alan.hayward@arm.com>
12591 David Sherwood <david.sherwood@arm.com>
12594 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12595 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12596 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12597 POINTER_AND_FP_REGS.
12599 2017-10-02 Richard Biener <rguenther@suse.de>
12601 PR tree-optimization/82355
12602 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12603 a mapping for the enclosing loop but avoid generating one for
12604 the loop tree root.
12605 (copy_bb_and_scalar_dependences): Remove premature codegen
12606 error on PHIs in blocks duplicated into multiple places.
12607 * graphite-scop-detection.c
12608 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12609 in the region use it as loop and nest to analyze the DR in.
12610 (try_generate_gimple_bb): Likewise.
12611 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12612 (add_loop_constraints): For blocks in a loop not in the region
12613 create a dimension with a single iteration.
12614 * sese.h (gbb_loop_at_index): Remove assert.
12616 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12618 * omp-expand.c (adjust_context_scope): New function.
12619 (expand_parallel_call): Call adjust_context_scope.
12621 2017-10-01 Jeff Law <law@redhat.com>
12623 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12624 dom_opt_dom_walker class with direct access to private members.
12625 Add comments. Call test_for_singularity.
12626 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12627 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12628 m_dummy_cond anymore.
12629 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12631 (pass_dominator:execute): Build the dummy_cond here and pass it
12632 to the dom_opt_dom_walker ctor.
12633 (test_for_singularity): New function.
12635 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12636 Maya Rashish <coypu@sdf.org>
12638 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12639 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12640 (alpha*-*-netbsd*) Use nbsd_tm_file.
12641 (arm*-*-netbsdelf*) Likewise.
12642 (i[34567]86-*-netbsdelf*) Likewise.
12643 (x86_64-*-netbsd*) Likewise.
12644 (mips*-*-netbsd*) Likewise.
12645 (powerpc-*-netbsd*) Likewise.
12646 (sh*-*-netbsd*) Likewise.
12647 (sparc-*-netbsdelf*) Likewise.
12648 (sparc64-*-netbsd*) Likewise.
12649 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12651 (vax-*-netbsdelf*) Likewise.
12652 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12653 (UINT_FAST8_TYPE) Likewise.
12654 (INT_FAST16_TYPE) Check CHAR_FAST16.
12655 (UINT_FAST16_TYPE) Likewise.
12657 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12660 * config/i386/i386.md
12661 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12662 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12663 *divmodsi4_zext_2): New define_insn_and_split.
12664 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12665 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12666 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12667 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12668 New define_insn_and_split.
12669 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12670 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12671 operands[1] having DImode when mode is SImode.
12673 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12674 always SImode for DIV and MOD in REG_EQUAL notes.
12676 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12678 PR middle-end/82319
12679 * match.pd: Fix handling of NaNs in pattern.
12681 2017-09-29 Jeff Law <law@redhat.com>
12683 * sbitmap.c (bitmap_bit_in_range_p): New function.
12684 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12685 * tree-ssa-dse.c (live_bytes_read): New function.
12686 (dse_classify_store): Ignore reads of dead bytes.
12688 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12689 typos and whitespace errors.
12690 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12691 * config/s390/s390.c (s390_emit_prologue): Likewise.
12693 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12696 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12697 with a symbol for LRA.
12699 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12701 PR rtl-optimization/82338
12702 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12704 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12706 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12707 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12708 (qsort_chk): Declare.
12709 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12710 (qsort_chk): New function.
12712 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12714 PR tree-optimization/82337
12715 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12716 phi definition if the PHI result appears in an abnormal PHI.
12717 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12718 basis appears in an abnormal PHI.
12720 2017-09-29 Richard Biener <rguenther@suse.de>
12722 * graphite-isl-ast-to-gimple.c
12723 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12724 (binary_op_to_tree): Use it.
12725 (get_rename_from_scev): Likewise.
12726 (copy_loop_phi_nodes): Likewise.
12727 (copy_bb_and_scalar_dependences): Likewise.
12728 (translate_pending_phi_nodes): Likewise.
12730 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12733 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12734 for movabsq $(i32 << shift), r64.
12736 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12738 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12739 index when encoding %esp as %rsp to avoid 0x67 prefix.
12741 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12743 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12744 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12746 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12748 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12749 Extensions with more than 16 double VFP registers.
12750 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12751 to_clear_mask and all code related to it. Replace the remaining
12752 entry by a sbitmap and adapt code accordingly.
12754 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12756 * brig-builtins.def: Change pure attributes to const.
12758 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12760 * config.gcc (default_gnu_indirect_function): Default to yes for
12761 sparc*-*-linux* with glibc.
12763 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12765 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12766 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12767 when creating .init_array and .fini_array sections with priority
12770 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12773 * config/aarch64/aarch64.c
12774 (aarch64_builtin_support_vector_misalignment): Always return false
12775 when misalignment is unknown.
12777 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12779 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12780 this function to return false if the definition used by the swap
12781 instruction is artificial, or if the memory address from which the
12782 constant value is loaded is not represented by a base address held
12783 in a register or if the base address register is a frame or stack
12784 pointer. Additionally, return false if the base address of the
12785 loaded constant is a SYMBOL_REF but is not considered to be a
12787 (replace_swapped_load_constant): New function.
12788 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12789 loaded constant vector with a load of a swapped constant vector.
12791 2017-09-27 Carl Love <cel@us.ibm.com>
12793 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12794 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12795 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12798 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12800 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12801 first, always call autopref_rank_data otherwise.
12803 2017-09-27 Richard Biener <rguenther@suse.de>
12805 * graphite-scop-detection.c (find_scop_parameters): Move
12806 loop bound handling ...
12807 (gather_bbs::before_dom_children): ... here, avoiding the need
12808 to build scop_info->loop_nest.
12809 (record_loop_in_sese): Remove.
12810 * sese.h (sese_info_t::loop_nest): Remove.
12811 * sese.c (new_sese_info): Do not allocate loop_nest.
12812 (free_sese_info): Do not free loop_nest.
12814 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12817 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12818 lhs from calls if the lhs has addressable type.
12820 2017-09-27 Richard Biener <rguenther@suse.de>
12822 * graphite.h (scop::max_alias_set): New member.
12823 * graphite-scop-detection.c: Remove references to non-existing
12824 --param in comments.
12825 (build_alias_sets): Record the maximum alias set used for drs.
12826 (build_scops): Support zero as unlimited for
12827 --param graphite-max-arrays-per-scop.
12828 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12829 and inline into ...
12830 (build_poly_sr_1): ... here. Compute alias set based on the
12831 maximum alias set used for drs rather than
12832 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12834 2017-09-27 Richard Biener <rguenther@suse.de>
12836 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12837 --param loop-block-tile-size=0 to disable tiling.
12839 2017-09-27 Richard Biener <rguenther@suse.de>
12841 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12842 (graphite-max-nb-scop-params): Document special value zero.
12843 * domwalk.h (dom_walker::STOP): New symbolical constant.
12844 (dom_walker::dom_walker): Add optional parameter for bb to
12846 (dom_walker::~dom_walker): Declare.
12847 (dom_walker::before_dom_children): Document STOP return value.
12848 (dom_walker::m_user_bb_to_rpo): New member.
12849 (dom_walker::m_bb_to_rpo): Likewise.
12850 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12851 mapping here if not provided by the user.
12852 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12853 provided by the user.
12854 (dom_walker::STOP): Define.
12855 (dom_walker::walk): Do not compute bb to RPO mapping here.
12856 Support STOP return value from before_dom_children to stop
12858 * graphite-optimize-isl.c (optimize_isl): If the schedule
12859 is the same still generate code if -fgraphite-identity
12860 or -floop-parallelize-all are given.
12861 * graphite-scop-detection.c: Include cfganal.h.
12862 (gather_bbs::gather_bbs): Get and pass through bb to RPO
12864 (gather_bbs::before_dom_children): Return STOP for BBs
12866 (build_scops): Compute bb to RPO mapping and pass it to
12867 the domwalk. Treat --param graphite-max-nb-scop-params=0
12868 as not limiting the number of params.
12869 * graphite.c (graphite_initialize): Remove limit on the
12870 number of basic-blocks in a function.
12871 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
12872 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
12873 default value of 10.
12875 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12877 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
12878 Adjust code to eliminate needing to do the shift right 32-bits
12879 operation after XSCVDPSPN.
12881 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12883 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
12884 ((X / Y) != 0 -> X >= Y): Likewise.
12886 2017-09-26 Carl Love <cel@us.ibm.com>
12888 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
12889 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
12890 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
12891 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
12892 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
12893 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
12894 definitions and overloading.
12895 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
12896 statement for P9V_BUILTIN_XST_LEN_R.
12897 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
12898 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
12899 define_expand and define_insn for the instructions and builtins.
12900 * doc/extend.texi: Update the built-in documentation file for the new
12901 built-in functions.
12902 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
12903 define_insn for the instructions
12905 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
12908 * gcc/config/netbsd-protos.h: New file.
12909 * gcc/config/netbsd.c: New file.
12910 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
12911 * gcc/config/t-netbsd: New file.
12912 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
12913 (tmake_file) Add t-netbsd.
12914 (extra_objs) Add netbsd.o.
12916 2017-09-26 Janus Weil <janus@gcc.gnu.org>
12920 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
12922 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12924 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
12925 sign extension from a vector register to a GPR by doing a 32-bit
12926 direct move and then an EXTSW.
12927 (extendsi<mode>2 splitter): Likewise.
12928 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
12929 right or vector extract after doing XSCVDPSPN. Use
12930 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
12932 (movdi_from_sf_zero_ext): Likewise.
12933 (reload_gpr_from_vsxsf): Likewise.
12934 (p8_mfvsrd_4_disf): Delete, no longer used.
12935 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
12936 then needing to move the SFmode to a GPR to use the XSCVDPSP
12937 instruction instead of FRSP and XSCVDPSPN.
12938 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
12939 it is adjacent to the other XSCVSPDP insns.
12940 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
12941 SFmode to be in traditional Altivec registers.
12942 (vsx_xscvdpspn): Eliminate useless alternative constraint.
12943 (vsx_xscvspdpn): Likewise.
12944 (vsx_xscvspdpn_scalar): Likewise.
12946 2017-09-26 Martin Jambor <mjambor@suse.cz>
12948 * tree-sra.c (compare_access_positions): Put integral types first,
12949 stabilize sorting of integral types, remove conditions putting
12950 non-full-precision integers last.
12951 (sort_and_splice_var_accesses): Disable scalarization if a
12952 non-integert would be represented by a non-full-precision integer.
12954 2017-09-26 Joseph Myers <joseph@codesourcery.com>
12956 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
12957 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
12958 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
12959 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
12960 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
12961 conditionals inside the function instead of around it. Call
12962 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
12963 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
12965 2017-09-26 Richard Biener <rguenther@suse.de>
12967 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
12969 (scop_detection::build_scop_breadth): ... this. Removed.
12970 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
12971 (scop_detection::harmful_stmt_in_bb): Likewise.
12972 (scop_detection::graphite_can_represent_stmt): Likewise.
12973 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
12974 (scop_detection::can_represent_loop): Remove recursion, fold in ...
12975 (scop_detection::can_represent_loop_1): ... this. Removed.
12976 (scop_detection::harmful_loop_in_region): Simplify after inlining
12977 the above and remove more quadraticness.
12978 (build_scops): Adjust.
12979 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
12982 2017-09-26 Jakub Jelinek <jakub@redhat.com>
12985 * config/i386/i386.c (ix86_print_operand_address_as): Only test
12986 REGNO (base) == SP_REG if base is a REG.
12988 PR middle-end/35691
12989 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
12990 if it is different SSA_NAME.
12991 (optimize_range_tests_cmp_bitwise): New function.
12992 (optimize_range_tests): Call it.
12994 2017-09-26 Richard Biener <rguenther@suse.de>
12996 PR tree-optimization/82321
12997 * graphite.c (canonicalize_loop_closed_ssa): Properly check
12998 for the def being inside the loop.
13000 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13002 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13004 * config/s390/s390-builtins.def: Fix constraint on op4.
13006 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13008 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13009 independent expanders.
13010 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13011 ("vec_ordered", "vec_unordered"): New expanders.
13013 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13015 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13018 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13020 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13021 vec_unpacks_lo_v16qi.
13022 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13024 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13026 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13027 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13028 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13030 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13032 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13034 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13036 ("*vec_slb<mode>"): New insn pattern.
13037 ("vec_shr_<mode>"): New expander.
13038 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13039 and force the shift count operand to V16QImode.
13040 ("vec_srb<mode>"): Set shift count mode to V16QI.
13042 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13044 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13045 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13046 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13048 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13051 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13053 2017-09-26 Richard Biener <rguenther@suse.de>
13055 PR tree-optimization/82320
13056 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13059 2017-09-25 Jeff Law <law@redhat.com>
13061 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13062 prototype for new argument.
13063 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13064 mostly extracted from rs6000_emit_allocate_stack.
13065 (rs6000_emit_probe_stack_range_stack_clash): New function.
13066 (rs6000_emit_allocate_stack): Call
13067 rs6000_emit_probe_stack_range_stack_clash as needed.
13068 (rs6000_emit_probe_stack_range): Add additional argument
13069 to call to gen_probe_stack_range{si,di}.
13070 (output_probe_stack_range): New.
13071 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13072 (output_probe_stack_range_stack_clash): New.
13073 (rs6000_emit_prologue): Emit notes into dump file as requested.
13074 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13075 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13076 Add additional operand and pass it to output_probe_stack_range.
13078 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13080 PR tree-optimization/82163
13081 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13082 (checking_verify_loop_closed_ssa): New parameter.
13083 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13084 (check_loop_closed_ssa_stmt): Delete.
13085 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13086 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13087 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13090 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13092 * brig-builtins.def: Treat HSAIL barrier builtins as
13093 setjmp/longjump style functions.
13095 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13097 * target.def (constant_alignment): New hook.
13098 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13099 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13100 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13101 * doc/tm.texi: Regenerate.
13102 * targhooks.h (default_constant_alignment): Declare.
13103 (constant_alignment_word_strings): Likewise.
13104 * targhooks.c (default_constant_alignment): New function.
13105 (constant_alignment_word_strings): Likewise.
13106 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13107 instead of CONSTANT_ALIGNMENT.
13108 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13109 (force_const_mem): Likewise.
13110 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13111 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13112 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13113 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13114 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13116 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13117 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13118 constant_alignment_word_strings.
13119 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13120 (CONSTANT_ALIGNMENT): Likewise.
13121 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13122 (arm_constant_alignment): New function.
13123 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13124 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13125 constant_alignment_word_strings.
13126 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13127 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13128 constant_alignment_word_strings.
13129 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13130 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13131 (cris_constant_alignment): New function.
13132 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13133 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13134 (epiphany_constant_alignment): New function.
13135 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13136 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13137 constant_alignment_word_strings.
13138 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13139 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13140 constant_alignment_word_strings.
13141 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13142 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13143 constant_alignment_word_strings.
13144 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13145 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13146 * config/i386/i386.c (ix86_constant_alignment): Make static.
13147 Use the same interface as the target hook.
13148 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13149 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13150 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13151 constant_alignment_word_strings.
13152 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13153 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13154 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13155 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13156 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13157 constant_alignment_word_strings.
13158 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13159 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13160 constant_alignment_word_strings.
13161 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13162 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13163 constant_alignment_word_strings.
13164 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13165 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13167 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13168 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13169 * config/mips/mips.c (mips_constant_alignment): New function.
13170 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13171 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13172 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13173 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13174 (mmix_constant_alignment): Make static. Use the same interface
13175 as the target hook.
13176 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13177 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13178 constant_alignment_word_strings.
13179 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13180 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13181 constant_alignment_word_strings.
13182 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13183 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13184 constant_alignment_word_strings.
13185 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13186 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13187 (rs6000_constant_alignment): New function.
13188 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13189 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13190 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13191 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13192 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13193 (rs6000_constant_alignment): New function.
13194 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13195 * config/s390/s390.c (s390_constant_alignment): New function.
13196 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13197 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13198 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13199 constant_alignment_word_strings.
13200 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13201 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13202 (sparc_constant_alignment): New function.
13203 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13204 * config/spu/spu.c (spu_constant_alignment): New function.
13205 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13206 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13207 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13208 constant_alignment_word_strings.
13209 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13210 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13211 constant_alignment_word_strings.
13212 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13213 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13214 constant_alignment_word_strings.
13215 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13216 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13217 (visium_constant_alignment): New function.
13218 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13219 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13220 (xtensa_constant_alignment): New function.
13221 * system.h (CONSTANT_ALIGNMENT): Poison.
13223 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13225 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13226 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13227 (rs6000_builtin_valid_without_lhs): New helper function.
13228 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13229 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13231 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13233 * target.h (vec_perm_indices): Use unsigned short rather than
13235 (auto_vec_perm_indices): Likewise.
13236 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13237 Use unsigned int rather than unsigned char.
13238 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13240 2017-09-25 Richard Biener <rguenther@suse.de>
13242 * cfgloop.h (sort_sibling_loops): Declare.
13243 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13244 (sort_sibling_loops): New function sorting the sibling loop list
13246 * graphite.c (graphite_transform_loops): Sort sibling loops.
13248 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13250 * target.def (vec_perm_const_ok): Change sel parameter to
13252 * optabs-query.c (can_vec_perm_p): Update accordingly.
13253 * doc/tm.texi: Regenerate.
13254 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13255 auto_vec_perm_indices and remove separate nelt field.
13256 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13257 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13258 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13259 (aarch64_expand_vec_perm_const): Update accordingly.
13260 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13261 to vec_perm_indices.
13262 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13263 auto_vec_perm_indices and remove separate nelt field.
13264 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13265 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13266 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13268 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13269 to vec_perm_indices.
13270 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13271 sel to vec_perm_indices.
13272 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13273 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13274 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13276 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13279 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13282 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13283 on the FUNCTION_DECL function context if it has a DIE that is a
13286 2017-09-25 Richard Biener <rguenther@suse.de>
13288 PR tree-optimization/82285
13289 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13292 2017-09-25 Tom de Vries <tom@codesourcery.com>
13296 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13299 2017-09-25 Richard Biener <rguenther@suse.de>
13301 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13302 ISL errors other than isl_error_quota happen. Dump if the
13303 schedule is the same.
13304 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13305 errors instead of aborting inside ISL.
13307 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13310 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13311 of libgcc_eh for m64.
13312 * config/i386/darwin64.h: Likewise.
13314 2017-09-25 Richard Biener <rguenther@suse.de>
13316 PR middle-end/82144
13317 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13318 attribute for incomplete types nor twice for complete ones.
13320 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13323 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13324 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13327 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13330 * config/i386/i386.opt (mprefer-avx256): Use
13331 ix86_target_flags variable.
13332 * config/i386/i386.c (ix86_target_string): Move
13333 -mprefer-avx256 to flag2_opts.
13335 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13337 PR middle-end/35691
13338 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13339 and x != -1 | y != -1 into (x & y) != -1.
13341 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13343 * config.gcc: Add new case statement to set
13344 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13345 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13346 s390x-*-linux* case statements. Added aarch64 to the list of
13347 supported architectures.
13349 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13351 PR tree-optimization/82289
13352 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13353 STMT_VINFO_RELEVANT_P.
13355 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13356 Alan Hayward <alan.hayward@arm.com>
13357 David Sherwood <david.sherwood@arm.com>
13359 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13360 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13361 (extract_range_from_binary_expr_1): Don't call
13362 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13364 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13365 Alan Hayward <alan.hayward@arm.com>
13366 David Sherwood <david.sherwood@arm.com>
13368 * target.def (preferred_vector_alignment): New hook.
13369 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13371 * doc/tm.texi: Regenerate.
13372 * targhooks.h (default_preferred_vector_alignment): Declare.
13373 * targhooks.c (default_preferred_vector_alignment): New function.
13374 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13376 (DR_TARGET_ALIGNMENT): New macro.
13377 (aligned_access_p): Update commentary.
13378 (vect_known_alignment_in_bytes): New function.
13379 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13381 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13382 Calculate the misalignment based on the target alignment rather than
13384 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13385 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13386 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13387 the target alignment, rather than masking the element misalignment
13388 with the number of elements in a vector. Also use the target
13389 alignment when calculating the maximum number of peels.
13390 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13391 instead of TYPE_ALIGN_UNIT.
13392 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13393 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13394 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13395 (vect_create_data_ref_ptr): Likewise.
13396 (vect_setup_realignment): Realign by ANDing with
13397 -DR_TARGET_MISALIGNMENT.
13398 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13399 the number of peels based on DR_TARGET_ALIGNMENT.
13400 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13401 with the guaranteed alignment boundary when deciding whether
13403 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13404 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13405 (ensure_base_align): Remove stmt_info parameter. Get the
13406 target base alignment from DR_TARGET_ALIGNMENT.
13407 (vectorizable_store): Update call accordingly. Interpret
13408 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13410 (vectorizable_load): Likewise.
13412 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13413 Alan Hayward <alan.hayward@arm.com>
13414 David Sherwood <david.sherwood@arm.com>
13416 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13417 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13418 (vect_enhance_data_refs_alignment): Likewise.
13420 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13422 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13423 error. Only quit immediately if parsing is complete.
13424 (BEGIN): Initialize fatal_err and parse_done.
13425 (begin fpu, end fpu): Check number of arguments.
13426 (begin arch, end arch): Likewise.
13427 (begin cpu, end cpu): Likewise.
13428 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13429 (optalias): Likewise.
13431 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13433 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13434 * config/arm/arm-isa.h: Delete. Move definitions to ...
13435 * arm-cpus.in: ... here. Use new feature and fgroup values.
13436 * config/arm/arm.c (arm_option_override): Use lower case for feature
13438 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13439 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13440 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13442 (print_isa_bits_for): New function.
13443 (gen_isa): New function.
13444 (gen_comm_data): Use print_isa_bits_for.
13445 (define feature): New keyword.
13446 (define fgroup): New keyword.
13447 * config/arm/t-arm (TM_H): Remove.
13448 (GTM_H): Add arm-isa.h.
13449 (arm-isa.h): Add rule to generate file.
13450 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13451 case for feature bit names.
13453 2017-09-22 Richard Biener <rguenther@suse.de>
13455 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13457 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13458 print of no dependency loops ...
13459 * graphite.c (graphite_transform_loops): ... here.
13460 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13462 (same_close_phi_node, remove_duplicate_close_phi,
13463 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13464 (canonicalize_loop_closed_ssa): ... here and simplify.
13465 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13466 (optimize_isl): Use dump_printf_loc to tell when we stopped
13467 optimizing because of an ISL timeout.
13469 2017-09-22 Richard Biener <rguenther@suse.de>
13471 PR tree-optimization/82291
13472 * tree-if-conv.c (predicate_mem_writes): Make sure to
13473 remove writes in blocks predicated with false.
13475 2017-09-22 Richard Biener <rguenther@suse.de>
13477 * sese.c: Include cfganal.h.
13478 (if_region_set_false_region): Remove.
13479 (create_if_region_on_edge): Likewise.
13480 (move_sese_in_condition): Re-implement without destroying
13483 2017-09-22 Richard Biener <rguenther@suse.de>
13485 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13486 Verify both BBs contain loop PHI nodes before dispatching to
13487 copy_loop_phi_args.
13488 (graphite_regenerate_ast_isl): Do not recompute dominators,
13489 do not verify three times. Restructure for clarity.
13490 * graphite-scop-detection.c (same_close_phi_node,
13491 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13492 defined_in_loop_p, canonicalize_loop_closed_ssa,
13493 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13494 checking and SSA rewrite, move to ...
13495 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13496 (graphite_initialize): Do not pass in ctx, do not reset the
13497 SCEV cache, compute only dominators.
13498 (graphite_transform_loops): Allocate ISL ctx after
13499 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13500 Maintain post-dominators only around build_scops.
13501 * sese.c (if_region_set_false_region): Make static. Free
13502 and recompute dominators.
13503 (move_sese_in_condition): Assert we don't get called with
13504 post-dominators computed.
13505 * sese.h (if_region_set_false_region): Remove.
13507 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13509 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13510 mode attribute for TARGET_AVX512VL.
13512 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13514 * config/i386/i386.opt (mprefer-avx256): New option.
13515 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13517 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13518 for TARGET_PREFER_AVX256.
13519 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13521 2017-09-21 Jeff Law <law@redhat.com>
13523 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13524 Fix dump output if the only stack space is for pushed registers.
13526 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13528 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13531 2017-09-21 Martin Sebor <msebor@redhat.com>
13534 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13535 code (in C++) or code that triggers warnings.
13537 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13539 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13541 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13543 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13544 * sched-rgn.c: Ditto.
13545 * sel-sched-ir.c: Ditto.
13547 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13549 * toplev.h (set_random_seed): Adjust return type.
13550 * toplev.c (init_local_tick): Move eager initialization of random_seed
13551 to get_random_seed. Adjust comment.
13552 (init_random_seed): Inline to get_random_seed, delete.
13553 (get_random_seed): Initialize random_seed lazily.
13554 (set_random_seed): Do not return previous value.
13555 (print_switch_value): Do not call get_random_seed.
13557 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13559 * cgraph.c (delete_function_version): New, broken out from...
13560 (cgraph_node::delete_function_version): ...here. Rename to
13561 cgraph_node::delete_function_version_by_decl. Update all uses.
13562 (cgraph_node::remove): Call delete_function_version.
13564 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13567 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13568 VAR_DECLs to which addressable non-volatile parameters are mapped
13569 and for id->retvar after the return value assignment. Clear
13570 id->retval and id->retbnd after inlining.
13572 2017-09-21 Richard Biener <rguenther@suse.de>
13574 PR tree-optimization/82276
13575 PR tree-optimization/82244
13576 * tree-vrp.c (build_assert_expr_for): Set
13577 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13579 (remove_range_assertions): Revert earlier change.
13581 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13584 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13586 2017-09-21 Richard Biener <rguenther@suse.de>
13588 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13589 Restore valid IL after code generation errors.
13590 * graphite.c (graphite_transform_loops): Diagnose code
13591 generation issues as MSG_MISSED_OPTIMIZATION and continue
13592 with processing SCOPs.
13594 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13595 Alan Hayward <alan.hayward@arm.com>
13596 David Sherwood <david.sherwood@arm.com>
13598 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13599 rather than choosing between plus_constant and gen_rtx_<CODE>.
13600 * expr.c (emit_push_insn): Likewise.
13601 (expand_expr_real_2): Likewise.
13603 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13604 Alan Hayward <alan.hayward@arm.com>
13605 David Sherwood <david.sherwood@arm.com>
13607 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13609 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13610 Alan Hayward <alan.hayward@arm.com>
13611 David Sherwood <david.sherwood@arm.com>
13613 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13614 calling tree_to_uhwi.
13616 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13617 Alan Hayward <alan.hayward@arm.com>
13618 David Sherwood <david.sherwood@arm.com>
13620 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13621 INTEGER_CST rather than a negative test for ADDR_EXPR.
13623 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13624 Alan Hayward <alan.hayward@arm.com>
13625 David Sherwood <david.sherwood@arm.com>
13627 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13628 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13630 2017-09-21 Richard Biener <rguenther@suse.de>
13632 PR tree-optimization/71351
13633 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13634 graphite_create_new_loop_guard): Remove, fold remaining parts
13636 (translate_isl_ast_node_for): ... here and simplify.
13638 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13641 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13642 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13643 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13644 alternative always use QI mode, for -Os imov (=R,R) alternative
13645 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13648 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13649 Jeff Law <law@redhat.com>
13651 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13652 (allocate_stack_space): New function, partially extracted from
13653 s390_emit_prologue.
13654 (s390_emit_prologue): Track offset to most recent stack probe.
13655 Code to allocate space moved into allocate_stack_space.
13656 Dump actions when no stack is allocated.
13657 (s390_prologue_plus_offset): New function.
13658 (s390_emit_stack_probe): Likewise.
13660 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13662 * common.opt (Wa, Wl, Wp, g, gz=): Add
13664 (gno-column-info): Remove.
13665 (gcolumn-info): Drop RejectNegative.
13666 (gno-): New prefix.
13667 (gno-record-gcc-switches): Remove.
13668 (grecord-gcc-switches): Drop RejectNegative.
13669 (gno-split-dwarf): Remove.
13670 (gsplit-dwarf): Drop RejectNegative.
13671 (gno-strict-dwarf): Remove.
13672 (gstrict-dwarf): Drop RejectNegative.
13673 * config/darwin.opt (gfull, gused): Add RejectNegative.
13674 * dwarf2out.c (gen_producer_string): Drop
13675 gno-record-gcc-switches handler.
13676 * optc-gen.awk: Add g to prefixes with negative forms.
13677 * opts-common.c (remapping_prefix_p): New.
13678 (find_opt): Check it.
13679 (generate_canonical_option): Test g prefix.
13680 (option_map): Add -gno- mapping.
13681 (add_misspelling_candidates): Check remapping_prefix_p.
13683 2017-09-20 Jeff Law <law@redhat.com>
13685 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13686 thinko in stack clash protection support.
13688 * explow.c (compute_stack_clash_protection_loop_data): Use
13689 CONST_INT_P instead of explicit test. Verify object is a
13690 CONST_INT_P before looking at INTVAL.
13691 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13692 instead of explicit test.
13694 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13697 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13698 address instead of to r1 and r11.
13700 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13702 * config.gcc: Support "knm".
13703 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13704 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13706 * config/i386/i386.c (m_KNM): Define.
13707 (processor_target_table): Add "knm".
13709 (ix86_option_override_internal): Add "knm".
13710 (ix86_issue_rate): Add PROCESSOR_KNM.
13711 (ix86_adjust_cost): Ditto.
13712 (ia32_multipass_dfa_lookahead): Ditto.
13713 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13714 (fold_builtin_cpu): Add M_INTEL_KNM.
13715 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13716 (processor_type): Add PROCESSOR_KNM.
13717 * config/i386/x86-tune.def: Add m_KNM.
13718 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13720 2017-09-20 Richard Biener <rguenther@suse.de>
13722 PR tree-optimization/80213
13723 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13724 are allowed in empty BBs as well.
13725 (canonicalize_loop_closed_ssa): Also look for other complex
13727 (scop_detection::get_sese): Include the loop-closed PHI block
13729 (scop_detection::merge_sese): Remove code adding extra blocks.
13730 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13731 (build_scops): Assert the final returned scop is invalid.
13733 2017-09-20 Richard Biener <rguenther@suse.de>
13735 PR tree-optimization/82264
13736 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13738 (vn_phi_lookup): Likewise.
13739 (vn_phi_insert): Likewise.
13741 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13743 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13744 that fits into uhwi or shwi, add DW_AT_const_value regardless
13745 of early_dwarf without going through RTL, using add_AT_unsigned
13748 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13749 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13750 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13751 (DEBUG_MACRO_SECTION): Likewise.
13752 (DEBUG_LTO_MACRO_SECTION): Likewise.
13753 (DEBUG_STR_DWO_SECTION): Likewise.
13754 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13755 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13756 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13757 (DEBUG_LTO_LINE_STR_SECTION): Define.
13758 (init_sections_and_labels): Initialize debug_line_str_section
13759 variable. Initialize debug_loc_section for -gdwarf-5 to
13760 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13762 2017-09-20 Richard Biener <rguenther@suse.de>
13764 * graphite-sese-to-poly.c (extract_affine): Properly handle
13765 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13767 2017-09-20 Richard Biener <rguenther@suse.de>
13769 PR tree-optimization/81373
13770 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13771 Force SESE live-out defs to be handled even if they are
13774 2017-09-19 Jeff Law <law@redhat.com>
13776 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13777 nothing for stack adjustments with REG_STACK_CHECK.
13778 * sched-deps.c (parse_add_or_inc): Reject insns with
13779 REG_STACK_CHECK from dependency breaking.
13780 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13781 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13782 * reg-notes.def (STACK_CHECK): New note.
13784 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13785 (ix86_expand_prologue): Dump stack clash info as needed.
13786 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13788 * function.c (dump_stack_clash_frame_info): New function.
13789 * function.h (dump_stack_clash_frame_info): Prototype.
13790 (enum stack_clash_probes): New enum.
13792 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13793 flag_stack_clash_protection.
13794 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13795 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13796 (arm_frame_pointer_required): Likewise.
13797 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13798 (ia64_expand_prologue): Likewise.
13799 * config/mips/mips.c (mips_expand_prologue): Likewise.
13800 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13801 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13802 (sparc_flat_expand_prologue): Likewise.
13803 * config/spu/spu.c (spu_expand_prologue): Likewise.
13805 * explow.c: Include "params.h".
13806 (anti_adjust_stack_and_probe_stack_clash): New function.
13807 (get_stack_check_protect): Likewise.
13808 (compute_stack_clash_protection_loop_data): Likewise.
13809 (emit_stack_clash_protection_loop_start): Likewise.
13810 (emit_stack_clash_protection_loop_end): Likewise.
13811 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13812 Use anti_adjust_stack_and_probe_stack_clash.
13813 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13814 (emit_stack_clash_protection_loop_start): Likewise.
13815 (emit_stack_clash_protection_loop_end): Likewise.
13816 * rtl.h (get_stack_check_protect): Prototype.
13817 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13818 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13819 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13821 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13823 * doc/tm.texi: Rebuilt.
13824 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13825 get_stack_check_protect.
13826 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13827 * config/arm/arm.c (arm_expand_prologue): Likewise.
13828 (arm_frame_pointer_required): Likewise.
13829 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13830 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13831 * config/mips/mips.c (mips_expand_prologue): Likewise.
13832 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13833 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13834 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13835 (sparc_flat_expand_prologue): Likewise.
13837 * common.opt (-fstack-clash-protection): New option.
13838 * flag-types.h (enum stack_check_type): Note difference between
13839 -fstack-check= and -fstack-clash-protection.
13840 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13841 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13842 * toplev.c (process_options): Issue warnings/errors for cases
13843 not handled with -fstack-clash-protection.
13844 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13845 (-fstack-check): Note additional problem with -fstack-check=generic.
13846 Note that -fstack-check is primarily for Ada and refer users
13847 to -fstack-clash-protection for stack-clash-protection.
13848 Document new params for stack clash protection.
13850 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13852 * config/i386/i386.c (ix86_split_long_move): Do not handle
13853 address used for LEA in a special way.
13855 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13857 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
13859 2017-09-19 Martin Sebor <msebor@redhat.com>
13862 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
13863 of incompatible types.
13865 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
13867 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13868 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
13869 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13870 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
13872 2017-09-19 Richard Biener <rguenther@suse.de>
13874 PR tree-optimization/82244
13875 * tree-vrp.c (remove_range_assertions): Do not propagate
13876 a constant to abnormals but replace the assert with a copy.
13878 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
13880 PR rtl-optimization/57878
13881 PR rtl-optimization/68988
13882 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
13883 avoidance test involving non_reload_pseudos. Move frequency test
13884 below the general fragmentation avoidance test.
13886 2017-09-19 Richard Biener <rguenther@suse.de>
13888 PR tree-optimization/69728
13889 * graphite-sese-to-poly.c (schedule_error): New global.
13890 (add_loop_schedule): Handle empty domain by failing the
13892 (build_original_schedule): Handle schedule_error.
13894 2017-09-19 Richard Biener <rguenther@suse.de>
13896 * graphite-scop-detection.c (scop_detection::can_represent_loop):
13897 Do not iterate to sibling loops but only to siblings of inner
13900 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
13903 * config/m68k/m68k.md (moveq feeding equality comparison): Check
13904 that the registers are different.
13906 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
13908 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
13909 to processor_model and "amdfam17h" to arch_names_table.
13910 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
13912 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13915 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
13917 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13918 Alan Hayward <alan.hayward@arm.com>
13919 David Sherwood <david.sherwood@arm.com>
13921 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
13923 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
13925 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
13926 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
13927 vect_schedule_slp_instance.
13928 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
13929 Update call to vect_slp_analyze_node_operations. Simplify return
13931 (vect_slp_analyze_bb_1): Update call accordingly.
13932 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
13933 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
13934 (vect_schedule_slp): Update call accordingly.
13936 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13937 Alan Hayward <alan.hayward@arm.com>
13938 David Sherwood <david.sherwood@arm.com>
13940 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
13941 with types that aren't in fact scalar.
13943 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13945 * tree-vect-slp.c (vect_record_max_nunits): New function,
13947 (vect_build_slp_tree_1): ...here.
13948 (vect_build_slp_tree_2): Call it for phis too.
13950 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13952 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
13953 to vect_get_vec_def_for_operand when getting the mask operand.
13955 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13956 Alan Hayward <alan.hayward@arm.com>
13957 David Sherwood <david.sherwood@arm.com>
13959 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
13962 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13963 Alan Hayward <alan.hayward@arm.com>
13964 David Sherwood <david.sherwood@arm.com>
13966 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
13967 calculation for vector booleans.
13969 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13970 Alan Hayward <alan.hayward@arm.com>
13971 David Sherwood <david.sherwood@arm.com>
13973 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
13975 (vect_transform_stmt): ...here.
13976 (vect_analyze_stmt): Use it instead of calling
13977 vectorizable_live_operation directly.
13979 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
13981 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
13982 non-SIMT targets in acc vector loops.
13984 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
13986 * configure.ac: Add arc and check if assembler supports gdwarf2.
13987 * configure: Regenerate.
13989 2017-09-18 Richard Biener <rguenther@suse.de>
13991 PR tree-optimization/82220
13992 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
13993 epilogue niters from the min_profitable_iters compute.
13995 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13998 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
13999 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14000 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14002 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14005 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14006 switching to a new text section.
14008 2017-09-18 Richard Biener <rguenther@suse.de>
14010 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14012 (build_alias_set): Reject aliases with no access function.
14014 2017-09-18 Richard Biener <rguenther@suse.de>
14016 PR tree-optimization/79622
14017 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14019 (build_cross_bb_scalars_use): Likewise.
14021 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14023 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14025 2017-09-18 Alan Modra <amodra@gmail.com>
14028 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14029 stack_pointer_rtx for count 0. Update comments. Break up
14030 large rtl expression.
14032 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14034 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14035 Increase to 20 bytes.
14036 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14037 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14038 or avx version of the stub.
14040 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14043 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14044 compute the minimum stack alignment. Also update preferred stack
14045 boundary for leaf functions.
14047 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14049 PR tree-optimization/82228
14050 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14053 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14055 * common/config/nds32/nds32-common.c
14056 (nds32_option_optimization_table): Refine formatting.
14057 (nds32_option_optimization_table): Use -fsched-pressure and
14058 -fomit-frame-pointer for specific optimization level.
14060 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14062 * config/nds32/nds32.c: Refine formatting and comments.
14063 * config/nds32/nds32.h: Likewise.
14064 * config/nds32/nds32.md: Likewise.
14065 * config/nds32/nds32-cost.c: Likewise.
14066 * config/nds32/nds32-isr.c: Likewise.
14067 * config/nds32/nds32-md-auxiliary.c: Likewise.
14068 * config/nds32/nds32-multiple.md: Likewise.
14069 * config/nds32/nds32-predicates.c: Likewise.
14071 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14072 Jakub Jelinek <jakub@redhat.com>
14074 Add support for -std=c++2a.
14075 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14077 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14079 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14082 * doc/extend.texi (Common Function Attributes): Add
14083 references to ARM, AArch64, and S/390 specific attributes.
14084 (Function Specific Option Pragmas): Add AArch64 and S/390
14085 to list of back ends that support the target pragma.
14087 2017-09-15 Nathan Sidwell <nathan@acm.org>
14089 * doc/standards.texi: Fix C++17 description. Update URLs for
14092 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14094 * common.opt (Wcast-align=strict): New warning option.
14095 * doc/invoke.texi: Document -Wcast-align=strict.
14097 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14099 * cgraph.h (cgraph_thunk_info): Add comments.
14100 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14101 assert for VIRTUAL_* arguments stricter.
14103 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14105 PR tree-optimization/71026
14106 * match.pd: Move RDIV patterns from fold-const.c
14107 * fold-const.c (distribute_real_division): Removed.
14108 (fold_binary_loc): Remove calls to distribute_real_divison.
14110 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14112 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14113 c++1z and gnu++1z as deprecated. Change other references to
14114 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14115 Change -Wc++1z-compat to -Wc++17-compat.
14116 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14117 * dwarf2out.c (highest_c_language): Handle C++17.
14118 (gen_compile_unit_die): Likewise.
14120 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14122 PR rtl-optimization/82192
14123 * combine.c (make_extraction): Don't look through non-paradoxical
14124 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14127 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14128 Alan Hayward <alan.hayward@arm.com>
14129 David Sherwood <david.sherwood@arm.com>
14131 * target.def (function_arg_offset): New hook.
14132 * targhooks.h (default_function_arg_offset): Declare.
14133 * targhooks.c (default_function_arg_offset): New function.
14134 * function.c (locate_and_pad_parm): Use
14135 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14136 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14137 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14138 * doc/tm.texi: Regenerate.
14139 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14140 * config/spu/spu.c (spu_function_arg_offset): New function.
14141 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14142 * system.h (FUNCTION_ARG_OFFSET): Poison.
14144 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14145 Alan Hayard <alan.hayward@arm.com>
14146 David Sherwood <david.sherwood@arm.com>
14148 * target.def (truly_noop_truncation): New hook.
14149 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14150 than TRULY_NOOP_TRUNCATION.
14151 * hooks.h (hook_bool_uint_uint_true): Declare.
14152 * hooks.c (hook_bool_uint_uint_true): New function.
14153 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14154 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14155 * doc/tm.texi: Regenerate.
14156 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14157 rather than TRULY_NOOP_TRUNCATION in comments.
14158 (simplify_comparison): Likewise.
14159 (record_truncated_value): Likewise.
14160 * expmed.c (extract_bit_field_1): Likewise.
14161 (extract_split_bit_field): Likewise.
14162 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14163 instead of TRULY_NOOP_TRUNCATION.
14164 * function.c (assign_parm_setup_block): Likewise.
14165 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14166 * rtlhooks.c: Include target.h.
14167 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14168 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14169 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14170 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14171 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14172 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14173 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14174 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14175 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14176 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14177 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14178 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14179 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14180 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14181 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14182 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14183 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14184 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14185 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14186 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14187 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14188 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14189 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14190 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14191 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14192 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14193 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14194 rather than TRULY_NOOP_TRUNCATION in comments.
14195 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14196 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14197 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14198 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14199 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14200 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14201 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14202 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14203 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14204 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14205 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14206 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14207 rather than TRULY_NOOP_TRUNCATION in comments.
14208 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14209 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14210 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14211 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14212 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14213 TRULY_NOOP_TRUNCATION condition.
14214 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14215 (TRULY_NOOP_TRUNCATION): Delete.
14216 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14217 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14218 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14219 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14220 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14221 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14222 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14223 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14224 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14225 rather than TRULY_NOOP_TRUNCATION in comments.
14226 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14227 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14228 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14229 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14230 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14231 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14233 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14236 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14237 (*cmp_ior): Likewise.
14238 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14239 (*ior_scc_scc_cmp): Likewise.
14240 (*and_scc_scc): Likewise.
14241 (*and_scc_scc_cmp): Likewise.
14243 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14244 Alan Hayard <alan.hayward@arm.com>
14245 David Sherwood <david.sherwood@arm.com>
14247 * target.def (can_change_mode_class): New hook.
14248 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14249 (hard_regno_nregs): Likewise.
14250 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14251 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14252 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14253 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14254 (LOAD_EXTEND_OP): Update accordingly.
14255 * doc/tm.texi: Regenerate.
14256 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14257 CANNOT_CHANGE_MODE_CLASS.
14258 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14259 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14260 * combine.c (simplify_set): Update accordingly.
14261 * emit-rtl.c (validate_subreg): Likewise.
14262 * recog.c (general_operand): Likewise.
14263 * regcprop.c (mode_change_ok): Likewise.
14264 * reload1.c (choose_reload_regs): Likewise.
14265 (inherit_piecemeal_p): Likewise.
14266 * rtlanal.c (simplify_subreg_regno): Likewise.
14267 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14268 instead of CANNOT_CHANGE_MODE_CLASS.
14269 (reload_cse_simplify_operands): Likewise.
14270 * reload.c (push_reload): Use targetm.can_change_mode_class
14271 instead of CANNOT_CHANGE_MODE_CLASS.
14272 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14273 REG_CANNOT_CHANGE_MODE_P.
14274 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14275 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14276 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14277 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14278 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14279 (arm_can_change_mode_class): New function.
14280 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14281 than CANNOT_CHANGE_MODE_CLASS in comments.
14282 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14283 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14284 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14285 (ix86_can_change_mode_class): ...this new function, inverting the
14286 sense of the return value.
14287 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14288 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14289 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14290 (ia64_can_change_mode_class): New function.
14291 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14292 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14293 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14294 (m32c_can_change_mode_class): ...this new function, inverting the
14295 sense of the return value.
14296 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14297 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14298 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14299 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14300 (mips_can_change_mode_class): ...this new function, inverting the
14301 sense of the return value.
14302 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14303 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14304 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14305 (msp430_can_change_mode_class): New function.
14306 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14307 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14308 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14309 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14310 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14311 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14312 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14313 (pa_cannot_change_mode_class): Replace with...
14314 (pa_can_change_mode_class): ...this new function, inverting the
14315 sense of the return value.
14316 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14317 than CANNOT_CHANGE_MODE_CLASS in comments.
14318 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14319 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14320 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14321 (pdp11_cannot_change_mode_class): Replace with...
14322 (pdp11_can_change_mode_class): ...this new function, inverting the
14323 sense of the return value.
14324 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14325 * config/powerpcspe/powerpcspe-protos.h
14326 (rs6000_cannot_change_mode_class_ptr): Delete.
14327 * config/powerpcspe/powerpcspe.c
14328 (rs6000_cannot_change_mode_class_ptr): Delete.
14329 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14330 (rs6000_option_override_internal): Assign to
14331 targetm.can_change_mode_class instead of
14332 rs6000_cannot_change_mode_class_ptr.
14333 (rs6000_cannot_change_mode_class): Replace with...
14334 (rs6000_can_change_mode_class): ...this new function, inverting the
14335 sense of the return value.
14336 (rs6000_debug_cannot_change_mode_class): Replace with...
14337 (rs6000_debug_can_change_mode_class): ...this new function.
14338 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14339 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14340 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14341 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14342 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14344 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14345 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14346 (rs6000_option_override_internal): Assign to
14347 targetm.can_change_mode_class instead of
14348 rs6000_cannot_change_mode_class_ptr.
14349 (rs6000_cannot_change_mode_class): Replace with...
14350 (rs6000_can_change_mode_class): ...this new function, inverting the
14351 sense of the return value.
14352 (rs6000_debug_cannot_change_mode_class): Replace with...
14353 (rs6000_debug_can_change_mode_class): ...this new function.
14354 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14355 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14356 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14357 (s390_can_change_mode_class): ...this new function, inverting the
14358 sense of the return value.
14359 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14360 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14361 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14362 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14363 (sh_cannot_change_mode_class): Replace with...
14364 (sh_can_change_mode_class): ...this new function, inverting the
14365 sense of the return value.
14366 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14367 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14368 (sparc_can_change_mode_class): New function.
14369 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14370 * config/spu/spu.c (spu_can_change_mode_class): New function.
14371 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14372 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14373 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14374 (visium_can_change_mode_class): New function.
14375 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14377 2017-09-15 Richard Biener <rguenther@suse.de>
14379 PR tree-optimization/82217
14380 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14381 but not undefined case.
14383 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14386 * postreload.c (reload_cse_simplify_operands): Skip
14387 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14389 2017-09-15 Richard Biener <rguenther@suse.de>
14391 PR tree-optimization/68823
14392 * graphite-scop-detection.c (build_alias_set): If we have a
14393 possible dependence check whether we can handle them by just
14394 looking at the DRs DR_ACCESS_FNs.
14395 (build_scops): If build_alias_set fails, fail the SCOP.
14397 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14399 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14400 to support float128 built-in functions that require the ISA 3.0
14402 (BU_FLOAT128_3_HW): Likewise.
14403 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14404 built-in functions.
14405 (FMAF128): Likewise.
14407 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14408 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14409 floating point instructions.
14410 (rs6000_invalid_builtin): Likewise.
14411 (rs6000_builtin_mask_names): Likewise.
14412 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14413 (RS6000_BTM_FLOAT128_HW): Likewise.
14414 (RS6000_BTM_COMMON): Likewise.
14415 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14417 * doc/extend.texi (RS/6000 built-in functions): Document the
14418 IEEE 128-bit floating point square root and fused multiply-add
14419 built-in functions.
14421 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14423 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14424 reg (r2) isn't in the set of registers defined in the prologue.
14426 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14427 Alan Hayward <alan.hayward@arm.com>
14428 David Sherwood <david.sherwood@arm.com>
14430 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14431 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14432 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14433 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14434 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14436 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14437 max_vectorization_factor.
14438 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14440 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14441 Alan Hayward <alan.hayward@arm.com>
14442 David Sherwood <david.sherwood@arm.com>
14444 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14445 (vect_worthwhile_without_simd_p): Declare.
14446 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14447 (vectorizable_reduction): Use it.
14448 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14449 (vectorizable_operation): Likewise.
14451 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14452 Alan Hayward <alan.hayward@arm.com>
14453 David Sherwood <david.sherwood@arm.com>
14455 * tree-vectorizer.h (vect_get_num_copies): New function.
14456 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14457 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14458 (vectorizable_induction): Likewise.
14459 (vectorizable_live_operation): Likewise.
14460 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14461 (vectorizable_bswap): Likewise.
14462 (vectorizable_call): Likewise.
14463 (vectorizable_conversion): Likewise.
14464 (vectorizable_assignment): Likewise.
14465 (vectorizable_shift): Likewise.
14466 (vectorizable_operation): Likewise.
14467 (vectorizable_store): Likewise.
14468 (vectorizable_load): Likewise.
14469 (vectorizable_condition): Likewise.
14470 (vectorizable_comparison): Likewise.
14471 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14473 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14474 Alan Hayward <alan.hayward@arm.com>
14475 David Sherwood <david.sherwood@arm.com>
14477 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14478 of vect_init_vector.
14480 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14481 Alan Hayward <alan.hayward@arm.com>
14482 David Sherwood <david.sherwood@arm.com>
14484 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14485 an inline wrapper that provides a location.
14486 (gimple_build_vector): Likewise.
14487 * gimple-fold.c (gimple_build_vector_from_val): New function.
14488 (gimple_build_vector): Likewise.
14489 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14490 functions to build the initial value. Always return a gimple value.
14491 (get_initial_defs_for_reduction): Likewise. Only compute
14493 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14494 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14495 (vectorizable_induction): Use gimple_build_vector rather than
14498 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14499 Alan Hayward <alan.hayward@arm.com>
14500 David Sherwood <david.sherwood@arm.com>
14502 * target.h (vec_perm_indices): New typedef.
14503 (auto_vec_perm_indices): Likewise.
14504 * optabs-query.h: Include target.h
14505 (can_vec_perm_p): Take a vec_perm_indices *.
14506 * optabs-query.c (can_vec_perm_p): Likewise.
14507 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14508 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14509 * tree-vect-generic.c (lower_vec_perm): Likewise.
14510 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14511 (vect_grouped_load_supported): Likewise.
14512 (vect_shift_permute_load_chain): Likewise.
14513 (vect_permute_store_chain): Use auto_vec_perm_indices.
14514 (vect_permute_load_chain): Likewise.
14515 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14516 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14517 Update uses of can_vec_perm_p.
14518 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14519 mode with a number of elements. Take a vec_perm_indices *.
14520 (vect_create_epilog_for_reduction): Update accordingly.
14521 Use auto_vec_perm_indices.
14522 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14523 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14524 (vect_transform_slp_perm_load): Likewise.
14525 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14526 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14527 (vect_gen_perm_mask_checked): Likewise.
14528 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14529 (vect_gen_perm_mask_checked): Likewise.
14530 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14531 (vectorizable_store): Likewise.
14532 (vectorizable_load): Likewise.
14533 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14534 (vectorizable_bswap): Likewise.
14536 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14537 Alan Hayward <alan.hayward@arm.com>
14538 David Sherwood <david.sherwood@arm.com>
14540 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14541 * tree.c (build_vector): Likewise.
14542 (build_vector_from_ctor): Update accordingly.
14543 (build_vector_from_val): Likewise.
14544 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14545 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14546 * tree-vect-generic.c (add_rshift): Likewise.
14547 (expand_vector_divmod): Likewise.
14548 (optimize_vector_constructor): Likewise.
14549 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14550 (vect_transform_slp_perm_load): Likewise.
14551 (vect_schedule_slp_instance): Likewise.
14552 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14553 (vectorizable_call): Likewise.
14554 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14555 * expmed.c (make_tree): Likewise.
14556 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14557 a vector passed to build_vector.
14558 (fold_convert_const): Likewise.
14559 (exact_inverse): Likewise.
14560 (fold_ternary_loc): Likewise.
14561 (fold_relational_const): Likewise.
14562 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14563 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14564 (const_unop): Likewise. Store the reduction accumulator in a
14565 variable rather than an array.
14566 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14567 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14568 the new vector, rather than constructing it after the input arrays.
14569 (native_interpret_vector): Use auto_vec<tree> when building
14570 a vector passed to build_vector. Add elements in order.
14571 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14572 auto_vec<tree> when building a vector passed to build_vector.
14573 (vect_create_epilog_for_reduction): Likewise.
14574 (vectorizable_induction): Likewise.
14575 (get_initial_def_for_reduction): Likewise. Fix indentation of
14577 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14579 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14580 passed to build_vector.
14582 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14583 Alan Hayward <alan.hayward@arm.com>
14584 David Sherwood <david.sherwood@arm.com>
14586 * tree-core.h (tree_base::u): Add an "nelts" field.
14587 (tree_vector): Use VECTOR_CST_NELTS as the length.
14588 * tree.c (tree_size): Likewise.
14589 (make_vector): Initialize VECTOR_CST_NELTS.
14590 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14591 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14592 TYPE_VECTOR_SUBPARTS.
14593 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14594 as the number of units, setting it from VECTOR_CST_NELTS.
14595 (const_vector_from_tree): Likewise.
14596 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14597 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14598 (fold_negate_expr_1): Likewise.
14599 (fold_convert_const): Likewise.
14600 (const_binop): Likewise. Differentiate the number of output and
14602 (const_unop): Likewise.
14603 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14604 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14605 in cases that did the opposite.
14607 2017-09-14 Richard Biener <rguenther@suse.de>
14609 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14612 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14614 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14616 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14619 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14620 if and where to split a bb, except for splitting before debug insn
14621 sequences followed by non-label real insn. Delete debug insns
14622 in between basic blocks.
14624 * combine.c (make_compound_operation_int): Formatting fixes.
14626 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14627 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14628 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14629 * config/sol2.h (LINK_EH_SPEC): Likewise.
14630 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14631 * config/s390/linux.h (LINK_SPEC): Likewise.
14632 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14633 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14634 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14635 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14636 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14637 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14638 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14639 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14641 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14643 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14645 (ENDFILE_LINUX_SPEC): Likewise.
14646 (LINK_EH_SPEC): Likewise.
14647 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14648 (LINK_OS_LINUX_SPEC32): Likewise.
14649 (LINK_OS_LINUX_SPEC64): Likewise.
14650 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14651 (LINK_OS_LINUX_SPEC): Likewise.
14653 2017-09-13 Martin Liska <mliska@suse.cz>
14655 PR middle-end/82154
14656 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14657 CASE_HIGH is NULL_TREE.
14659 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14660 Alan Hayward <alan.hayward@arm.com>
14661 David Sherwood <david.sherwood@arm.com>
14663 * target.def (secondary_memory_needed): New hook.
14664 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14665 instead of SECONDARY_MEMORY_NEEDED.
14666 (secondary_memory_needed_mode): Likewise.
14667 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14668 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14669 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14670 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14671 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14672 * doc/tm.texi: Regenerate.
14673 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14674 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14675 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14676 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14677 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14678 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14679 mode argument first and change the reg_class arguments to reg_class_t.
14680 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14681 Make static. Update the call to inline_secondary_memory_needed.
14682 (ix86_register_move_cost): Update the call to
14683 inline_secondary_memory_needed.
14684 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14685 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14687 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14688 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14690 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14691 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14692 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14693 and match hook interface. Add comment from mips.h.
14694 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14695 * config/mmix/mmix.md (truncdfsf2): Refer to
14696 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14698 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14699 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14700 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14701 (pa_secondary_memory_needed): New function.
14702 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14703 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14704 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14705 (pdp11_secondary_memory_needed): Make static and match hook interface.
14706 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14707 * config/powerpcspe/powerpcspe-protos.h
14708 (rs6000_secondary_memory_needed_ptr): Delete.
14709 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14711 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14712 (rs6000_option_override_internal): Assign to
14713 targetm.secondary_memory_needed rather than
14714 rs6000_secondary_memory_needed_ptr.
14715 (rs6000_secondary_memory_needed): Match hook interface.
14716 (rs6000_debug_secondary_memory_needed): Likewise.
14717 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14718 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14719 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14720 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14721 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14722 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14724 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14725 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14726 (rs6000_option_override_internal): Assign to
14727 targetm.secondary_memory_needed rather than
14728 rs6000_secondary_memory_needed_ptr.
14729 (rs6000_secondary_memory_needed): Match hook interface.
14730 (rs6000_debug_secondary_memory_needed): Likewise.
14731 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14732 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14733 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14734 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14735 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14736 (sparc_secondary_memory_needed): New function.
14737 * lra-constraints.c (check_and_process_move): Refer to
14738 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14740 (curr_insn_transform): Likewise.
14741 (process_alt_operands): Use targetm.secondary_memory_needed
14742 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14743 (check_secondary_memory_needed_p): Likewise.
14744 (choose_split_class): Likewise.
14745 * reload.c: Unconditionally include code that was previously
14746 conditional on SECONDARY_MEMORY_NEEDED.
14747 (push_secondary_reload): Use targetm.secondary_memory_needed
14748 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14749 (push_reload): Likewise.
14750 * reload1.c: Unconditionally include code that was previously
14751 conditional on SECONDARY_MEMORY_NEEDED.
14752 (choose_reload_regs): Use targetm.secondary_memory_needed
14753 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14754 (gen_reload): Likewise.
14755 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14757 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14758 Alan Hayward <alan.hayward@arm.com>
14759 David Sherwood <david.sherwood@arm.com>
14761 * target.def (secondary_memory_needed_mode): New hook:
14762 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14763 * targhooks.h (default_secondary_memory_needed_mode): New function.
14764 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14765 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14766 * doc/tm.texi: Regenerate.
14767 * lra-constraints.c (check_and_process_move): Use
14768 targetm.secondary_memory_needed_mode instead of
14769 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14770 (curr_insn_transform): Likewise.
14771 * reload.c (get_secondary_mem): Likewise.
14772 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14773 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14775 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14776 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14777 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14778 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14779 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14781 * config/powerpcspe/powerpcspe-protos.h
14782 (rs6000_secondary_memory_needed_mode): Delete.
14783 * config/powerpcspe/powerpcspe.c
14784 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14785 (rs6000_secondary_memory_needed_mode): Make static.
14786 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14787 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14789 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14791 (rs6000_secondary_memory_needed_mode): Make static.
14792 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14793 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14794 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14795 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14796 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14798 (sparc_secondary_memory_needed_mode): New function.
14799 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14801 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14803 * config/aarch64/constraints.md (Umq): New constraint.
14804 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14806 (mov<mode>): Update condition.
14808 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14810 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14811 when bitposition is the same.
14813 2017-09-13 Richard Biener <rguenther@suse.de>
14815 * dwarf2out.c (output_die_symbol): Remove.
14816 (output_die): Do not output a DIEs symbol.
14818 2017-09-13 Richard Biener <rguenther@suse.de>
14820 PR middle-end/82128
14821 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14822 default-def to avoid breaking iterator update with the weird
14823 interaction with cgraph_update_edges_for_call_stmt_node.
14825 2017-09-13 Richard Biener <rguenther@suse.de>
14827 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14828 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14829 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14830 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14832 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14834 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14835 Disable pc relative literal load irrespective of
14836 TARGET_FIX_ERR_A53_84341 for default.
14838 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14840 * config/sparc/sparc.c (output_return): Output the source location of
14841 the insn in the delay slot, if any.
14842 (output_sibcall): Likewise.
14844 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14847 * common.opt (-static-pie): New alias.
14848 (shared): Negate static-pie.
14849 (-no-pie): Update help text.
14851 (static-pie): New option.
14852 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14853 -static-pie support.
14854 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14855 (LINK_EH_SPEC): Likewise.
14856 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14857 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14858 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14859 * gcc.c (LINK_COMMAND_SPEC): Likewise.
14860 (init_gcc_specs): Likewise.
14861 (init_spec): Likewise.
14862 (display_help): Update help message for -pie.
14863 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
14866 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
14868 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
14869 (movdi_aarch64): Likewise.
14870 (movti_aarch64): Likewise.
14872 2017-09-12 Simon Wright <simon@pushface.org>
14875 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
14876 calculation of the minor version, always output as 0.
14878 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14881 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14882 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
14883 on it early, rather than manual conversion late. For
14884 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
14885 instead of performing manual conversion.
14887 2017-09-12 Carl Love <cel@us.ibm.com>
14889 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
14890 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
14892 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14893 VMULOSW): Add definitions.
14894 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14895 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14896 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14897 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14898 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14900 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14902 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
14904 (movti_aarch64): Likewise.
14905 (movdf_aarch64): Likewise.
14906 (movtf_aarch64): Likewise.
14907 (load_pairdi): Likewise.
14908 (store_pairdi): Likewise.
14909 (load_pairdf): Likewise.
14910 (store_pairdf): Likewise.
14911 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
14912 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
14913 (ldr_got_small_<mode>): Likewise.
14914 (ldr_got_small_28k_<mode>): Likewise.
14915 (ldr_got_tiny): Likewise.
14916 * config/aarch64/iterators.md (ldst_sz): New.
14917 (ldpstp_sz): Likewise.
14918 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
14920 (thunderx_load): Split load_8 to load_16.
14921 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
14923 (thunderx2t99_storepair_basic): Split store_8 to store_16.
14924 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
14925 (xgene1_store_pair): Split store_8 to store_16.
14926 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
14927 (falkor_st_0_st_sd): Split store_8 to store_16.
14929 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14931 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
14932 and store1/2/3/4 to store_4/8/12/16.
14933 * config/aarch64/aarch64.md: Update for rename.
14934 * config/arm/arm.md: Likewise.: Likewise.
14935 * config/arm/arm.c: Likewise.
14936 * config/arm/thumb1.md: Likewise.
14937 * config/arm/thumb2.md: Likewise.
14938 * config/arm/vfp.md: Likewise.
14939 * config/arm/arm-generic.md: Likewise.
14940 * config/arm/arm1020e.md: Likewise.
14941 * config/arm/arm1026ejs.md: Likewise.
14942 * config/arm/arm1136jfs.md: Likewise.
14943 * config/arm/arm926ejs.md: Likewise.
14944 * config/arm/cortex-a15.md: Likewise.
14945 * config/arm/cortex-a17.md: Likewise.
14946 * config/arm/cortex-a5.md: Likewise.
14947 * config/arm/cortex-a53.md: Likewise.
14948 * config/arm/cortex-a57.md: Likewise.
14949 * config/arm/cortex-a7.md: Likewise.
14950 * config/arm/cortex-a8.md: Likewise.
14951 * config/arm/cortex-a9.md: Likewise.
14952 * config/arm/cortex-m4.md: Likewise.
14953 * config/arm/cortex-m7.md: Likewise.
14954 * config/arm/cortex-r4.md: Likewise.
14955 * config/arm/exynos-m1.md: Likewise.
14956 * config/arm/fa526.md: Likewise.
14957 * config/arm/fa606te.md: Likewise.
14958 * config/arm/fa626te.md: Likewise.
14959 * config/arm/fa726te.md: Likewise.
14960 * config/arm/fmp626.md: Likewise.
14961 * config/arm/iwmmxt.md: Likewise.
14962 * config/arm/ldmstm.md: Likewise.
14963 * config/arm/marvell-pj4.md: Likewise.
14964 * config/arm/xgene1.md: Likewise.
14965 * config/aarch64/thunderx.md: Likewise.
14966 * config/aarch64/thunderx2t99.md: Likewise.
14967 * config/aarch64/falkor.md: Likewise.
14969 2017-09-12 Martin Liska <mliska@suse.cz>
14971 * attribs.c (private_lookup_attribute): New function.
14972 * attribs.h (private_lookup_attribute): Declared here.
14973 (lookup_attribute): Called from this place.
14975 2017-09-12 Richard Biener <rguenther@suse.de>
14977 PR tree-optimization/82157
14978 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
14979 stmts with side-effects.
14981 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14982 Alan Hayward <alan.hayward@arm.com>
14983 David Sherwood <david.sherwood@arm.com>
14985 * target.def (hard_regno_nregs): New hook.
14986 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
14987 * targhooks.h (default_hard_regno_nregs): Declare.
14988 * targhooks.c (default_hard_regno_nregs): New function.
14989 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
14990 (TARGET_HARD_REGNO_NREGS): ...this hook.
14991 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
14992 (CLASS_MAX_NREGS): Likewise.
14993 * doc/tm.texi: Regenerate.
14994 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
14995 instead of HARD_REGNO_NREGS.
14996 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
14997 HARD_REGNO_NREGS in the comment.
14998 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
14999 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15000 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15001 Return an unsigned int.
15002 (TARGET_HARD_REGNO_NREGS): Redefine.
15003 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15004 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15005 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15006 (arc_hard_regno_nregs): New function.
15007 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15008 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15009 (arm_hard_regno_nregs): New function.
15010 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15011 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15012 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15013 (TARGET_HARD_REGNO_NREGS): Redefine.
15014 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15015 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15016 (HARD_REGNO_NREGS): Delete.
15017 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15018 (cr16_hard_regno_nregs): New function.
15019 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15020 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15021 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15022 (cris_hard_regno_nregs): New function.
15023 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15024 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15025 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15026 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15027 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15028 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15029 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15030 (frv_hard_regno_nregs): Make static. Take and return an
15032 (frv_class_max_nregs): Remove outdated copy of documentation.
15033 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15034 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15035 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15036 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15037 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15038 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15039 (TARGET_HARD_REGNO_NREGS): Redefine.
15040 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15041 (CLASS_MAX_NREGS): Update comment.
15042 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15043 (ia64_hard_regno_nregs): New function.
15044 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15045 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15046 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15047 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15048 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15050 (m32c_hard_regno_nregs): Likewise. Make static.
15051 (TARGET_HARD_REGNO_NREGS): Redefine.
15052 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15053 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15054 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15055 (m68k_hard_regno_nregs): New function.
15056 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15057 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15058 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15059 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15060 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15061 Take and return an unsigned int.
15062 (TARGET_HARD_REGNO_NREGS): Redefine.
15063 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15064 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15065 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15066 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15067 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15068 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15069 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15070 (msp430_hard_regno_nregs): Make static. Take and return an
15072 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15073 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15074 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15075 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15076 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15077 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15078 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15079 (TARGET_HARD_REGNO_NREGS): Redefine.
15080 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15081 (PA_HARD_REGNO_NREGS): ...this.
15082 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15083 (PA_HARD_REGNO_NREGS): ...this.
15084 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15085 (pa_hard_regno_nregs): New function.
15086 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15087 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15088 (pdp11_hard_regno_nregs): New function.
15089 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15090 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15091 (rs6000_hard_regno_nregs_hook): New function.
15092 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15093 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15094 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15095 Take and return an unsigned int. Move earlier in file.
15096 (TARGET_HARD_REGNO_NREGS): Redefine.
15097 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15098 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15099 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15100 (rl78_hard_regno_nregs): Make static. Take and return an
15102 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15103 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15104 (rs6000_hard_regno_nregs_hook): New function.
15105 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15106 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15107 (TARGET_HARD_REGNO_NREGS): Redefine.
15108 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15109 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15110 instead of HARD_REGNO_NREGS.
15111 (s390_hard_regno_nregs): New function.
15112 (s390_hard_regno_mode_ok): Add comment from s390.h.
15113 (TARGET_HARD_REGNO_NREGS): Redefine.
15114 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15115 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15116 (sh_hard_regno_nregs): New function.
15117 (sh_pass_in_reg_p): Use it.
15118 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15119 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15120 (sparc_hard_regno_nregs): New function.
15121 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15122 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15123 (spu_function_arg_advance): Use it, supplying a valid register number.
15124 (TARGET_HARD_REGNO_NREGS): Redefine.
15125 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15126 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15127 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15128 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15129 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15130 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15131 (CLASS_MAX_NREGS): Remove copy of old documentation.
15132 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15133 (visium_hard_regno_nregs): New function.
15134 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15135 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15136 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15137 xtensa_hard_regno_nregs): New function.
15138 * system.h (HARD_REGNO_NREGS): Poison.
15140 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15142 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15143 hard_regno_nregs instead of HARD_REGNO_NREGS.
15144 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15145 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15146 (c6x_expand_epilogue): Likewise.
15147 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15148 (frv_read_iacc_argument): Likewise.
15149 * config/sh/sh.c: Include regs.h.
15150 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15151 (regs_used): Likewise.
15152 (output_stack_adjust): Likewise.
15153 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15154 * expmed.c: Include regs.h.
15155 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15156 * ree.c: Include regs.h.
15157 (combine_reaching_defs): Use hard_regno_nregs instead of
15159 (add_removable_extension): Likewise.
15161 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15163 * regs.h (hard_regno_nregs): Turn into a function.
15164 (end_hard_regno): Update accordingly.
15165 * caller-save.c (setup_save_areas): Likewise.
15166 (save_call_clobbered_regs): Likewise.
15167 (replace_reg_with_saved_mem): Likewise.
15168 (insert_restore): Likewise.
15169 (insert_save): Likewise.
15170 * combine.c (can_change_dest_mode): Likewise.
15171 (move_deaths): Likewise.
15172 (distribute_notes): Likewise.
15173 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15174 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15175 (rs6000_split_multireg_move): Likewise.
15176 (rs6000_register_move_cost): Likewise.
15177 (rs6000_memory_move_cost): Likewise.
15178 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15179 (rs6000_split_multireg_move): Likewise.
15180 (rs6000_register_move_cost): Likewise.
15181 (rs6000_memory_move_cost): Likewise.
15182 * cselib.c (cselib_reset_table): Likewise.
15183 (cselib_lookup_1): Likewise.
15184 * emit-rtl.c (set_mode_and_regno): Likewise.
15185 * function.c (aggregate_value_p): Likewise.
15186 * ira-color.c (setup_profitable_hard_regs): Likewise.
15187 (check_hard_reg_p): Likewise.
15188 (calculate_saved_nregs): Likewise.
15189 (assign_hard_reg): Likewise.
15190 (improve_allocation): Likewise.
15191 (calculate_spill_cost): Likewise.
15192 * ira-emit.c (modify_move_list): Likewise.
15193 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15194 (ira_hard_reg_in_set_p): Likewise.
15195 * ira.c (setup_reg_mode_hard_regset): Likewise.
15196 (clarify_prohibited_class_mode_regs): Likewise.
15197 (check_allocation): Likewise.
15198 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15199 (lra_setup_reg_renumber): Likewise.
15200 (setup_try_hard_regno_pseudos): Likewise.
15201 (spill_for): Likewise.
15202 (assign_hard_regno): Likewise.
15203 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15204 * lra-constraints.c (in_class_p): Likewise.
15205 (lra_constraint_offset): Likewise.
15206 (simplify_operand_subreg): Likewise.
15207 (lra_constraints): Likewise.
15208 (split_reg): Likewise.
15209 (split_if_necessary): Likewise.
15210 (invariant_p): Likewise.
15211 (inherit_in_ebb): Likewise.
15212 * lra-lives.c (process_bb_lives): Likewise.
15213 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15214 (get_hard_regs): Likewise.
15215 (do_remat): Likewise.
15216 * lra-spills.c (assign_spill_hard_regs): Likewise.
15217 * mode-switching.c (create_pre_exit): Likewise.
15218 * postreload.c (reload_combine_recognize_pattern): Likewise.
15219 * recog.c (peep2_find_free_register): Likewise.
15220 * regcprop.c (kill_value_regno): Likewise.
15221 (set_value_regno): Likewise.
15222 (copy_value): Likewise.
15223 (maybe_mode_change): Likewise.
15224 (find_oldest_value_reg): Likewise.
15225 (copyprop_hardreg_forward_1): Likewise.
15226 * regrename.c (check_new_reg_p): Likewise.
15227 (regrename_do_replace): Likewise.
15228 * reload.c (push_reload): Likewise.
15229 (combine_reloads): Likewise.
15230 (find_dummy_reload): Likewise.
15231 (operands_match_p): Likewise.
15232 (find_reloads): Likewise.
15233 (find_equiv_reg): Likewise.
15234 (reload_adjust_reg_for_mode): Likewise.
15235 * reload1.c (count_pseudo): Likewise.
15236 (count_spilled_pseudo): Likewise.
15237 (find_reg): Likewise.
15238 (clear_reload_reg_in_use): Likewise.
15239 (free_for_value_p): Likewise.
15240 (allocate_reload_reg): Likewise.
15241 (choose_reload_regs): Likewise.
15242 (reload_adjust_reg_for_temp): Likewise.
15243 (emit_reload_insns): Likewise.
15244 (delete_output_reload): Likewise.
15245 * rtlanal.c (subreg_get_info): Likewise.
15246 * sched-deps.c (sched_analyze_reg): Likewise.
15247 * sel-sched.c (init_regs_for_mode): Likewise.
15248 (mark_unavailable_hard_regs): Likewise.
15249 (choose_best_reg_1): Likewise.
15250 (verify_target_availability): Likewise.
15251 * valtrack.c (dead_debug_insert_temp): Likewise.
15252 * var-tracking.c (track_loc_p): Likewise.
15253 (emit_note_insn_var_location): Likewise.
15254 * varasm.c (make_decl_rtl): Likewise.
15255 * reginfo.c (choose_hard_reg_mode): Likewise.
15256 (init_reg_modes_target): Refer directly to
15257 this_target_regs->x_hard_regno_nregs.
15259 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15261 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15262 instead of hard_regno_nregs.
15264 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15266 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15267 end_hard_regno instead of hard_regno_nregs.
15268 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15269 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15270 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15271 * ira-color.c (improve_allocation): Likewise.
15272 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15273 * lra-lives.c (mark_regno_live): Likewise.
15274 (mark_regno_dead): Likewise.
15275 * lra-remat.c (operand_to_remat): Likewise.
15276 * lra.c (collect_non_operand_hard_regs): Likewise.
15277 * postreload.c (reload_combine_note_store): Likewise.
15278 (move2add_valid_value_p): Likewise.
15279 * reload.c (regno_clobbered_p): Likewise.
15281 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15283 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15285 * config/v850/v850.c (v850_reorg): Likewise.
15286 * reload.c (refers_to_regno_for_reload_p): Likewise.
15287 (find_equiv_reg): Likewise.
15288 * reload1.c (reload_reg_reaches_end_p): Likewise.
15290 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15292 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15294 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15295 * config/arm/arm.c (output_move_neon): Likewise.
15296 (arm_attr_length_move_neon): Likewise.
15297 (neon_split_vcombine): Likewise.
15298 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15299 (c6x_mark_reg_written): Likewise.
15300 (c6x_dwarf_register_span): Likewise.
15301 * config/i386/i386.c (ix86_save_reg): Likewise.
15302 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15303 (rws_access_reg): Likewise.
15304 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15305 * mode-switching.c (create_pre_exit): Likewise.
15306 * ree.c (combine_reaching_defs): Likewise.
15307 (add_removable_extension): Likewise.
15308 * regcprop.c (find_oldest_value_reg): Likewise.
15309 (copyprop_hardreg_forward_1): Likewise.
15310 * reload.c (reload_inner_reg_of_subreg): Likewise.
15311 (push_reload): Likewise.
15312 (combine_reloads): Likewise.
15313 (find_dummy_reload): Likewise.
15314 (reload_adjust_reg_for_mode): Likewise.
15315 * reload1.c (find_reload_regs): Likewise.
15316 (forget_old_reloads_1): Likewise.
15317 (reload_reg_free_for_value_p): Likewise.
15318 (reload_adjust_reg_for_temp): Likewise.
15319 (emit_reload_insns): Likewise.
15320 (delete_output_reload): Likewise.
15321 * sel-sched.c (choose_best_reg_1): Likewise.
15322 (choose_best_pseudo_reg): Likewise.
15324 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15325 Alan Hayward <alan.hayward@arm.com>
15326 David Sherwood <david.sherwood@arm.com>
15328 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15329 * target.def (slow_unaligned_access): New hook.
15330 * targhooks.h (default_slow_unaligned_access): Declare.
15331 * targhooks.c (default_slow_unaligned_access): New function.
15332 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15333 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15334 * doc/tm.texi: Regenerate.
15335 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15336 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15337 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15339 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15340 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15342 (rs6000_slow_unaligned_access): New function.
15343 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15344 (expand_block_compare): Likewise.
15345 (expand_strn_compare): Likewise.
15346 (rs6000_rtx_costs): Likewise.
15347 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15348 (riscv_slow_unaligned_access): Likewise.
15349 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15350 (riscv_slow_unaligned_access_p): ...this and make static.
15351 (riscv_option_override): Update accordingly.
15352 (riscv_slow_unaligned_access): New function.
15353 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15354 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15355 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15356 (rs6000_slow_unaligned_access): New function.
15357 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15358 (rs6000_rtx_costs): Likewise.
15359 * config/rs6000/rs6000-string.c (expand_block_compare)
15360 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15361 of SLOW_UNALIGNED_ACCESS.
15362 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15363 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15364 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15365 of SLOW_UNALIGNED_ACCESS.
15366 * expmed.c (simple_mem_bitfield_p): Likewise.
15367 * expr.c (alignment_for_piecewise_move): Likewise.
15368 (emit_group_load_1): Likewise.
15369 (emit_group_store): Likewise.
15370 (copy_blkmode_from_reg): Likewise.
15371 (emit_push_insn): Likewise.
15372 (expand_assignment): Likewise.
15373 (store_field): Likewise.
15374 (expand_expr_real_1): Likewise.
15375 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15376 * lra-constraints.c (simplify_operand_subreg): Likewise.
15377 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15378 * gimple-ssa-store-merging.c: Likewise in block comment at start
15380 * tree-ssa-strlen.c: Include target.h.
15381 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15382 of SLOW_UNALIGNED_ACCESS.
15383 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15385 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15387 PR rtl-optimization/82185
15388 * expmed.c (emit_store_flag_int): Only test tem if it has been
15391 2017-09-12 Richard Biener <rguenther@suse.de>
15393 PR middle-end/82149
15394 * match.pd ((FTYPE) N CMP CST): Fix typo.
15396 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15398 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15400 (mips_near_type_p): Add 'short_call' attribute as a synonym
15402 * doc/extend.texi (short_call): Document new function attribute.
15404 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15407 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15408 assertion check that in the condition.
15409 (get_atomic_generic_size): Likewise. Before testing if parameter
15410 has pointer type, if it has array type, call for C++
15411 default_conversion to perform array-to-pointer conversion.
15413 2017-09-12 Richard Biener <rguenther@suse.de>
15415 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15416 for operations we cannot scalarize.
15418 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15420 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15421 vectors heap vectors. Clean up comments.
15422 Make visited_bbs a reference.
15423 (profitable_jump_thread_path): Make GC
15424 vectors heap vectors. Clean up comments.
15426 (convert_and_register_jump_thread_path): Make GC vectors heap
15428 (check_subpath_and_update_thread_path): Same. Clean up comments.
15429 Make visited_bbs a reference.
15430 (handle_phi): Abstract common code to to
15431 register_jump_thread_path_if_profitable.
15432 Rename VAR_BB to DEF_BB.
15434 Make GC vectors heap vectors.
15435 Make visited_bbs a reference.
15436 (handle_assignment): Same.
15437 (register_jump_thread_path_if_profitable): New.
15438 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15440 Make GC vectors heap vectors. Clean up comments.
15441 Make visited_bbs a reference.
15442 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15443 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15446 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15449 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15450 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15452 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15454 Revert r251800 and r251799.
15456 2017-09-11 Martin Jambor <mjambor@suse.cz>
15459 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15460 arguments in advance.
15461 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15462 use it to find predecessor edges.
15463 (naive_outof_ssa): Collect vector of predecessors.
15465 2017-09-08 Jason Merrill <jason@redhat.com>
15467 PR c++/70029 - ICE with ref-qualifier and -flto
15468 * langhooks.h (struct lang_hooks_for_types): Add
15469 copy_lang_qualifiers.
15470 * attribs.c (build_type_attribute_qual_variant): Use it.
15471 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15473 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15474 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15476 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15479 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15480 (*mulsi3_sp64): New instruction.
15481 (mulsi3): New expander.
15483 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15485 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15487 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15489 * sancov.c: Include memmodel.h.
15491 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15494 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15497 2017-09-07 Carl Love <cel@us.ibm.com>
15499 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15500 the sldi instruction.
15502 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15504 * sancov.c: Include tm_p.h.
15506 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15509 * output.h (switch_to_other_text_partition): New declaration.
15510 * varasm.c (switch_to_other_text_partition): New function.
15511 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15512 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15513 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15514 to the other text partition before emitting LCL label and switch back
15515 after emitting the word after it.
15517 2017-09-07 Richard Biener <rguenther@suse.de>
15519 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15520 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15521 critical edges here, after loop init.
15522 (pass_data_pre): Remove PROP_no_crit_edges flags.
15523 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15524 for valueization of call args to avoid leaking VN_TOP.
15525 (visit_use): Assert we do not visit default defs.
15526 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15527 Use error_mark_node to more easily detect leaking VN_TOP.
15528 All default-defs are varying, not VN_TOP. Mark them visited.
15529 (run_scc_vn): Make code match comment.
15531 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15533 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15534 OPTION_MASK_FLOAT128_KEYWORD.
15535 (POWERPC_MASKS): Likewise.
15536 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15537 support for the -mfloat128-type option, and make -mfloat128
15538 default on PowerPC Linux systems. Define or undefine
15539 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15540 Define __float128 to be __ieee128 if IEEE 128-bit support is
15541 enabled, or undefine it.
15542 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15543 Delete defining __FLOAT128_TYPE__.
15544 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15545 -mfloat128-type option and make -mfloat128 default on PowerPC
15547 (TARGET_FLOAT128_TYPE): Likewise.
15548 (-mfloat128-type): Likewise.
15549 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15550 Delete the -mfloat128-type option and make -mfloat128 default on
15551 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15552 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15553 128-bit floating point is enabled. Change tests from using
15554 -mfloat128-type to -mfloat128.
15555 (rs6000_mangle_type): Use the correct mangling for the __float128
15556 type even if normal long double is restricted to 64-bits.
15557 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15559 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15560 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15561 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15563 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15564 documentation for -mfloat128.
15566 2017-09-06 Olivier Hainque <hainque@adacore.com>
15568 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15570 2017-09-06 Wish Wu <wishwu007@gmail.com>
15571 Jakub Jelinek <jakub@redhat.com>
15573 * asan.c (initialize_sanitizer_builtins): Add
15574 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15575 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15576 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15577 BT_FN_VOID_UINT64_PTR variables.
15578 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15579 (BT_FN_VOID_UINT16_UINT16): Likewise.
15580 (BT_FN_VOID_UINT32_UINT32): Likewise.
15581 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15582 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15583 (BT_FN_VOID_UINT64_PTR): Likewise.
15584 * common.opt (flag_sanitize_coverage): New variable.
15585 (fsanitize-coverage=trace-pc): Remove.
15586 (fsanitize-coverage=): Add.
15587 * flag-types.h (enum sanitize_coverage_code): New enum.
15588 * fold-const.c (fold_range_test): Disable non-short-circuit
15589 optimization if flag_sanitize_coverage.
15590 (fold_truth_andor): Likewise.
15591 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15592 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15593 (coverage_sanitizer_opts): New array.
15594 (get_closest_sanitizer_option): Add OPTS argument, handle also
15595 OPT_fsanitize_coverage_.
15596 (parse_sanitizer_options): Adjusted to also handle
15597 OPT_fsanitize_coverage_.
15598 (common_handle_option): Add OPT_fsanitize_coverage_.
15599 * sancov.c (instrument_comparison, instrument_switch): New function.
15600 (sancov_pass): Add trace-cmp support.
15601 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15602 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15603 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15604 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15605 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15606 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15607 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15608 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15609 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15610 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15612 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15614 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15615 error. Only quit immediately if parsing is complete.
15616 (BEGIN): Initialize fatal_err and parse_done.
15617 (begin fpu, end fpu): Check number of arguments.
15618 (begin arch, end arch): Likewise.
15619 (begin cpu, end cpu): Likewise.
15620 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15621 (optalias): Likewise.
15623 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15625 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15626 * config/arm/arm-isa.h: Delete. Move definitions to ...
15627 * arm-cpus.in: ... here. Use new feature and fgroup values.
15628 * config/arm/arm.c (arm_option_override): Use lower case for feature
15630 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15631 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15632 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15634 (print_isa_bits_for): New function.
15635 (gen_isa): New function.
15636 (gen_comm_data): Use print_isa_bits_for.
15637 (define feature): New keyword.
15638 (define fgroup): New keyword.
15639 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15640 (arm-isa.h): Add rule to generate file.
15641 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15642 case for feature bit names.
15644 2017-09-06 Richard Biener <rguenther@suse.de>
15646 * tree-ssa-pre.c (NECESSARY): Remove.
15647 (create_expression_by_pieces): Do not touch pass-local flags.
15648 (insert_into_preds_of_block): Likewise.
15649 (do_pre_regular_insertion): Likewise.
15650 (eliminate_insert): Likewise.
15651 (eliminate_dom_walker::before_dom_children): Likewise.
15652 (fini_eliminate): Do not look at inserted_exprs.
15653 (mark_operand_necessary): Remove.
15654 (remove_dead_inserted_code): Replace with simple work-list
15655 algorithm based on inserted_exprs and SSA uses.
15656 (pass_pre::execute): Re-order fini_eliminate and
15657 remove_dead_inserted_code.
15659 2017-09-06 Olivier Hainque <hainque@adacore.com>
15661 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15662 for VxWorks 7. Adjust surrounding comments.
15664 2017-09-06 Richard Biener <rguenther@suse.de>
15666 * gimple-ssa-strength-reduction.c
15667 (find_candidates_dom_walker::before_dom_children): Also allow
15670 2017-09-06 Richard Biener <rguenther@suse.de>
15672 PR tree-optimization/82108
15673 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15674 for gap in the non-permutation SLP case.
15676 2017-09-06 Martin Jambor <mjambor@suse.cz>
15678 PR tree-optimization/82078
15679 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15680 add_access_to_work_queue...
15681 (build_accesses_from_assign): ...here.
15682 (propagate_all_subaccesses): Make sure racc is the group
15683 representative, if there is one.
15685 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15687 PR middle-end/82095
15688 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15691 2017-09-06 Richard Biener <rguenther@suse.de>
15693 * gimple-ssa-strength-reduction.c
15694 (find_candidates_dom_walker::before_doom_children): Use a
15695 type and not a mode check.
15697 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15700 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15702 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15704 2017-09-05 Jeff Law <law@redhat.com>
15706 PR tree-optimization/64910
15707 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15708 cases where we have 3 or more operands.
15710 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15712 PR middle-end/81768
15713 * omp-low.c (lower_omp_for): Recompute tree invariant if
15714 gimple_omp_for_initial/final is ADDR_EXPR.
15716 PR middle-end/81768
15717 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15718 into gimple val before gimplification fo the COND_EXPR.
15720 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15722 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15723 REGION_COPY argument.
15724 (thread_through_all_blocks): Remove unused argument to
15725 duplicate_thread_path.
15727 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15728 Alan Hayward <alan.hayward@arm.com>
15729 David Sherwood <david.sherwood@arm.com>
15731 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15732 Take a scalar_mode rather than a machine_mode.
15733 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15734 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15735 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15736 (aarch64_gen_adjusted_ldpstp): Likewise.
15737 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15739 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15740 Alan Hayward <alan.hayward@arm.com>
15741 David Sherwood <david.sherwood@arm.com>
15743 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15744 Take a scalar_int_mode instead of a machine_mode.
15745 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15746 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15747 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15748 (aarch64_simd_attr_length_rglist): Delete.
15749 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15750 a scalar_int_mode instead of a machine_mode.
15751 (aarch64_add_offset): Likewise.
15752 (aarch64_internal_mov_immediate): Likewise
15753 (aarch64_add_constant_internal): Likewise.
15754 (aarch64_add_constant): Likewise.
15755 (aarch64_movw_imm): Likewise.
15756 (aarch64_rtx_arith_op_extract_p): Likewise.
15757 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15758 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15759 Remove assert that the mode isn't a vector.
15760 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15761 (aarch64_expand_mov_immediate): Update calls after above changes.
15762 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15763 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15764 (aarch64_move_imm): Likewise.
15765 (aarch64_can_const_movi_rtx_p): Likewise.
15766 (aarch64_strip_extend): Likewise.
15767 (aarch64_extr_rtx_p): Likewise.
15768 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15769 a CONST_INT when the mode parameter is VOIDmode.
15770 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15772 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15774 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15775 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15776 (bitwise_type_for_mode): Update accordingly.
15778 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15780 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15781 * stor-layout.c (mode_for_size_tree): Likewise.
15782 (mode_for_array): Update accordingly.
15783 (layout_decl): Likewise.
15784 (compute_record_mode): Likewise. Only set the mode once.
15786 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15788 * target.def (get_mask_mode): Change return type to opt_mode.
15790 * doc/tm.texi: Regenerate.
15791 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15792 * targhooks.c (default_get_mask_mode): Likewise.
15793 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15794 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15795 targetm.get_mask_mode.
15796 * tree.c (build_truth_vector_type): Likewise.
15798 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15800 * machmode.h (mode_for_vector): Return an opt_mode.
15801 * stor-layout.c (mode_for_vector): Likewise.
15802 (mode_for_int_vector): Update accordingly.
15803 (layout_type): Likewise.
15804 * config/i386/i386.c (emit_memmov): Likewise.
15805 (ix86_expand_set_or_movmem): Likewise.
15806 (ix86_expand_vector_init): Likewise.
15807 (ix86_get_mask_mode): Likewise.
15808 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15810 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15811 * expmed.c (extract_bit_field_1): Likewise.
15812 * expr.c (expand_expr_real_2): Likewise.
15813 * optabs-query.c (can_vec_perm_p): Likewise.
15814 (can_vec_mask_load_store_p): Likewise.
15815 * optabs.c (expand_vec_perm): Likewise.
15816 * targhooks.c (default_get_mask_mode): Likewise.
15817 * tree-vect-stmts.c (vectorizable_store): Likewise.
15818 (vectorizable_load): Likewise.
15819 (get_vectype_for_scalar_type_and_size): Likewise.
15821 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15823 * machmode.h (mode_for_int_vector): New function.
15824 * stor-layout.c (mode_for_int_vector): Likewise.
15825 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15826 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15827 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15828 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15829 (s390_expand_vcond): Likewise.
15831 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15833 * machmode.h (opt_machine_mode): New type.
15834 (opt_mode<T>): Allow construction from anything that can be
15836 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15837 (mode_for_size): Return an opt_machine_mode.
15838 * stor-layout.c (mode_for_size): Likewise.
15839 (mode_for_size_tree): Update call accordingly.
15840 (bitwise_mode_for_mode): Likewise.
15841 (make_fract_type): Likewise.
15842 (make_accum_type): Likewise.
15843 * caller-save.c (replace_reg_with_saved_mem): Update call
15845 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15846 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15847 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15848 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15849 * expmed.c (extract_bit_field_1): Likewise.
15850 * reload.c (get_secondary_mem): Likewise.
15851 * varasm.c (assemble_integer): Likewise.
15852 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15855 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15857 * machmode.h (decimal_float_mode_for_size): New function.
15858 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
15859 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
15860 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
15861 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
15862 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
15863 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
15865 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15867 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
15868 (get_builtin_sync_mode): Likewise.
15869 (expand_ifn_atomic_compare_exchange): Likewise.
15870 (expand_builtin_atomic_clear): Likewise.
15871 (expand_builtin_atomic_test_and_set): Likewise.
15872 (fold_builtin_atomic_always_lock_free): Likewise.
15873 * calls.c (compute_argument_addresses): Likewise.
15874 (emit_library_call_value_1): Likewise.
15875 (store_one_arg): Likewise.
15876 * combine.c (combine_instructions): Likewise.
15877 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
15878 * config/arm/arm.c (arm_function_value): Likewise.
15879 (aapcs_allocate_return_reg): Likewise.
15880 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
15881 * config/i386/i386.c (construct_container): Likewise.
15882 (ix86_gimplify_va_arg): Likewise.
15883 (ix86_expand_sse_cmp): Likewise.
15884 (emit_memmov): Likewise.
15885 (emit_memset): Likewise.
15886 (expand_small_movmem_or_setmem): Likewise.
15887 (ix86_expand_pextr): Likewise.
15888 (ix86_expand_pinsr): Likewise.
15889 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
15890 * config/microblaze/microblaze.c (microblaze_block_move_straight):
15892 * config/mips/mips.c (mips_function_value_1) Likewise.
15893 (mips_block_move_straight): Likewise.
15894 (mips_expand_ins_as_unaligned_store): Likewise.
15895 * config/powerpcspe/powerpcspe.c
15896 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15897 (rs6000_darwin64_record_arg_flush): Likewise.
15898 * config/rs6000/rs6000.c
15899 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15900 (rs6000_darwin64_record_arg_flush): Likewise.
15901 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
15902 (sparc_function_value_1): Likewise.
15903 * config/spu/spu.c (adjust_operand): Likewise.
15904 (spu_emit_branch_or_set): Likewise.
15905 (arith_immediate_p): Likewise.
15906 * emit-rtl.c (gen_lowpart_common): Likewise.
15907 * expr.c (expand_expr_real_1): Likewise.
15908 * function.c (assign_parm_setup_block): Likewise.
15909 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
15910 * reload1.c (alter_reg): Likewise.
15911 * stor-layout.c (mode_for_vector): Likewise.
15912 (layout_type): Likewise.
15914 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15916 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
15917 (spu_convert_move): Likewise.
15918 * lower-subreg.c (resolve_simple_move): Likewise.
15920 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15923 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
15924 define_insn to a define_expand.
15925 (altivec_vsum2sws_direct): New define_insn.
15926 (altivec_vsumsws): Convert from a define_insn to a define_expand.
15928 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15930 * config/arm/arm.c (arm_option_params_internal): Improve setting of
15933 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
15938 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
15939 realign stack if stack alignment needed is less than incoming
15942 2017-09-05 Marek Polacek <polacek@redhat.com>
15945 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
15948 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15950 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
15952 2017-09-05 Richard Biener <rguenther@suse.de>
15954 PR tree-optimization/82084
15955 * fold-const.c (can_native_encode_string_p): Handle wide characters.
15957 2017-09-05 Richard Biener <rguenther@suse.de>
15959 PR tree-optimization/82102
15960 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
15962 2017-09-05 Martin Liska <mliska@suse.cz>
15964 PR tree-optimization/82032
15965 * tree-cfg.c (generate_range_test): New function.
15966 * tree-cfg.h (generate_range_test): Declared here.
15967 * tree-cfgcleanup.c (convert_single_case_switch): New function.
15968 (cleanup_control_expr_graph): Use it.
15969 * tree-switch-conversion.c (try_switch_expansion): Remove
15971 (emit_case_nodes): Use generate_range_test.
15973 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15976 * config/i386/i386.md (*<btsc><mode>_mask): Add
15977 TARGET_USE_BT to insn constraint.
15978 (*btr<mode>_mask): Ditto.
15980 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
15982 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
15983 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
15985 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
15988 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
15989 TARGET_NEON and TARGET_IWMMXT.
15990 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
15991 TARGET_NEON and TARGET_IWMMXT.
15992 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
15994 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15996 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
15997 (ix86_rewrite_tls_address): Ditto.
15998 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
15999 (ix86_rewrite_tls_address_1): Ditto.
16000 (ix86_rewrite_tls_address): Ditto.
16001 * config/i386/predicates.md (tls_address_pattern): New predicate.
16002 * config/i386/i386.md (TLS address splitter): New splitter.
16004 2017-09-04 Richard Biener <rguenther@suse.de>
16006 PR tree-optimization/82084
16007 * fold-const.h (can_native_encode_string_p): Declare.
16008 * fold-const.c (can_native_encode_string_p): Factor out from ...
16009 (native_encode_string): ... here.
16010 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16011 vectorizing stores from constants we later cannot handle.
16013 2017-09-04 Marek Polacek <polacek@redhat.com>
16016 * doc/invoke.texi: Update -Wtautological-compare documentation.
16018 2017-09-04 Jeff Law <law@redhat.com>
16020 PR tree-optimization/64910
16021 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16022 swap the first and last operand if the last is a constant.
16024 2017-09-04 Marek Polacek <polacek@redhat.com>
16027 * convert.c (do_narrow): When sanitizing signed integer overflows,
16028 bail out for signed types.
16029 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16031 2017-09-04 Richard Biener <rguenther@suse.de>
16033 PR tree-optimization/82060
16034 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16035 Move devirtualization after stmt folding and before EH/AB/noreturn
16036 cleanup to get the stmt refs canonicalized. Use a bool instead
16037 of gimple_modified_p since that doesn't work for NOPs. Schedule
16038 NOPs generated by folding for removal.
16040 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16041 Alan Hayward <alan.hayward@arm.com>
16042 David Sherwood <david.sherwood@arm.com>
16044 * coretypes.h (pad_direction): New enum.
16045 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16046 (FUNCTION_ARG_PADDING): Likewise.
16047 * target.def (function_arg_padding): New hook.
16048 * targhooks.h (default_function_arg_padding): Declare.
16049 * targhooks.c (default_function_arg_padding): New function.
16050 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16051 (TARGET_FUNCTION_ARG_PADDING): ...this.
16052 * doc/tm.texi: Regenerate.
16053 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16054 instead of direction.
16055 (compute_argument_addresses): Likewise.
16056 (load_register_parameters): Likewise.
16057 (emit_library_call_value_1): Likewise.
16058 (store_one_arg): Use targetm.calls.function_arg_padding instead
16059 of FUNCTION_ARG_PADDING.
16060 (must_pass_in_stack_var_size_or_pad): Likewise.
16061 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16062 (emit_group_store): Likewise.
16063 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16064 instead of FUNCTION_ARG_PADDING.
16065 (emit_push_insn): Likewise, and propagate enum change throughout
16067 * function.h (direction): Delete.
16068 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16070 * function.c (assign_parm_find_stack_rtl): Likewise.
16071 (assign_parm_setup_block_p): Likewise.
16072 (assign_parm_setup_block): Likewise.
16073 (gimplify_parameters): Likewise.
16074 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16075 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16077 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16078 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16079 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16080 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16081 (aarch64_function_arg_padding): ...this new function.
16082 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16083 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16084 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16085 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16086 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16087 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16088 (arm_pad_arg_upward): Replace with...
16089 (arm_function_arg_padding): ...this new function.
16090 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16092 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16093 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16094 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16095 (ia64_hpux_function_arg_padding): Replace with...
16096 (ia64_function_arg_padding): ...this new function. Use pad_direction
16097 instead of direction. Check for TARGET_HPUX.
16098 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16099 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16100 (iq2000_function_arg_padding): New function.
16101 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16102 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16103 (mips_function_arg_padding): ...this new function.
16104 (mips_pad_reg_upward): Update accordingly.
16105 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16106 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16107 targetm.calls.function_arg_padding.
16108 (FUNCTION_ARG_PADDING): Delete.
16109 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16110 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16111 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16112 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16113 (nios2_block_reg_padding): Return pad_direction instead of direction.
16114 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16115 instead of direction.
16116 (nios2_function_arg_padding): Likewise. Make static.
16117 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16118 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16119 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16120 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16121 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16122 (pa_function_arg_padding): Make static. Return pad_direction instead
16124 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16125 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16126 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16127 instead of direction. Use targetm.calls.function_arg_padding.
16128 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16129 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16130 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16131 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16132 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16134 (function_arg_padding): Rename to...
16135 (rs6000_function_arg_padding): ...this. Make static. Return
16136 pad_direction instead of direction.
16137 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16138 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16139 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16140 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16141 instead of direction. Use targetm.calls.function_arg_padding.
16142 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16143 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16144 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16145 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16146 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16147 (function_arg_padding): Rename to...
16148 (rs6000_function_arg_padding): ...this. Make static. Return
16149 pad_direction instead of direction.
16150 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16151 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16152 * config/s390/s390.c (s390_function_arg_padding): New function.
16153 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16154 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16155 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16156 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16157 (function_arg_padding): Rename to...
16158 (sparc_function_arg_padding): ...this. Make static. Return
16159 pad_direction instead of direction.
16160 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16161 * config/spu/spu.c (spu_function_arg_padding): New function.
16162 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16163 * system.h (FUNCTION_ARG_PADDING): Poison.
16165 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16166 Alan Hayward <alan.hayward@arm.com>
16167 David Sherwood <david.sherwood@arm.com>
16169 * target.def (modes_tieable_p): New hook.
16170 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16171 (TARGET_MODES_TIEABLE_P): ...this.
16172 * doc/tm.texi.in: Regenerate.
16173 * hooks.h (hook_bool_mode_mode_true): Declare.
16174 * hooks.c (hook_bool_mode_mode_true): New function.
16175 * combine.c (subst): Use targetm.modes_tieable_p instead of
16177 * dse.c (find_shift_sequence): Likewise.
16178 * expmed.c (extract_low_bits): Likewise.
16179 * lower-subreg.c: Include target.h.
16180 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16182 * rtlanal.c (rtx_cost): Likewise.
16183 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16184 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16185 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16186 (TARGET_MODES_TIEABLE_P): Redefine.
16187 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16188 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16189 (TARGET_MODES_TIEABLE_P): Redefine.
16190 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16191 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16192 (arc_modes_tieable_p): New function.
16193 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16194 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16195 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16196 (arm_modes_tieable_p): Make static.
16197 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16198 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16199 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16200 (TARGET_MODES_TIEABLE_P): Redefine.
16201 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16202 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16203 (TARGET_MODES_TIEABLE_P): Redefine.
16204 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16205 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16206 (cr16_modes_tieable_p): New function.
16207 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16208 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16209 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16210 (TRULY_NOOP_TRUNCATION): Update comment.
16211 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16212 (TRULY_NOOP_TRUNCATION): Update comment.
16213 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16214 (frv_modes_tieable_p): New function.
16215 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16216 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16217 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16218 (TARGET_MODES_TIEABLE_P): Redefine.
16219 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16220 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16221 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16222 (TARGET_MODES_TIEABLE_P): Redefine.
16223 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16224 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16225 (ia64_modes_tieable_p): New function.
16226 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16227 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16228 (iq2000_modes_tieable_p): New function.
16229 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16230 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16231 (lm32_modes_tieable_p): New function.
16232 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16233 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16234 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16235 (TARGET_MODES_TIEABLE_P): Redefine.
16236 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16237 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16238 (m32r_modes_tieable_p): New function.
16239 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16240 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16241 (m68k_modes_tieable_p): New function.
16242 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16243 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16244 (mcore_modes_tieable_p): New function.
16245 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16246 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16248 (TARGET_MODES_TIEABLE_P): Redefine.
16249 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16250 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16251 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16252 (TARGET_MODES_TIEABLE_P): Redefine.
16253 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16254 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16255 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16256 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16257 (mn10300_modes_tieable_p): ...this and make static.
16258 (TARGET_MODES_TIEABLE_P): Redefine.
16259 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16260 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16261 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16262 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16263 (msp430_modes_tieable_p): Make static.
16264 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16265 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16266 (TARGET_MODES_TIEABLE_P): Redefine.
16267 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16268 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16269 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16270 (TARGET_MODES_TIEABLE_P): Redefine.
16271 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16272 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16273 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16274 (TARGET_MODES_TIEABLE_P): Redefine.
16275 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16276 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16277 (pdp11_modes_tieable_p): New function.
16278 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16279 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16280 (rs6000_modes_tieable_p): New function.
16281 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16282 * config/powerpcspe/powerpcspe.md: Update comment.
16283 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16284 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16285 (TARGET_MODES_TIEABLE_P): Redefine.
16286 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16287 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16288 (rl78_modes_tieable_p): New function.
16289 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16290 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16291 (rs6000_modes_tieable_p): New function.
16292 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16293 * config/rs6000/rs6000.md: Update comment.
16294 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16295 * config/rx/rx.c (rx_modes_tieable_p): New function.
16296 (TARGET_MODES_TIEABLE_P): Redefine.
16297 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16298 * config/s390/s390.c (s390_modes_tieable_p): New function.
16299 (TARGET_MODES_TIEABLE_P): Redefine.
16300 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16301 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16302 (sh_modes_tieable_p): New function.
16303 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16304 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16305 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16306 (sparc_modes_tieable_p): Make static.
16307 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16308 * config/spu/spu.c (spu_modes_tieable_p): New function.
16309 (TARGET_MODES_TIEABLE_P): Redefine.
16310 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16311 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16312 (TARGET_MODES_TIEABLE_P): Redefine.
16313 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16314 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16315 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16316 * config/v850/v850.c (v850_modes_tieable_p): New function.
16317 (TARGET_MODES_TIEABLE_P): Redefine.
16318 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16319 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16320 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16321 (visium_modes_tieable_p): New function.
16322 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16323 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16324 (xtensa_modes_tieable_p): New function.
16325 * system.h (MODES_TIEABLE_P): Poison.
16327 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16328 Alan Hayward <alan.hayward@arm.com>
16329 David Sherwood <david.sherwood@arm.com>
16331 * target.def (hard_regno_mode_ok): New hook.
16332 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16333 (TARGET_HARD_REGNO_MODE_OK): ...this.
16334 * doc/tm.texi.in: Regenerate.
16335 * hooks.h (hook_bool_uint_mode_true): Declare.
16336 * hooks.c (hook_bool_uint_mode_true): New function.
16337 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16338 HARD_REGNO_MODE_OK.
16339 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16340 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16341 instead of HARD_REGNO_MODE_OK.
16342 * caller-save.c: Include target.h.
16343 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16344 HARD_REGNO_MODE_OK.
16345 * combine.c (can_combine_p): Likewise.
16346 (combinable_i3pat): Likewise.
16347 (can_change_dest_mode): Likewise.
16348 * expr.c (init_expr_target): Likewise.
16349 (convert_move): Likewise.
16350 (convert_modes): Likewise.
16351 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16352 (setup_prohibited_mode_move_regs): Likewise.
16353 * ira.h (target_ira): Likewise.
16354 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16355 * lra-constraints.c (process_alt_operands): Likewise.
16356 (split_reg): Likewise.
16357 * recog.c (peep2_find_free_register): Likewise.
16358 * ree.c (combine_reaching_defs): Likewise.
16359 * regcprop.c (maybe_mode_change): Likewise.
16360 * reginfo.c (init_reg_sets_1): Likewise.
16361 (choose_hard_reg_mode): Likewise.
16362 (simplifiable_subregs): Likewise.
16363 * regrename.c (check_new_reg_p): Likewise.
16364 * reload.c (find_valid_class): Likewise.
16365 (find_valid_class_1): Likewise.
16366 (reload_inner_reg_of_subreg): Likewise.
16367 (push_reload): Likewise.
16368 (combine_reloads): Likewise.
16369 (find_dummy_reload): Likewise.
16370 (find_reloads): Likewise.
16371 * reload1.c (find_reg): Likewise.
16372 (set_reload_reg): Likewise.
16373 (allocate_reload_reg): Likewise.
16374 (choose_reload_regs): Likewise.
16375 (reload_adjust_reg_for_temp): Likewise.
16376 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16377 (simplify_subreg_regno): Likewise.
16378 * sel-sched.c (init_regs_for_mode): Likewise.
16379 * varasm.c (make_decl_rtl): Likewise.
16380 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16381 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16382 HARD_REGNO_MODE_OK.
16383 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16384 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16385 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16386 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16387 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16388 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16389 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16390 (arc_mode_class): Delete.
16391 (HARD_REGNO_MODE_OK): Delete.
16392 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16393 (arc_hard_regno_mode_ok): Rename old array to...
16394 (arc_hard_regno_mode_ok_modes): ...this.
16395 (arc_conditional_register_usage): Update accordingly.
16396 (arc_mode_class): Make static.
16397 (arc_hard_regno_mode_ok): New function.
16398 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16399 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16400 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16401 (arm_hard_regno_mode_ok): Make static.
16402 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16403 HARD_REGNO_MODE_OK.
16404 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16405 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16406 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16408 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16409 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16410 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16411 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16412 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16413 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16414 * config/bfin/predicates.md (valid_reg_operand): Use
16415 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16416 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16417 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16418 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16419 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16420 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16421 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16422 (cr16_hard_regno_mode_ok): Make static and return a bool.
16423 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16424 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16425 (cris_hard_regno_mode_ok): New function.
16426 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16427 (epiphany_mode_class): Delete.
16428 (HARD_REGNO_MODE_OK): Delete.
16429 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16430 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16431 (hard_regno_mode_ok): Rename to...
16432 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16433 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16434 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16435 HARD_REGNO_MODE_OK.
16436 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16437 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16438 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16439 (frv_hard_regno_mode_ok): Make static and return a bool.
16440 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16441 HARD_REGNO_MODE_OK.
16442 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16443 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16444 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16445 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16447 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16448 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16449 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16450 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16452 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16453 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16454 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16455 (ia64_hard_regno_mode_ok): New function.
16456 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16457 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16458 (iq2000_hard_regno_mode_ok): New function.
16459 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16460 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16461 (lm32_hard_regno_mode_ok): New function.
16462 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16463 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16464 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16465 instead of HARD_REGNO_MODE_OK.
16466 (m32c_hard_regno_ok): Rename to...
16467 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16468 (m32c_cannot_change_mode_class): Update accordingly.
16469 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16470 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16471 (m32r_mode_class): Delete.
16472 (HARD_REGNO_MODE_OK): Delete.
16473 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16474 (m32r_hard_regno_mode_ok): Rename to...
16475 (m32r_hard_regno_modes): ...this.
16476 (m32r_mode_class): Make static.
16477 (m32r_hard_regno_mode_ok): New function.
16478 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16479 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16480 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16481 (m68k_hard_regno_mode_ok): Make static.
16482 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16483 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16484 (mcore_hard_regno_mode_ok): New function.
16485 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16486 (HARD_REGNO_MODE_OK): Delete.
16487 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16489 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16490 (microblaze_hard_regno_mode_ok): New function.
16491 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16492 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16493 (mips_hard_regno_mode_ok): Delete.
16494 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16495 (mips_hard_regno_mode_ok_p): ...this and make static.
16496 (mips_hard_regno_mode_ok_p): Rename to...
16497 (mips_hard_regno_mode_ok_uncached): ...this.
16498 (mips_hard_regno_mode_ok): New function.
16499 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16500 of HARD_REGNO_MODE_OK.
16501 (mips_option_override): Update after above name changes.
16502 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16503 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16504 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16505 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16506 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16507 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16508 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16509 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16510 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16511 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16512 (msp430_hard_regno_mode_ok): Make static and return a bool.
16513 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16514 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16515 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16517 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16518 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16519 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16520 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16521 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16522 (PA_HARD_REGNO_MODE_OK): ...this
16523 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16524 (PA_HARD_REGNO_MODE_OK): ...this.
16525 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16526 (pa_hard_regno_mode_ok): New function.
16527 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16528 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16529 (pdp11_hard_regno_mode_ok): New function.
16530 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16531 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16533 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16535 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16536 (rs6000_hard_regno_mode_ok): Rename to...
16537 (rs6000_hard_regno_mode_ok_uncached): ...this.
16538 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16539 (rs6000_hard_regno_mode_ok): New function.
16540 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16541 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16542 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16543 (riscv_hard_regno_mode_ok): ...this and make static.
16544 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16545 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16546 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16547 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16548 (rl78_hard_regno_mode_ok): Make static and return bool.
16549 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16550 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16552 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16553 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16554 (rs6000_hard_regno_mode_ok): Rename to...
16555 (rs6000_hard_regno_mode_ok_uncached): ...this.
16556 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16557 (rs6000_hard_regno_mode_ok): New function.
16558 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16559 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16560 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16561 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16562 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16563 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16564 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16565 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16566 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16567 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16568 (sh_hard_regno_mode_ok): Make static.
16569 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16570 instead of HARD_REGNO_MODE_OK.
16571 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16572 (sparc_mode_class): Delete.
16573 (HARD_REGNO_MODE_OK): Delete.
16574 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16575 (hard_regno_mode_classes): Make static.
16576 (sparc_mode_class): Likewise.
16577 (sparc_hard_regno_mode_ok): New function.
16578 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16579 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16580 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16582 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16583 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16584 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16585 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16586 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16587 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16588 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16589 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16590 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16591 (visium_hard_regno_mode_ok): New function.
16592 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16593 instead of HARD_REGNO_MODE_OK.
16594 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16595 (HARD_REGNO_MODE_OK): Delete.
16596 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16597 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16598 (xtensa_option_override): Update accordingly.
16599 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16600 (xtensa_hard_regno_mode_ok): New function.
16601 * system.h (HARD_REGNO_MODE_OK): Poison.
16603 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16604 Alan Hayward <alan.hayward@arm.com>
16605 David Sherwood <david.sherwood@arm.com>
16607 * target.def (hard_regno_call_part_clobbered): New hook.
16608 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16609 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16610 * doc/tm.texi: Regenerate.
16611 * hooks.h (hook_bool_uint_mode_false): Declare.
16612 * hooks.c (hook_bool_uint_mode_false): New function.
16613 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16614 * cselib.c (cselib_process_insn): Use
16615 targetm.hard_regno_call_part_clobbered instead of
16616 HARD_REGNO_CALL_PART_CLOBBERED.
16617 * ira-conflicts.c (ira_build_conflicts): Likewise.
16618 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16619 * lra-constraints.c (need_for_call_save_p): Likewise.
16620 * lra-lives.c: Include target.h.
16621 (check_pseudos_live_through_calls): Use
16622 targetm.hard_regno_call_part_clobbered instead of
16623 HARD_REGNO_CALL_PART_CLOBBERED.
16624 * regcprop.c: Include target.h.
16625 (copyprop_hardreg_forward_1): Use
16626 targetm.hard_regno_call_part_clobbered instead of
16627 HARD_REGNO_CALL_PART_CLOBBERED.
16628 * reginfo.c (choose_hard_reg_mode): Likewise.
16629 * regrename.c (check_new_reg_p): Likewise.
16630 * reload.c (find_equiv_reg): Likewise.
16631 * reload1.c (emit_reload_insns): Likewise.
16632 * sched-deps.c (deps_analyze_insn): Likewise.
16633 * sel-sched.c (init_regs_for_mode): Likewise.
16634 (mark_unavailable_hard_regs): Likewise.
16635 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16636 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16637 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16639 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16640 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16641 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16643 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16645 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16646 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16647 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16649 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16650 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16651 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16653 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16654 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16656 * config/powerpcspe/powerpcspe.c
16657 (rs6000_hard_regno_call_part_clobbered): New function.
16658 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16659 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16660 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16662 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16663 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16664 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16666 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16667 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16668 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16670 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16671 Alan Hayward <alan.hayward@arm.com>
16672 David Sherwood <david.sherwood@arm.com>
16674 * rtl.h (subreg_memory_offset): Declare.
16675 * emit-rtl.c (subreg_memory_offset): New function.
16676 * expmed.c (store_bit_field_1): Use it.
16677 * expr.c (undefined_operand_subword_p): Likewise.
16678 * simplify-rtx.c (simplify_subreg): Likewise.
16680 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16682 PR rtl-optimization/57448
16685 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16686 using atomic_load pattern.
16687 (expand_atomic_store): Likewise.
16689 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16692 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16693 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16694 handling. Use replace_call_with_value with NULL instead of
16695 gsi_replace, unlink_stmt_vdef and release_defs.
16697 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16700 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16702 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16705 * rtl.h (emit_library_call_value_1): Declare.
16706 (emit_library_call): Replace declaration with a series of overloads.
16707 Remove the parameter count argument.
16708 (emit_library_call_value): Likewise.
16709 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16710 with an "rtx_mode_t *".
16711 (emit_library_call_value): Delete.
16712 (emit_library_call): Likewise.
16713 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16714 (asan_emit_allocas_unpoison): Likewise.
16715 * builtins.c (expand_builtin_powi): Likewise.
16716 (expand_asan_emit_allocas_unpoison): Likewise.
16717 * cfgexpand.c (expand_main_function): Likewise.
16718 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16719 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16720 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16721 * config/arm/arm.c (arm_trampoline_init): Likewise.
16722 (arm_call_tls_get_addr): Likewise.
16723 (arm_expand_divmod_libfunc): Likewise.
16724 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16725 (smulsi3_highpart): Likewise.
16726 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16727 (c6x_expand_compare): Likewise.
16728 (c6x_expand_movmem): Likewise.
16729 * config/frv/frv.c (frv_trampoline_init): Likewise.
16730 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16731 (ix86_expand_divmod_libfunc): Likewise.
16732 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16733 (ia64_expand_compare): Likewise.
16734 (ia64_profile_hook): Likewise.
16735 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16736 (nonlocal_goto): Likewise.
16737 (restore_stack_nonlocal): Likewise.
16738 * config/m32r/m32r.c (block_move_call): Likewise.
16739 (m32r_trampoline_init): Likewise.
16740 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16741 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16742 (m68k_call_m68k_read_tp): Likewise.
16743 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16744 (microblaze_expand_divide): Likewise.
16745 * config/mips/mips.h (mips_args): Likewise.
16746 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16747 (MIPS_ICACHE_SYNC): Likewise.
16748 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16749 (nios2_trampoline_init): Likewise.
16750 * config/pa/pa.c (hppa_tls_call): Likewise.
16751 (pa_trampoline_init): Likewise.
16752 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16753 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16754 (expand_strn_compare): Likewise.
16755 (rs6000_generate_compare): Likewise.
16756 (rs6000_expand_float128_convert): Likewise.
16757 (output_profile_hook): Likewise.
16758 (rs6000_trampoline_init): Likewise.
16759 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16760 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16761 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16762 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16763 (rs6000_generate_compare): Likewise.
16764 (rs6000_expand_float128_convert): Likewise.
16765 (output_profile_hook): Likewise.
16766 (rs6000_trampoline_init): Likewise.
16767 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16768 * config/sh/sh.c (sh_trampoline_init): Likewise.
16769 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16770 (sparc_emit_float_lib_cmp): Likewise.
16771 (sparc32_initialize_trampoline): Likewise.
16772 (sparc64_initialize_trampoline): Likewise.
16773 (sparc_profile_hook): Likewise.
16774 * config/spu/spu.c (ea_load_store): Likewise.
16775 * config/spu/spu.md (floatunssidf2): Likewise.
16776 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16777 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16778 * config/visium/visium.c (expand_block_move_4): Likewise.
16779 (expand_block_move_2): Likewise.
16780 (expand_block_move_1): Likewise.
16781 (expand_block_set_4): Likewise.
16782 (expand_block_set_2): Likewise.
16783 (expand_block_set_1): Likewise.
16784 (visium_trampoline_init): Likewise.
16785 (visium_profile_hook): Likewise.
16786 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16787 (xtensa_setup_frame_addresses): Likewise.
16788 (xtensa_trampoline_init): Likewise.
16789 * except.c (sjlj_emit_function_enter): Likewise.
16790 (sjlj_emit_function_exit): Likewise.
16791 * explow.c (allocate_dynamic_stack_space): Likewise.
16792 (probe_stack_range): Likewise.
16793 * expr.c (convert_mode_scalar): Likewise.
16794 * optabs.c (expand_binop): Likewise.
16795 (expand_twoval_binop_libfunc): Likewise.
16796 (expand_unop): Likewise.
16797 (prepare_cmp_insn): Likewise.
16798 (prepare_float_lib_cmp): Likewise.
16799 (expand_float): Likewise.
16800 (expand_fix): Likewise.
16801 (expand_fixed_convert): Likewise.
16802 (maybe_emit_sync_lock_test_and_set): Likewise.
16803 (expand_atomic_compare_and_swap): Likewise.
16804 (expand_mem_thread_fence): Likewise.
16805 (expand_atomic_fetch_op): Likewise.
16807 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16809 * doc/generic.texi (OpenACC): Adjust URL.
16810 * doc/invoke.texi (C Dialect Options): Ditto.
16812 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16814 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16815 predicate for operand 1. Add (m,<S>) constraint.
16816 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16817 Prevent memory operand 1 with register operand 2.
16819 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16821 PR rtl-optimization/82024
16822 * combine.c (try_combine): If the combination result is a PARALLEL,
16823 and we only need to retain the SET in there that would be placed
16824 at I2, check that we can place that at I3 instead, before doing so.
16826 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16829 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16831 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16832 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16835 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16836 Jeff Law <law@redhat.com>
16838 * varasm.c (bss_initializer_p): Do not put constants into .bss
16839 (categorize_decl_for_section): Handle bss_initializer_p returning
16840 false when DECL_INITIAL is NULL.
16842 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16845 * config/s390/s390.c (s390_can_inline_p): New function.
16847 2017-09-01 Jeff Law <law@redhat.com>
16849 PR tree-optimization/82052
16850 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16851 Always initialize the returned slot after a hash table miss
16852 when INSERT is true.
16854 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16856 * config/s390/s390.md (mem_signal_fence): Remove.
16857 * doc/md.texi (mem_signal_fence): Remove.
16858 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
16860 * target-insns.def (mem_signal_fence): Remove.
16862 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16865 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
16868 * asan.c (create_odr_indicator): Strip name encoding from assembler
16869 name before appending it after __odr_asan_.
16871 2017-09-01 Martin Liska <mliska@suse.cz>
16873 PR tree-optimization/82059
16874 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
16875 frequency only when an edge is redirected.
16877 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16879 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
16880 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
16881 (arc_conditional_register_usage): Remove ARC600 lp_count
16883 (arc_file_start): Emit Tag_ARC_CPU_variation.
16884 (arc_can_use_doloop_p): New conditions to use ZOLs.
16885 (hwloop_fail): New function.
16886 (hwloop_optimize): Likewise.
16887 (hwloop_pattern_reg): Likewise.
16888 (arc_doloop_hooks): New struct, to be used with reorg_loops.
16889 (arc_reorg_loops): New function, calls reorg_loops.
16890 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
16891 (arc600_corereg_hazard): Remove ZOL checking, case handled by
16893 (arc_loop_hazard): Remove function, functionality moved into
16895 (arc_hazard): Remove arc_loop_hazard call.
16896 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
16897 into hwloop_optimize.
16898 (arc_label_align): Remove ZOL handling.
16899 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
16900 * config/arc/arc.md (doloop_begin): Remove pattern.
16901 (doloop_begin_i): Likewise.
16902 (doloop_end_i): Likewise.
16903 (doloop_fallback): Likewise.
16904 (doloop_fallback_m): Likewise.
16905 (doloop_end): Reimplement expand.
16906 (arc_lp): New pattern for LP instruction.
16907 (loop_end): New pattern.
16908 (loop_fail): Likewise.
16909 (decrement_and_branch_until_zero): Likewise.
16910 * config/arc/arc.opt (mlpc-width): New option.
16911 * doc/invoke.texi (mlpc-width): Document option.
16913 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16915 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
16916 (arc_ccfsm_advance): Fix checking for delay slots.
16917 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
16919 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16921 * config/arc/arc.md (movqi_insn): Add stores to save constant long
16923 (movhi_insn): Update store instruction constraint which are saving
16924 6-bit short immediates.
16925 (movsi_insn): Consider also short scaled load operations.
16926 (zero_extendhisi2_i): Use Usd constraint instead of T.
16927 (extendhisi2_i): Add q constraint.
16928 (arc_clzsi2): Add type and length attributes.
16929 (arc_ctzsi2): Likewise.
16930 * config/arc/constraints.md (Usc): Update constraint, the
16931 assembler can parse two relocations for a single instruction.
16933 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16935 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
16936 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
16938 2017-08-31 Olivier Hainque <hainque@adacore.com>
16940 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
16941 match as powerpc-wrs-vxworks*.
16943 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
16945 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
16946 register constraint for by-element operand.
16947 (aarch64_mls_elt_merge<mode>): Likewise.
16949 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16951 * config/arc/arc.c (arc_can_follow_jump): Check for short
16954 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16956 * config.gcc: Use g.opt for arc.
16957 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
16958 functionality moved to ...
16959 (legitimate_scaled_address_p): New function, ...here.
16960 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
16961 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
16962 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
16964 (arc_override_options): Handle G option.
16965 (arc_output_pic_addr_const): Correct function definition.
16966 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
16967 (arc_decl_anon_ns_mem_p): Delete.
16968 (arc_in_small_data_p): Overhaul this function to take into
16969 consideration the value given via G option.
16970 (arc_rewrite_small_data_1): Renamed and corrected old
16971 arc_rewrite_small_data function.
16972 (arc_rewrite_small_data): New function.
16973 (small_data_pattern): Don't use pic_offset_table_rtx.
16974 * config/arc/arc.h (CC1_SPEC): Recognize G option.
16975 * config/arc/simdext.md (movmisalignv2hi): Use
16976 prepare_move_operands function.
16978 (movmisalign*): Likewise.
16979 * doc/invoke.texi (ARC options): Document -G option.
16981 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16983 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
16985 * config/arc/arc.c (arc_print_operand): Output scalled address for
16986 sdata whenever is possible.
16987 (arc_in_small_data_p): Allow sdata for 64bit datum when double
16988 load/stores are available.
16989 (compact_sda_memory_operand): Check for the alignment required by
16990 code density instructions.
16991 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
16993 * config/arc/constraints.md (Usd): Update constraint.
16994 (Us0): New constraint.
16995 (Usc): Update constraint.
16997 2017-08-31 Richard Biener <rguenther@suse.de>
16999 PR middle-end/82054
17000 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17001 function only once.
17003 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17005 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17006 Resize type_signature.
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/aarch64.c (aarch64_base_register_rtx_p): Only allow
17013 subregs whose inner modes can be stored in GPRs.
17014 (aarch64_classify_index): Likewise.
17016 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17017 Alan Hayward <alan.hayward@arm.com>
17018 David Sherwood <david.sherwood@arm.com>
17020 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17021 (V_INT_EQUIV): ...this.
17022 (v_cmp_result): Rename to...
17023 (v_int_equiv): ...this.
17024 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17025 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17026 (copysign<mode>3): Likewise.
17027 (aarch64_simd_bsl<mode>_internal): Likewise.
17028 (aarch64_simd_bsl<mode>): Likewise.
17029 (vec_cmp<mode><mode>): Likewise.
17030 (vcond<mode><mode>): Likewise.
17031 (vcond<v_cmp_mixed><mode>): Likewise.
17032 (vcondu<mode><v_cmp_mixed>): Likewise.
17033 (aarch64_cm<optab><mode>): Likewise.
17034 (aarch64_cmtst<mode>): Likewise.
17035 (aarch64_fac<optab><mode>): Likewise.
17036 (vec_perm_const<mode>): Likewise.
17037 (vcond_mask_<mode><v_cmp_result>): Rename to...
17038 (vcond_mask_<mode><v_int_equiv>): ...this.
17039 (vec_cmp<mode><v_cmp_result>): Rename to...
17040 (vec_cmp<mode><v_int_equiv>): ...this.
17042 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17043 Alan Hayward <alan.hayward@arm.com>
17044 David Sherwood <david.sherwood@arm.com>
17046 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17048 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17049 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17050 (UNSPEC_LD4_DREG): New unspecs.
17051 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17052 (aarch64_ld2<mode>_dreg_be): Replace with...
17053 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17055 (aarch64_ld3<mode>_dreg_le)
17056 (aarch64_ld3<mode>_dreg_be): Replace with...
17057 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17059 (aarch64_ld4<mode>_dreg_le)
17060 (aarch64_ld4<mode>_dreg_be): Replace with...
17061 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17064 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17066 PR tree-optimization/81987
17067 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17068 insert an initializer in a location not dominated by the stride
17071 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17073 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17074 on the entire header of the finally block in the fallthru case.
17076 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17078 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17080 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17082 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17083 rs6000_emit_move_from_cr and call renamed function.
17084 (rs6000_emit_prologue): Call renamed functions.
17085 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17086 movesi_from_cr, remove volatile CRs.
17088 2017-08-30 Jon Beniston <jon@beniston.com>
17089 Richard Biener <rguenther@suse.de>
17091 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17092 of VECTOR_MODE_P check.
17093 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17094 element vector types.
17096 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17098 * df.h (df_read_modify_subreg_p): Remove in favor of...
17099 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17100 const_rtx instead of an rtx.
17101 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17102 * df-problems.c (df_word_lr_mark_ref): Likewise.
17103 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17104 (mark_pseudo_reg_dead): Likewise.
17105 (mark_ref_dead): Likewise.
17106 * reginfo.c (init_subregs_of_mode): Likewise.
17107 * sched-deps.c (sched_analyze_1): Likewise.
17108 * df-scan.c (df_def_record_1): Likewise.
17109 (df_uses_record): Likewise.
17110 (df_read_modify_subreg_p): Remove in favor of...
17111 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17112 const_rtx instead of an rtx.
17114 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17115 Alan Hayward <alan.hayward@arm.com>
17116 David Sherwood <david.sherwood@arm.com>
17118 * rtl.h (partial_subreg_p): New function.
17119 * caller-save.c (save_call_clobbered_regs): Use it.
17120 * calls.c (expand_call): Likewise.
17121 * combine.c (combinable_i3pat): Likewise.
17122 (simplify_set): Likewise.
17123 (make_extraction): Likewise.
17124 (make_compound_operation_int): Likewise.
17125 (gen_lowpart_or_truncate): Likewise.
17126 (force_to_mode): Likewise.
17127 (make_field_assignment): Likewise.
17128 (reg_truncated_to_mode): Likewise.
17129 (record_truncated_value): Likewise.
17130 (move_deaths): Likewise.
17131 * cse.c (record_jump_cond): Likewise.
17132 (cse_insn): Likewise.
17133 * cselib.c (cselib_lookup_1): Likewise.
17134 * expmed.c (extract_bit_field_using_extv): Likewise.
17135 * function.c (assign_parm_setup_reg): Likewise.
17136 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17137 * ira-build.c (create_insn_allocnos): Likewise.
17138 * lra-coalesce.c (merge_pseudos): Likewise.
17139 * lra-constraints.c (match_reload): Likewise.
17140 (simplify_operand_subreg): Likewise.
17141 (curr_insn_transform): Likewise.
17142 * lra-lives.c (process_bb_lives): Likewise.
17143 * lra.c (new_insn_reg): Likewise.
17144 (lra_substitute_pseudo): Likewise.
17145 * regcprop.c (mode_change_ok): Likewise.
17146 (maybe_mode_change): Likewise.
17147 (copyprop_hardreg_forward_1): Likewise.
17148 * reload.c (push_reload): Likewise.
17149 (find_reloads): Likewise.
17150 (find_reloads_subreg_address): Likewise.
17151 * reload1.c (alter_reg): Likewise.
17152 (eliminate_regs_1): Likewise.
17153 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17155 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17157 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17158 back to if statements, including unpack.
17160 2017-08-30 Martin Liska <mliska@suse.cz>
17162 PR inline-asm/82001
17163 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17165 (func_checker::compare_asm_inputs_outputs): ... this function.
17166 (func_checker::compare_gimple_asm): Use the function to compare
17167 also ASM constrains.
17168 * ipa-icf-gimple.h: Rename the function.
17170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17171 Alan Hayward <alan.hayward@arm.com>
17172 David Sherwood <david.sherwood@arm.com>
17174 * coretypes.h (complex_mode): New type.
17175 * gdbhooks.py (build_pretty_printer): Handle it.
17176 * machmode.h (complex_mode): New class.
17177 (complex_mode::includes_p): New function.
17178 (is_complex_int_mode): Likewise.
17179 (is_complex_float_mode): Likewise.
17180 * genmodes.c (get_mode_class): Handle complex mode classes.
17181 * function.c (expand_function_end): Use is_complex_int_mode.
17183 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17184 Alan Hayward <alan.hayward@arm.com>
17185 David Sherwood <david.sherwood@arm.com>
17187 * coretypes.h (scalar_mode_pod): New typedef.
17188 * gdbhooks.py (build_pretty_printer): Handle it.
17189 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17190 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17191 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17192 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17193 as_a <scalar_mode>.
17195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17196 Alan Hayward <alan.hayward@arm.com>
17197 David Sherwood <david.sherwood@arm.com>
17199 * machmode.h (mode_for_vector): Take a scalar_mode instead
17201 * stor-layout.c (mode_for_vector): Likewise.
17202 * explow.c (promote_mode): Use as_a <scalar_mode>.
17203 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17206 Alan Hayward <alan.hayward@arm.com>
17207 David Sherwood <david.sherwood@arm.com>
17209 * target.def (preferred_simd_mode): Take a scalar_mode
17210 instead of a machine_mode.
17211 * targhooks.h (default_preferred_simd_mode): Likewise.
17212 * targhooks.c (default_preferred_simd_mode): Likewise.
17213 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17214 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17215 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17216 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17217 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17218 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17219 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17220 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17222 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17223 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17224 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17225 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17226 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17227 * doc/tm.texi: Regenerate.
17228 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17232 Alan Hayward <alan.hayward@arm.com>
17233 David Sherwood <david.sherwood@arm.com>
17235 * target.def (scalar_mode_supported_p): Take a scalar_mode
17236 instead of a machine_mode.
17237 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17238 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17239 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17240 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17241 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17242 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17243 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17244 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17245 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17246 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17247 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17248 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17249 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17250 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17252 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17253 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17254 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17255 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17256 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17258 * doc/tm.texi: Regenerate.
17260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17261 Alan Hayward <alan.hayward@arm.com>
17262 David Sherwood <david.sherwood@arm.com>
17264 * coretypes.h (opt_scalar_mode): New typedef.
17265 * gdbhooks.py (build_pretty_printers): Handle it.
17266 * machmode.h (mode_iterator::get_2xwider): Add overload for
17268 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17270 * expr.c (convert_mode_scalar): Likewise.
17271 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17272 * optabs.c (expand_float): Likewise.
17273 (expand_fix): Likewise.
17274 * tree-vect-stmts.c (vectorizable_conversion): 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 * optabs.c (expand_float): Explicitly check for scalars before
17281 using a branching expansion.
17282 (expand_fix): Likewise.
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 * expr.c (convert_mode): Split scalar handling out into...
17289 (convert_mode_scalar): ...this new function. Treat the modes
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 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17298 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17300 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17301 Alan Hayward <alan.hayward@arm.com>
17302 David Sherwood <david.sherwood@arm.com>
17304 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17305 rather than a machine_mode.
17306 (fixed_from_string): Likewise.
17307 (fixed_convert): Likewise.
17308 (fixed_convert_from_int): Likewise.
17309 (fixed_convert_from_real): Likewise.
17310 (real_convert_from_fixed): Likewise.
17311 * fixed-value.c (fixed_from_double_int): Likewise.
17312 (fixed_from_string): Likewise.
17313 (fixed_convert): Likewise.
17314 (fixed_convert_from_int): Likewise.
17315 (fixed_convert_from_real): Likewise.
17316 (real_convert_from_fixed): Likewise.
17317 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17320 Alan Hayward <alan.hayward@arm.com>
17321 David Sherwood <david.sherwood@arm.com>
17323 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17324 of separate mode class checks. Do not allow vector modes here.
17325 (immed_wide_int_const): Use as_a <scalar_mode>.
17326 * explow.c (trunc_int_for_mode): Likewise.
17327 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17328 (wi::shwi): Likewise.
17329 (wi::min_value): Likewise.
17330 (wi::max_value): Likewise.
17331 * dwarf2out.c (loc_descriptor): Likewise.
17332 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17333 for CONST_WIDE_INT.
17335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17336 Alan Hayward <alan.hayward@arm.com>
17337 David Sherwood <david.sherwood@arm.com>
17339 * tree.h (SCALAR_TYPE_MODE): New macro.
17340 * expr.c (expand_expr_addr_expr_1): Use it.
17341 (expand_expr_real_2): Likewise.
17342 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17343 (fold_convert_const_fixed_from_int): Likewise.
17344 (fold_convert_const_fixed_from_real): Likewise.
17345 (native_encode_fixed): Likewise
17346 (native_encode_complex): Likewise
17347 (native_encode_vector): Likewise.
17348 (native_interpret_fixed): Likewise.
17349 (native_interpret_real): Likewise.
17350 (native_interpret_complex): Likewise.
17351 (native_interpret_vector): Likewise.
17352 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17353 (simd_clone_adjust_argument_types): Likewise.
17354 (simd_clone_init_simd_arrays): Likewise.
17355 (simd_clone_adjust): Likewise.
17356 * stor-layout.c (layout_type): Likewise.
17357 * tree.c (build_minus_one_cst): Likewise.
17358 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17359 * tree-inline.c (estimate_move_cost): Likewise.
17360 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17361 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17362 (vectorizable_reduction): Likewise.
17363 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17364 (vect_recog_mixed_size_cond_pattern): Likewise.
17365 (check_bool_pattern): Likewise.
17366 (adjust_bool_pattern): Likewise.
17367 (search_type_for_mask_1): Likewise.
17368 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17369 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17370 (vectorizable_load): Likewise.
17371 (vectorizable_store): Likewise.
17372 * ubsan.c (ubsan_encode_value): Likewise.
17373 * varasm.c (output_constant): Likewise.
17375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17376 Alan Hayward <alan.hayward@arm.com>
17377 David Sherwood <david.sherwood@arm.com>
17379 * coretypes.h (scalar_mode): New class.
17380 * machmode.h (scalar_mode): Likewise.
17381 (scalar_mode::includes_p): New function.
17382 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17383 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17384 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17385 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17386 * expmed.c (store_bit_field_1): Likewise.
17387 (extract_bit_field_1): Likewise.
17388 * expr.c (write_complex_part): Likewise.
17389 (read_complex_part): Likewise.
17390 (emit_move_complex_push): Likewise.
17391 (expand_expr_real_2): Likewise.
17392 * function.c (assign_parm_setup_reg): Likewise.
17393 (assign_parms_unsplit_complex): Likewise.
17394 * optabs.c (expand_binop): Likewise.
17395 * rtlanal.c (subreg_get_info): Likewise.
17396 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17397 * varasm.c (output_constant_pool_2): Likewise.
17399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17400 Alan Hayward <alan.hayward@arm.com>
17401 David Sherwood <david.sherwood@arm.com>
17403 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17405 (expmed_mult_highpart_optab): Likewise.
17406 (expmed_mult_highpart): Likewise.
17408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17409 Alan Hayward <alan.hayward@arm.com>
17410 David Sherwood <david.sherwood@arm.com>
17412 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17413 instead of a machine_mode.
17414 (builtin_memset_read_str): Likewise.
17415 * builtins.c (c_readstr): Likewise.
17416 (builtin_memcpy_read_str): Likewise.
17417 (builtin_strncpy_read_str): Likewise.
17418 (builtin_memset_read_str): Likewise.
17419 (builtin_memset_gen_str): Likewise.
17420 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17421 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17422 instead of a machine_mode.
17423 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17425 (make_extraction): Likewise.
17426 (try_widen_shift_mode): Take and return scalar_int_modes instead
17428 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17429 a scalar_int_mode instead of a machine_mode.
17430 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17431 (avr_addr_space_pointer_mode): Likewise.
17432 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17433 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17434 (msp430_unwind_word_mode): Likewise.
17435 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17436 (spu_addr_space_pointer_mode): Likewise.
17437 (spu_addr_space_address_mode): Likewise.
17438 (spu_libgcc_cmp_return_mode): Likewise.
17439 (spu_libgcc_shift_count_mode): Likewise.
17440 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17441 (rl78_addr_space_pointer_mode): Likewise.
17442 (fl78_unwind_word_mode): Likewise.
17443 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17445 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17446 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17447 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17448 (mips_valid_pointer_mode): Likewise.
17449 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17450 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17451 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17453 (ft32_addr_space_address_mode): Likewise.
17454 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17455 scalar_int_mode instead of a machine_mode.
17456 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17458 (m32c_addr_space_address_mode): Likewise.
17459 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17460 (rs6000_eh_return_filter_mode): Likewise.
17461 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17462 (rs6000_eh_return_filter_mode): Likewise.
17463 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17464 (s390_libgcc_shift_count_mode): Likewise.
17465 (s390_unwind_word_mode): Likewise.
17466 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17468 * target.def (mode_rep_extended): Likewise.
17469 (valid_pointer_mode): Likewise.
17470 (addr_space.valid_pointer_mode): Likewise.
17471 (eh_return_filter_mode): Return a scalar_int_mode rather than
17473 (libgcc_cmp_return_mode): Likewise.
17474 (libgcc_shift_count_mode): Likewise.
17475 (unwind_word_mode): Likewise.
17476 (addr_space.pointer_mode): Likewise.
17477 (addr_space.address_mode): Likewise.
17478 * doc/tm.texi: Regenerate.
17479 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17481 (do_jump): Use scalar_int_mode for local variables.
17482 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17483 rather than a machine_mode.
17484 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17485 (scompare_loc_descriptor_wide): Likewise.
17486 (scompare_loc_descriptor_narrow): Likewise.
17487 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17489 * except.c (sjlj_emit_dispatch_table): Likewise.
17490 (expand_builtin_eh_copy_values): Likewise.
17491 * explow.c (convert_memory_address_addr_space_1): Likewise.
17492 Take a scalar_int_mode rather than a machine_mode.
17493 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17494 than a machine_mode.
17495 (memory_address_addr_space): Use scalar_int_mode for local variables.
17496 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17497 rather than a machine_mode.
17498 * expmed.c (mask_rtx): Likewise.
17499 (init_expmed_one_conv): Likewise.
17500 (expand_mult_highpart_adjust): Likewise.
17501 (extract_high_half): Likewise.
17502 (expmed_mult_highpart_optab): Likewise.
17503 (expmed_mult_highpart): Likewise.
17504 (expand_smod_pow2): Likewise.
17505 (expand_sdiv_pow2): Likewise.
17506 (emit_store_flag_int): Likewise.
17507 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17509 (extract_low_bits): Likewise.
17510 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17512 * expr.c (pieces_addr::adjust): Likewise.
17513 (can_store_by_pieces): Likewise.
17514 (store_by_pieces): Likewise.
17515 (clear_by_pieces_1): Likewise.
17516 (expand_expr_addr_expr_1): Likewise.
17517 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17518 (expand_expr_real_1): Likewise.
17519 (try_casesi): Likewise.
17520 * final.c (shorten_branches): Likewise.
17521 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17522 type of "mode" to machine_mode.
17523 * internal-fn.c (expand_arith_overflow_result_store): Take a
17524 scalar_int_mode rather than a machine_mode.
17525 (expand_mul_overflow): Use scalar_int_mode for local variables.
17526 * loop-doloop.c (doloop_modify): Likewise.
17527 (doloop_optimize): Likewise.
17528 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17529 than a machine_mode.
17530 (expand_doubleword_shift_condmove): Likewise.
17531 (expand_doubleword_shift): Likewise.
17532 (expand_doubleword_clz): Likewise.
17533 (expand_doubleword_popcount): Likewise.
17534 (expand_doubleword_parity): Likewise.
17535 (expand_absneg_bit): Use scalar_int_mode for local variables.
17536 (prepare_float_lib_cmp): Likewise.
17537 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17538 rather than a machine_mode.
17539 (convert_memory_address_addr_space): Likewise.
17540 (get_mode_bounds): Likewise.
17541 (get_address_mode): Return a scalar_int_mode rather than a
17543 * rtlanal.c (get_address_mode): Likewise.
17544 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17545 than a machine_mode.
17546 * targhooks.c (default_mode_rep_extended): Likewise.
17547 (default_valid_pointer_mode): Likewise.
17548 (default_addr_space_valid_pointer_mode): Likewise.
17549 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17550 than a machine_mode.
17551 (default_libgcc_cmp_return_mode): Likewise.
17552 (default_libgcc_shift_count_mode): Likewise.
17553 (default_unwind_word_mode): Likewise.
17554 (default_addr_space_pointer_mode): Likewise.
17555 (default_addr_space_address_mode): Likewise.
17556 * targhooks.h (default_eh_return_filter_mode): Likewise.
17557 (default_libgcc_cmp_return_mode): Likewise.
17558 (default_libgcc_shift_count_mode): Likewise.
17559 (default_unwind_word_mode): Likewise.
17560 (default_addr_space_pointer_mode): Likewise.
17561 (default_addr_space_address_mode): Likewise.
17562 (default_mode_rep_extended): Take a scalar_int_mode rather than
17564 (default_valid_pointer_mode): Likewise.
17565 (default_addr_space_valid_pointer_mode): Likewise.
17566 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17568 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17569 rather than a machine_mode.
17570 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17571 for local variables.
17572 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17573 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17574 than a machine_mode.
17576 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17577 Alan Hayward <alan.hayward@arm.com>
17578 David Sherwood <david.sherwood@arm.com>
17580 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17581 the mode argument to scalar_int_mode.
17582 (do_jump_by_parts_zero_rtx): Likewise.
17583 (do_jump_by_parts_equality_rtx): Likewise.
17584 (do_jump_by_parts_greater): Take a mode argument.
17585 (do_jump_by_parts_equality): Likewise.
17586 (do_jump_1): Update calls accordingly.
17588 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17589 Alan Hayward <alan.hayward@arm.com>
17590 David Sherwood <david.sherwood@arm.com>
17592 * is-a.h (safe_dyn_cast): New function.
17593 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17594 (jump_table_for_label): Likewise.
17595 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17596 instead of an rtx_insn *.
17597 (shorten_branches): Use dyn_cast instead of LABEL_P and
17598 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17599 rtx_jump_table_data::get_data_mode.
17600 (final_scan_insn): Likewise.
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 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17607 trying to combine a full-register integer set with a subreg
17610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17611 Alan Hayward <alan.hayward@arm.com>
17612 David Sherwood <david.sherwood@arm.com>
17614 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17615 that is always either address_mode or pointer_mode.
17617 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17618 Alan Hayward <alan.hayward@arm.com>
17619 David Sherwood <david.sherwood@arm.com>
17621 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17622 when the two are known to be equal.
17624 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17625 Alan Hayward <alan.hayward@arm.com>
17626 David Sherwood <david.sherwood@arm.com>
17628 * simplify-rtx.c (simplify_const_unary_operation): Use
17629 is_a <scalar_int_mode> instead of checking for a nonzero
17630 precision. Forcibly convert op_mode to a scalar_int_mode
17631 in that case. More clearly differentiate the operand and
17632 result modes and use the former when deciding what the value
17633 of a count-bits operation should be. Use is_int_mode instead
17634 of checking for a MODE_INT. Remove redundant check for whether
17635 this mode has a zero precision.
17637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17638 Alan Hayward <alan.hayward@arm.com>
17639 David Sherwood <david.sherwood@arm.com>
17641 * optabs.c (widen_leading): Change the type of the mode argument
17642 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17643 (widen_bswap): Likewise.
17644 (expand_parity): Likewise.
17645 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17646 (expand_ffs): Likewise.
17647 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17656 (expand_expr_real_1): Update call accordingly.
17658 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17659 Alan Hayward <alan.hayward@arm.com>
17660 David Sherwood <david.sherwood@arm.com>
17662 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17663 value_mode arguments. Use scalar_int_mode internally.
17664 (store_bit_field_1): Rename the new integer mode from imode
17665 to op0_mode and use it instead of GET_MODE (op0). Update calls
17666 to store_split_bit_field, store_bit_field_using_insv and
17667 store_fixed_bit_field.
17668 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17669 Use scalar_int_mode internally. Use a bit count rather than a mode
17670 when calculating the largest bit size for get_best_mode.
17671 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17672 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17673 Remove assertion that OP0 has a scalar integer mode.
17674 (store_split_bit_field): Add op0_mode and value_mode arguments.
17675 Update calls to extract_fixed_bit_field.
17676 (extract_bit_field_using_extv): Add an op0_mode argument.
17677 Use scalar_int_mode internally.
17678 (extract_bit_field_1): Rename the new integer mode from imode to
17679 op0_mode and use it instead of GET_MODE (op0). Update calls to
17680 extract_split_bit_field, extract_bit_field_using_extv and
17681 extract_fixed_bit_field.
17682 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17683 to extract_split_bit_field and extract_fixed_bit_field_1.
17684 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17685 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17686 on the target mode.
17687 (extract_split_bit_field): Add an op0_mode argument. Update call
17688 to extract_fixed_bit_field.
17690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17691 Alan Hayward <alan.hayward@arm.com>
17692 David Sherwood <david.sherwood@arm.com>
17694 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17695 * explow.c (hard_function_value): Likewise.
17696 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17697 convert_to_mode call outside the loop.
17698 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17699 for the mode iterator. Require the mode specified by max_pieces
17701 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17703 (copy_blkmode_to_reg): Likewise.
17704 (set_storage_via_setmem): Likewise.
17705 * optabs.c (prepare_cmp_insn): Likewise.
17706 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17707 * stor-layout.c (finish_bitfield_representative): Likewise.
17709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17710 Alan Hayward <alan.hayward@arm.com>
17711 David Sherwood <david.sherwood@arm.com>
17713 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17714 * expr.c (convert_move): Use them.
17715 (convert_modes): Likewise.
17716 (store_expr_with_bounds): Likewise.
17718 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17719 Alan Hayward <alan.hayward@arm.com>
17720 David Sherwood <david.sherwood@arm.com>
17722 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17723 parameter for the mode of "x". Remove the "known_x", "known_mode"
17724 and "known_ret" arguments. Change the type of the mode argument
17725 to scalar_int_mode.
17726 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17727 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17728 (reg_num_sign_bit_copies_for_combine): Likewise.
17729 * rtlanal.c (nonzero_bits1): Likewise.
17730 (num_sign_bit_copies1): Likewise.
17731 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17732 (reg_num_sign_bit_copies_general): Likewise.
17733 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17734 (reg_nonzero_bits_general): Likewise.
17736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17737 Alan Hayward <alan.hayward@arm.com>
17738 David Sherwood <david.sherwood@arm.com>
17740 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17741 than in subroutines. Return 1 for non-integer modes.
17742 (cached_num_sign_bit_copies): Change the type of the mode parameter
17743 to scalar_int_mode.
17744 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17745 classes. Handle CONST_INT_P first and then check whether X also
17746 has a scalar integer mode. Check the same thing for inner registers
17747 of a SUBREG and for values that are being extended or truncated.
17749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17750 Alan Hayward <alan.hayward@arm.com>
17751 David Sherwood <david.sherwood@arm.com>
17753 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17754 in subroutines. Return the mode mask for non-integer modes.
17755 (cached_nonzero_bits): Change the type of the mode parameter
17756 to scalar_int_mode.
17757 (nonzero_bits1): Likewise. Remove early exit for other mode
17758 classes. Handle CONST_INT_P first and then check whether X
17759 also has a scalar integer mode.
17761 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17762 Alan Hayward <alan.hayward@arm.com>
17763 David Sherwood <david.sherwood@arm.com>
17765 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17766 Return a scalar_int_mode and assert that the size is greater than
17768 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17769 (op_by_pieces_d::op_by_pieces_d): Likewise.
17770 (op_by_pieces_d::run): Likewise.
17771 (can_store_by_pieces): Likewise.
17773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17774 Alan Hayward <alan.hayward@arm.com>
17775 David Sherwood <david.sherwood@arm.com>
17777 * combine.c (extract_left_shift): Add a mode argument and update
17779 (make_compound_operation_int): Change the type of the mode parameter
17780 to scalar_int_mode and update the call to extract_left_shift.
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_and_const_int): Change the type of the mode
17787 parameter to scalar_int_mode.
17788 (simplify_and_const_int_1): Likewise. Update recursive call.
17790 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17791 Alan Hayward <alan.hayward@arm.com>
17792 David Sherwood <david.sherwood@arm.com>
17794 * combine.c (simplify_compare_const): Check that the mode is a
17795 scalar_int_mode (rather than VOIDmode) before testing its
17797 (simplify_comparison): Move COMPARISON_P handling out of the
17798 loop and restrict the latter part of the loop to scalar_int_modes.
17799 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17800 and when considering SUBREG_REGs. Use is_int_mode instead of
17801 checking GET_MODE_CLASS against MODE_INT.
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 (try_widen_shift_mode): Move check for equal modes to...
17808 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17809 shift_unit_mode and for modes involved in scalar shifts.
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 * combine.c (force_int_to_mode): New function, split out from...
17816 (force_to_mode): ...here. Keep xmode up-to-date and use it
17817 instead of GET_MODE (x).
17819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17820 Alan Hayward <alan.hayward@arm.com>
17821 David Sherwood <david.sherwood@arm.com>
17823 * optabs-query.h (extraction_insn::struct_mode): Change type to
17824 opt_scalar_int_mode and update comment.
17825 (extraction_insn::field_mode): Change type to scalar_int_mode.
17826 (extraction_insn::pos_mode): Likewise.
17827 * combine.c (make_extraction): Update accordingly.
17828 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17829 (get_optab_extraction_insn): Likewise.
17830 * recog.c (simplify_while_replacing): Likewise.
17831 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17832 parameter to opt_scalar_int_mode.
17834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17835 Alan Hayward <alan.hayward@arm.com>
17836 David Sherwood <david.sherwood@arm.com>
17838 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17839 to a scalar_int_mode instead of a machine_mode.
17840 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17841 (get_best_mode): Return a boolean and use a pointer argument to store
17842 the selected mode. Replace the limit mode parameter with a bit limit.
17843 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17844 for the values returned by bit_field_mode_iterator::next_mode.
17845 (store_bit_field): Update call to get_best_mode.
17846 (store_fixed_bit_field): Likewise.
17847 (extract_fixed_bit_field): Likewise.
17848 * expr.c (optimize_bitfield_assignment_op): Likewise.
17849 * fold-const.c (optimize_bit_field_compare): Likewise.
17850 (fold_truth_andor_1): Likewise.
17851 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17852 Update for new type of m_mode.
17853 (get_best_mode): As above.
17855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17856 Alan Hayward <alan.hayward@arm.com>
17857 David Sherwood <david.sherwood@arm.com>
17859 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
17860 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
17861 (store_bit_field): Check is_a <scalar_int_mode> before calling
17862 strict_volatile_bitfield_p.
17863 (extract_bit_field): Likewise.
17865 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17866 Alan Hayward <alan.hayward@arm.com>
17867 David Sherwood <david.sherwood@arm.com>
17869 * target.def (cstore_mode): Return a scalar_int_mode.
17870 * doc/tm.texi: Regenerate.
17871 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
17872 * targhooks.h (default_cstore_mode): Likewise.
17873 * targhooks.c (default_cstore_mode): Likewise, using a forced
17875 * expmed.c (emit_cstore): Expect the target of the cstore to be
17878 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17879 Alan Hayward <alan.hayward@arm.com>
17880 David Sherwood <david.sherwood@arm.com>
17882 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
17884 (niter_desc): Likewise mode.
17885 (iv_analyze): Add a mode parameter.
17887 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
17888 and change its type to scalar_int_mode.
17889 * loop-iv.c: Update commentary at head of file.
17890 (iv_constant): Pass the mode paraeter before the rtx it describes
17891 and change its type to scalar_int_mode. Remove VOIDmode handling.
17892 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
17893 (iv_extend): Likewise.
17894 (shorten_into_mode): Likewise.
17895 (iv_add): Use scalar_int_mode.
17896 (iv_mult): Likewise.
17897 (iv_shift): Likewise.
17898 (canonicalize_iv_subregs): Likewise.
17899 (get_biv_step_1): Pass the outer_mode parameter before the rtx
17900 it describes and change its mode to scalar_int_mode. Also change
17901 the type of the returned inner_mode to scalar_int_mode.
17902 (get_biv_step): Likewise, turning outer_mode from a pointer
17903 into a direct parameter. Update call to get_biv_step_1.
17904 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
17905 iv_constant and get_biv_step.
17906 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
17907 and change its type to scalar_int_mode. Don't initialise iv->mode
17908 to VOIDmode and remove later checks for its still being VOIDmode.
17909 Update calls to iv_analyze_op and iv_analyze_expr. Check
17910 is_a <scalar_int_mode> when changing the mode under consideration.
17911 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
17912 Update call to iv_analyze_expr.
17913 (iv_analyze_op): Add a mode parameter. Reject subregs whose
17914 inner register is not also a scalar_int_mode. Update call to
17916 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
17917 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
17918 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
17919 separate mode class checks. Update calls to iv_analyze. Remove
17920 fix-up of VOIDmodes after iv_analyze_biv.
17921 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
17922 don't have a scalar_int_mode. Update call to biv_p.
17924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17925 Alan Hayward <alan.hayward@arm.com>
17926 David Sherwood <david.sherwood@arm.com>
17928 * cfgexpand.c (convert_debug_memory_address): Use
17929 as_a <scalar_int_mode>.
17930 * combine.c (expand_compound_operation): Likewise.
17931 (make_extraction): Likewise.
17932 (change_zero_ext): Likewise.
17933 (simplify_comparison): Likewise.
17934 * cse.c (cse_insn): Likewise.
17935 * dwarf2out.c (minmax_loc_descriptor): Likewise.
17936 (mem_loc_descriptor): Likewise.
17937 (loc_descriptor): Likewise.
17938 * expmed.c (init_expmed_one_mode): Likewise.
17939 (synth_mult): Likewise.
17940 (emit_store_flag_1): Likewise.
17941 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
17942 of a comparison with size.
17943 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
17944 (reduce_to_bit_field_precision): Likewise.
17945 * function.c (expand_function_end): Likewise.
17946 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
17947 * loop-doloop.c (doloop_modify): Likewise.
17948 * optabs.c (expand_binop): Likewise.
17949 (expand_unop): Likewise.
17950 (expand_copysign_absneg): Likewise.
17951 (prepare_cmp_insn): Likewise.
17952 (maybe_legitimize_operand): Likewise.
17953 * recog.c (const_scalar_int_operand): Likewise.
17954 * rtlanal.c (get_address_mode): Likewise.
17955 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17956 (simplify_cond_clz_ctz): Likewise.
17957 * tree-nested.c (get_nl_goto_field): Likewise.
17958 * tree.c (build_vector_type_for_mode): Likewise.
17959 * var-tracking.c (use_narrower_mode): Likewise.
17961 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17962 Alan Hayward <alan.hayward@arm.com>
17963 David Sherwood <david.sherwood@arm.com>
17965 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
17966 * builtins.c (expand_builtin_signbit): Use it.
17967 * cfgexpand.c (expand_debug_expr): Likewise.
17968 * dojump.c (do_jump): Likewise.
17969 (do_compare_and_jump): Likewise.
17970 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
17971 * expmed.c (make_tree): Likewise.
17972 * expr.c (expand_expr_real_2): Likewise.
17973 (expand_expr_real_1): Likewise.
17974 (try_casesi): Likewise.
17975 * fold-const-call.c (fold_const_call_ss): Likewise.
17976 * fold-const.c (unextend): Likewise.
17977 (extract_muldiv_1): Likewise.
17978 (fold_single_bit_test): Likewise.
17979 (native_encode_int): Likewise.
17980 (native_encode_string): Likewise.
17981 (native_interpret_int): Likewise.
17982 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
17983 * internal-fn.c (expand_addsub_overflow): Likewise.
17984 (expand_neg_overflow): Likewise.
17985 (expand_mul_overflow): Likewise.
17986 (expand_arith_overflow): Likewise.
17987 * match.pd: Likewise.
17988 * stor-layout.c (layout_type): Likewise.
17989 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17990 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
17991 * tree-ssanames.c (get_range_info): Likewise.
17992 * tree-switch-conversion.c (array_value_type) Likewise.
17993 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
17994 (vect_recog_divmod_pattern): Likewise.
17995 (vect_recog_mixed_size_cond_pattern): Likewise.
17996 * tree-vrp.c (extract_range_basic): Likewise.
17997 (simplify_float_conversion_using_ranges): Likewise.
17998 * tree.c (int_fits_type_p): Likewise.
17999 * ubsan.c (instrument_bool_enum_load): Likewise.
18000 * varasm.c (mergeable_string_section): Likewise.
18001 (narrowing_initializer_constant_valid_p): Likewise.
18002 (output_constant): Likewise.
18004 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18005 Alan Hayward <alan.hayward@arm.com>
18006 David Sherwood <david.sherwood@arm.com>
18008 * machmode.h (NARROWEST_INT_MODE): New macro.
18009 * expr.c (alignment_for_piecewise_move): Use it instead of
18010 GET_CLASS_NARROWEST_MODE (MODE_INT).
18011 (push_block): Likewise.
18012 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18014 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18017 Alan Hayward <alan.hayward@arm.com>
18018 David Sherwood <david.sherwood@arm.com>
18020 * postreload.c (move2add_valid_value_p): Change the type of the
18021 mode parameter to scalar_int_mode.
18022 (move2add_use_add2_insn): Add a mode parameter and use it instead
18024 (move2add_use_add3_insn): Likewise.
18025 (reload_cse_move2add): Update accordingly.
18027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18028 Alan Hayward <alan.hayward@arm.com>
18029 David Sherwood <david.sherwood@arm.com>
18031 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18033 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18034 * optabs.c (expand_unop): Likewise.
18036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18037 Alan Hayward <alan.hayward@arm.com>
18038 David Sherwood <david.sherwood@arm.com>
18040 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18041 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18042 (popcount_loc_descriptor): Likewise.
18043 (bswap_loc_descriptor): Likewise.
18044 (rotate_loc_descriptor): Likewise.
18045 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18046 calling the functions above.
18048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18049 Alan Hayward <alan.hayward@arm.com>
18050 David Sherwood <david.sherwood@arm.com>
18052 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18054 (try_combine): Likewise.
18055 (simplify_if_then_else): Likewise.
18056 * cse.c (cse_insn): Likewise.
18057 * dwarf2out.c (mem_loc_descriptor): Likewise.
18058 * emit-rtl.c (gen_lowpart_common): Likewise.
18059 * simplify-rtx.c (simplify_truncation): Likewise.
18060 (simplify_binary_operation_1): Likewise.
18061 (simplify_const_relational_operation): Likewise.
18062 (simplify_ternary_operation): Likewise.
18063 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): 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 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18070 * reload.c (push_reload): Likewise.
18071 (find_reloads): Likewise.
18073 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18074 Alan Hayward <alan.hayward@arm.com>
18075 David Sherwood <david.sherwood@arm.com>
18077 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18078 (make_compound_operation_int): Likewise.
18079 (change_zero_ext): Likewise.
18080 * expr.c (convert_move): Likewise.
18081 (convert_modes): Likewise.
18082 * fwprop.c (forward_propagate_subreg): Likewise.
18083 * loop-iv.c (get_biv_step_1): Likewise.
18084 * optabs.c (widen_operand): Likewise.
18085 * postreload.c (move2add_valid_value_p): Likewise.
18086 * recog.c (simplify_while_replacing): Likewise.
18087 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18088 (simplify_binary_operation_1): Likewise. Remove redundant
18089 mode equality check.
18091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18092 Alan Hayward <alan.hayward@arm.com>
18093 David Sherwood <david.sherwood@arm.com>
18095 * combine.c (combine_simplify_rtx): Add checks for
18096 is_a <scalar_int_mode>.
18097 (simplify_if_then_else): Likewise.
18098 (make_field_assignment): Likewise.
18099 (simplify_comparison): Likewise.
18100 * ifcvt.c (noce_try_bitop): Likewise.
18101 * loop-invariant.c (canonicalize_address_mult): Likewise.
18102 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18104 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18105 Alan Hayward <alan.hayward@arm.com>
18106 David Sherwood <david.sherwood@arm.com>
18108 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18109 is_a <scalar_int_mode> instead of != BLKmode.
18111 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18112 Alan Hayward <alan.hayward@arm.com>
18113 David Sherwood <david.sherwood@arm.com>
18115 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18116 instead of != VOIDmode.
18117 * combine.c (if_then_else_cond): Likewise.
18118 (change_zero_ext): Likewise.
18119 * dwarf2out.c (mem_loc_descriptor): Likewise.
18120 (loc_descriptor): Likewise.
18121 * rtlanal.c (canonicalize_condition): Likewise.
18122 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18124 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18125 Alan Hayward <alan.hayward@arm.com>
18126 David Sherwood <david.sherwood@arm.com>
18128 * simplify-rtx.c (simplify_binary_operation_1): Use
18129 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18132 Alan Hayward <alan.hayward@arm.com>
18133 David Sherwood <david.sherwood@arm.com>
18135 * wide-int.h (int_traits<unsigned char>) New class.
18136 (int_traits<unsigned short>) Likewise.
18137 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18138 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18140 * combine.c (set_nonzero_bits_and_sign_copies): Use
18141 is_a <scalar_int_mode>.
18142 (find_split_point): Likewise.
18143 (combine_simplify_rtx): Likewise.
18144 (simplify_logical): Likewise.
18145 (expand_compound_operation): Likewise.
18146 (expand_field_assignment): Likewise.
18147 (make_compound_operation): Likewise.
18148 (extended_count): Likewise.
18149 (change_zero_ext): Likewise.
18150 (simplify_comparison): Likewise.
18151 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18152 (ucompare_loc_descriptor): Likewise.
18153 (minmax_loc_descriptor): Likewise.
18154 (mem_loc_descriptor): Likewise.
18155 (loc_descriptor): Likewise.
18156 * expmed.c (init_expmed_one_mode): Likewise.
18157 * lra-constraints.c (lra_constraint_offset): Likewise.
18158 * optabs.c (prepare_libcall_arg): Likewise.
18159 * postreload.c (move2add_note_store): Likewise.
18160 * reload.c (operands_match_p): Likewise.
18161 * rtl.h (load_extend_op): Likewise.
18162 * rtlhooks.c (gen_lowpart_general): Likewise.
18163 * simplify-rtx.c (simplify_truncation): Likewise.
18164 (simplify_unary_operation_1): Likewise.
18165 (simplify_binary_operation_1): Likewise.
18166 (simplify_const_binary_operation): Likewise.
18167 (simplify_const_relational_operation): Likewise.
18168 (simplify_subreg): Likewise.
18169 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18170 * var-tracking.c (adjust_mems): Likewise.
18171 (prepare_call_arguments): Likewise.
18173 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18174 Alan Hayward <alan.hayward@arm.com>
18175 David Sherwood <david.sherwood@arm.com>
18177 * machmode.h (is_int_mode): New fuction.
18178 * combine.c (find_split_point): Use it.
18179 (combine_simplify_rtx): Likewise.
18180 (simplify_if_then_else): Likewise.
18181 (simplify_set): Likewise.
18182 (simplify_shift_const_1): Likewise.
18183 (simplify_comparison): Likewise.
18184 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18185 * cse.c (notreg_cost): Likewise.
18186 (cse_insn): Likewise.
18187 * cselib.c (cselib_lookup_1): Likewise.
18188 * dojump.c (do_jump_1): Likewise.
18189 (do_compare_rtx_and_jump): Likewise.
18190 * dse.c (get_call_args): Likewise.
18191 * dwarf2out.c (rtl_for_decl_init): Likewise.
18192 (native_encode_initializer): Likewise.
18193 * expmed.c (emit_store_flag_1): Likewise.
18194 (emit_store_flag): Likewise.
18195 * expr.c (convert_modes): Likewise.
18196 (store_field): Likewise.
18197 (expand_expr_real_1): Likewise.
18198 * fold-const.c (fold_read_from_constant_string): Likewise.
18199 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18200 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18201 * optabs.c (expand_binop): Likewise.
18202 (expand_unop): Likewise.
18203 (expand_abs_nojump): Likewise.
18204 (expand_one_cmpl_abs_nojump): Likewise.
18205 * simplify-rtx.c (mode_signbit_p): Likewise.
18206 (val_signbit_p): Likewise.
18207 (val_signbit_known_set_p): Likewise.
18208 (val_signbit_known_clear_p): Likewise.
18209 (simplify_relational_operation_1): Likewise.
18210 * tree.c (vector_type_mode): Likewise.
18212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18213 Alan Hayward <alan.hayward@arm.com>
18214 David Sherwood <david.sherwood@arm.com>
18216 * machmode.h (smallest_mode_for_size): Fix formatting.
18217 (smallest_int_mode_for_size): New function.
18218 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18219 instead of smallest_mode_for_size.
18220 * combine.c (make_extraction): Likewise.
18221 * config/arc/arc.c (arc_expand_movmem): Likewise.
18222 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18223 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18224 * config/s390/s390.c (s390_expand_insv): Likewise.
18225 * config/sparc/sparc.c (assign_int_registers): Likewise.
18226 * config/spu/spu.c (spu_function_value): Likewise.
18227 (spu_function_arg): Likewise.
18228 * coverage.c (get_gcov_type): Likewise.
18229 (get_gcov_unsigned_t): Likewise.
18230 * dse.c (find_shift_sequence): Likewise.
18231 * expmed.c (store_bit_field_1): Likewise.
18232 * expr.c (convert_move): Likewise.
18233 (store_field): Likewise.
18234 * internal-fn.c (expand_arith_overflow): Likewise.
18235 * optabs-query.c (get_best_extraction_insn): Likewise.
18236 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18237 * stor-layout.c (layout_type): Likewise.
18238 (initialize_sizetypes): Likewise.
18239 * targhooks.c (default_get_mask_mode): Likewise.
18240 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18243 Alan Hayward <alan.hayward@arm.com>
18244 David Sherwood <david.sherwood@arm.com>
18246 * machmode.h (opt_mode::else_blk): New function.
18247 (int_mode_for_mode): Declare.
18248 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18249 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18251 * cfgexpand.c (expand_debug_expr): Likewise.
18252 * combine.c (gen_lowpart_or_truncate): Likewise.
18253 (gen_lowpart_for_combine): Likewise.
18254 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18255 * config/avr/avr.c (avr_to_int_mode): Likewise.
18256 (avr_out_plus_1): Likewise.
18257 (avr_out_plus): Likewise.
18258 (avr_out_round): Likewise.
18259 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18260 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18261 (s390_expand_vcond): Likewise.
18262 * config/spu/spu.c (spu_split_immediate): Likewise.
18263 (spu_expand_mov): Likewise.
18264 * dse.c (get_stored_val): Likewise.
18265 * expmed.c (store_bit_field_1): Likewise.
18266 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18268 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18269 (extract_low_bits): Likewise.
18270 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18271 handling rather than repeating the check.
18272 (emit_group_store): Likewise.
18273 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18274 * optabs.c (expand_absneg_bit): Likewise.
18275 (expand_copysign_absneg): Likewise.
18276 (expand_copysign_bit): Likewise.
18277 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18278 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18279 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18280 * var-tracking.c (prepare_call_arguments): Likewise.
18281 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18282 int_mode_for_mode instead of mode_for_size.
18283 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18286 Alan Hayward <alan.hayward@arm.com>
18287 David Sherwood <david.sherwood@arm.com>
18289 * machmode.h (int_mode_for_size): New function.
18290 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18291 instead of mode_for_size.
18292 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18294 * combine.c (expand_field_assignment): Use int_mode_for_size
18295 instead of mode_for_size.
18296 (make_extraction): Likewise.
18297 (simplify_shift_const_1): Likewise.
18298 (simplify_comparison): Likewise.
18299 * dojump.c (do_jump): Likewise.
18300 * dwarf2out.c (mem_loc_descriptor): Likewise.
18301 * emit-rtl.c (init_derived_machine_modes): Likewise.
18302 * expmed.c (flip_storage_order): Likewise.
18303 (convert_extracted_bit_field): Likewise.
18304 * expr.c (copy_blkmode_from_reg): Likewise.
18305 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18306 * internal-fn.c (expand_mul_overflow): Likewise.
18307 * lower-subreg.c (simple_move): Likewise.
18308 * optabs-libfuncs.c (init_optabs): Likewise.
18309 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18310 * tree.c (vector_type_mode): Likewise.
18311 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18312 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18313 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18314 * tree-vect-stmts.c (vectorizable_load): Likewise.
18315 (vectorizable_store): Likewise.
18317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18318 Alan Hayward <alan.hayward@arm.com>
18319 David Sherwood <david.sherwood@arm.com>
18321 * coretypes.h (pod_mode): New type.
18322 (scalar_int_mode_pod): New typedef.
18323 * machmode.h (pod_mode): New class.
18324 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18325 * genmodes.c (emit_mode_int_n): Update accordingly.
18326 * lower-subreg.h (target_lower_subreg): Change type to
18327 scalar_int_mode_pod.
18328 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18329 scalar_int_mode_pod.
18331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18332 Alan Hayward <alan.hayward@arm.com>
18333 David Sherwood <david.sherwood@arm.com>
18335 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18336 machine_mode to scalar_int_mode.
18337 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18338 (rs6000_option_override_internal): Remove cast to int.
18339 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18340 machine_mode to scalar_int_mode.
18341 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18342 (rs6000_option_override_internal): Remove cast to int.
18343 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18344 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18346 * config/s390/s390.c (s390_expand_builtin): Likewise.
18347 * coretypes.h (scalar_int_mode): New type.
18348 (opt_scalar_int_mode): New typedef.
18349 * machmode.h (scalar_int_mode): New class.
18350 (scalar_int_mode::includes_p): New function.
18351 (byte_mode): Change type to scalar_int_mode.
18352 (word_mode): Likewise.
18353 (ptr_mode): Likewise.
18354 * emit-rtl.c (byte_mode): Likewise.
18355 (word_mode): Likewise.
18356 (ptr_mode): Likewise.
18357 (init_derived_machine_modes): Update accordingly.
18358 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18359 and MODE_PARTIAL_INT.
18360 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18361 opt_scalar_int_mode.
18363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18364 Alan Hayward <alan.hayward@arm.com>
18365 David Sherwood <david.sherwood@arm.com>
18367 * target.def (libgcc_floating_mode_supported_p): Take a
18369 * doc/tm.texi: Regenerate.
18370 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18372 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18373 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18376 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18377 Alan Hayward <alan.hayward@arm.com>
18378 David Sherwood <david.sherwood@arm.com>
18380 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18381 * doc/tm.texi: Regenerate.
18382 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18383 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18384 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18385 * targhooks.h (default_floatn_mode): Likewise.
18386 * targhooks.c (default_floatn_mode): Likewise.
18387 * tree.c (build_common_tree_nodes): Update accordingly.
18389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18390 Alan Hayward <alan.hayward@arm.com>
18391 David Sherwood <david.sherwood@arm.com>
18393 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18394 (mode_iterator::iterate_p): Likewise.
18395 (mode_iterator::get_wider): Likewise.
18396 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18399 Alan Hayward <alan.hayward@arm.com>
18400 David Sherwood <david.sherwood@arm.com>
18402 * coretypes.h (opt_scalar_float_mode): New typedef.
18403 * machmode.h (float_mode_for_size): New function.
18404 * emit-rtl.c (double_mode): Delete.
18405 (init_emit_once): Use float_mode_for_size.
18406 * stor-layout.c (layout_type): Likewise.
18407 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18410 Alan Hayward <alan.hayward@arm.com>
18411 David Sherwood <david.sherwood@arm.com>
18413 * output.h (assemble_real): Take a scalar_float_mode.
18414 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18415 * config/arm/arm.md (consttable_4): Likewise.
18416 (consttable_8): Likewise.
18417 (consttable_16): Likewise.
18418 * config/mips/mips.md (consttable_float): Likewise.
18419 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18420 * varasm.c (assemble_real): Take a scalar_float_mode.
18421 (output_constant_pool_2): Update accordingly.
18422 (output_constant): Likewise.
18424 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18425 Alan Hayward <alan.hayward@arm.com>
18426 David Sherwood <david.sherwood@arm.com>
18428 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18429 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18430 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18431 (native_encode_real): Likewise.
18432 (native_interpret_real): Likewise.
18433 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18434 * tree-vrp.c (simplify_float_conversion_using_ranges): 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 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18441 <scalar_float_mode>. Simplify.
18442 (gen_extend_conv_libfunc): Likewise.
18444 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18445 Alan Hayward <alan.hayward@arm.com>
18446 David Sherwood <david.sherwood@arm.com>
18448 * coretypes.h (scalar_float_mode): New type.
18449 * machmode.h (mode_traits::from_int): Use machine_mode if
18450 USE_ENUM_MODES is defined.
18451 (is_a): New function.
18453 (dyn_cast): Likewise.
18454 (scalar_float_mode): New class.
18455 (scalar_float_mode::includes_p): New function.
18456 (is_float_mode): Likewise.
18457 * gdbhooks.py (MachineModePrinter): New class.
18458 (build_pretty_printer): Use it for scalar_float_mode.
18459 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18460 (format_helper::format_helper): Turn into a template.
18461 * genmodes.c (get_mode_class): New function.
18462 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18463 or machine_mode if none.
18464 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18465 as_a <scalar_float_mode>.
18466 * dwarf2out.c (mem_loc_descriptor): Likewise.
18467 (insert_float): Likewise.
18468 (add_const_value_attribute): Likewise.
18469 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18470 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18471 (expand_unop): Update accordingly.
18472 (expand_abs_nojump): Likewise.
18473 (expand_copysign_absneg): Take a scalar_float_mode.
18474 (expand_copysign_bit): Likewise.
18475 (expand_copysign): Update accordingly.
18477 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18478 Alan Hayward <alan.hayward@arm.com>
18479 David Sherwood <david.sherwood@arm.com>
18481 * coretypes.h (opt_mode): New class.
18482 * machmode.h (opt_mode): Likewise.
18483 (opt_mode::else_void): New function.
18484 (opt_mode::require): Likewise.
18485 (opt_mode::exists): Likewise.
18486 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18487 (GET_MODE_2XWIDER_MODE): Likewise.
18488 (mode_iterator::get_wider): Update accordingly.
18489 (mode_iterator::get_2xwider): Likewise.
18490 (mode_iterator::get_known_wider): Likewise, turning into a template.
18491 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18492 forcing a wider mode to exist.
18493 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18494 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18495 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18496 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18497 * lower-subreg.c (init_lower_subreg): Likewise.
18498 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18499 on the final iteration.
18500 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18501 a wider mode exists before asking for a move pattern.
18502 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18503 forcing a wider mode to exist.
18504 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18505 returning false if no such mode exists.
18506 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18507 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18508 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18509 Avoid checking for a MODE_INT if we already know the mode is not a
18511 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18512 forcing a wider mode to exist.
18513 (expmed_mult_highpart_optab): Likewise.
18514 (expmed_mult_highpart): Likewise.
18515 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18517 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18518 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18519 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18520 * internal-fn.c (expand_mul_overflow): Update use of
18521 GET_MODE_2XWIDER_MODE.
18522 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18523 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18524 GET_MODE_WIDER_MODE.
18525 (convert_plusminus_to_widen): Likewise.
18526 * tree-switch-conversion.c (array_value_type): Likewise.
18527 * var-tracking.c (emit_note_insn_var_location): Likewise.
18528 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18529 Return false inside rather than outside the loop if no wider mode
18531 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18532 and GET_MODE_2XWIDER_MODE
18533 (can_compare_p): Use else_void.
18534 * gdbhooks.py (OptMachineModePrinter): New class.
18535 (build_pretty_printer): Use it for opt_mode.
18537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18538 Alan Hayward <alan.hayward@arm.com>
18539 David Sherwood <david.sherwood@arm.com>
18541 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18542 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18544 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18545 Alan Hayward <alan.hayward@arm.com>
18546 David Sherwood <david.sherwood@arm.com>
18548 * machmode.h (mode_traits): New structure.
18549 (get_narrowest_mode): New function.
18550 (mode_iterator::start): Likewise.
18551 (mode_iterator::iterate_p): Likewise.
18552 (mode_iterator::get_wider): Likewise.
18553 (mode_iterator::get_known_wider): Likewise.
18554 (mode_iterator::get_2xwider): Likewise.
18555 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18556 (FOR_EACH_MODE): Likewise.
18557 (FOR_EACH_MODE_FROM): Likewise.
18558 (FOR_EACH_MODE_UNTIL): Likewise.
18559 (FOR_EACH_WIDER_MODE): Likewise.
18560 (FOR_EACH_2XWIDER_MODE): Likewise.
18561 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18562 * combine.c (simplify_comparison): Likewise
18563 * config/i386/i386.c (type_natural_mode): Likewise.
18564 * cse.c (cse_insn): Likewise.
18565 * dse.c (find_shift_sequence): Likewise.
18566 * emit-rtl.c (init_derived_machine_modes): Likewise.
18567 (init_emit_once): Likewise.
18568 * explow.c (hard_function_value): Likewise.
18569 * expmed.c (extract_fixed_bit_field_1): Likewise.
18570 (extract_bit_field_1): Likewise.
18571 (expand_divmod): Likewise.
18572 (emit_store_flag_1): Likewise.
18573 * expr.c (init_expr_target): Likewise.
18574 (convert_move): Likewise.
18575 (alignment_for_piecewise_move): Likewise.
18576 (widest_int_mode_for_size): Likewise.
18577 (emit_block_move_via_movmem): Likewise.
18578 (copy_blkmode_to_reg): Likewise.
18579 (set_storage_via_setmem): Likewise.
18580 (compress_float_constant): Likewise.
18581 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18582 * optabs-query.c (get_best_extraction_insn): Likewise.
18583 * optabs.c (expand_binop): Likewise.
18584 (expand_twoval_unop): Likewise.
18585 (expand_twoval_binop): Likewise.
18586 (widen_leading): Likewise.
18587 (widen_bswap): Likewise.
18588 (expand_parity): Likewise.
18589 (expand_unop): Likewise.
18590 (prepare_cmp_insn): Likewise.
18591 (prepare_float_lib_cmp): Likewise.
18592 (expand_float): Likewise.
18593 (expand_fix): Likewise.
18594 (expand_sfix_optab): Likewise.
18595 * postreload.c (move2add_use_add2_insn): Likewise.
18596 * reg-stack.c (reg_to_stack): Likewise.
18597 * reginfo.c (choose_hard_reg_mode): Likewise.
18598 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18599 * stor-layout.c (mode_for_size): Likewise.
18600 (smallest_mode_for_size): Likewise.
18601 (mode_for_vector): Likewise.
18602 (finish_bitfield_representative): Likewise.
18603 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18604 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18605 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18606 * var-tracking.c (prepare_call_arguments): Likewise.
18608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18609 Alan Hayward <alan.hayward@arm.com>
18610 David Sherwood <david.sherwood@arm.com>
18612 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18613 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18614 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18615 * machmode.h (mode_size): Move earlier in file.
18616 (mode_precision): Likewise.
18617 (mode_inner): Likewise.
18618 (mode_nunits): Likewise.
18619 (mode_unit_size): Likewise.
18620 (unit_unit_precision): Likewise.
18621 (mode_wider): Likewise.
18622 (mode_2xwider): Likewise.
18623 (machine_mode): New class.
18624 (mode_to_bytes): New function.
18625 (mode_to_bits): Likewise.
18626 (mode_to_precision): Likewise.
18627 (mode_to_inner): Likewise.
18628 (mode_to_unit_size): Likewise.
18629 (mode_to_unit_precision): Likewise.
18630 (mode_to_nunits): Likewise.
18631 (GET_MODE_SIZE): Use mode_to_bytes.
18632 (GET_MODE_BITSIZE): Use mode_to_bits.
18633 (GET_MODE_PRECISION): Use mode_to_precision.
18634 (GET_MODE_INNER): Use mode_to_inner.
18635 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18636 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18637 (GET_MODE_NUNITS): Use mode_to_nunits.
18638 * system.h (ALWAYS_INLINE): New macro.
18639 * config/powerpcspe/powerpcspe-c.c
18640 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18641 int for arg1_mode and arg2_mode.
18643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18644 Alan Hayward <alan.hayward@arm.com>
18645 David Sherwood <david.sherwood@arm.com>
18647 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18648 Prefix mode names with E_ in case statements.
18649 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18650 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18651 (aarch64_split_simd_move): Likewise.
18652 (aarch64_gen_storewb_pair): Likewise.
18653 (aarch64_gen_loadwb_pair): Likewise.
18654 (aarch64_gen_store_pair): Likewise.
18655 (aarch64_gen_load_pair): Likewise.
18656 (aarch64_get_condition_code_1): Likewise.
18657 (aarch64_constant_pool_reload_icode): Likewise.
18658 (get_rsqrte_type): Likewise.
18659 (get_rsqrts_type): Likewise.
18660 (get_recpe_type): Likewise.
18661 (get_recps_type): Likewise.
18662 (aarch64_gimplify_va_arg_expr): Likewise.
18663 (aarch64_simd_container_mode): Likewise.
18664 (aarch64_emit_load_exclusive): Likewise.
18665 (aarch64_emit_store_exclusive): Likewise.
18666 (aarch64_expand_compare_and_swap): Likewise.
18667 (aarch64_gen_atomic_cas): Likewise.
18668 (aarch64_emit_bic): Likewise.
18669 (aarch64_emit_atomic_swap): Likewise.
18670 (aarch64_emit_atomic_load_op): Likewise.
18671 (aarch64_evpc_trn): Likewise.
18672 (aarch64_evpc_uzp): Likewise.
18673 (aarch64_evpc_zip): Likewise.
18674 (aarch64_evpc_ext): Likewise.
18675 (aarch64_evpc_rev): Likewise.
18676 (aarch64_evpc_dup): Likewise.
18677 (aarch64_gen_ccmp_first): Likewise.
18678 (aarch64_gen_ccmp_next): Likewise.
18679 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18680 (alpha_emit_xfloating_libcall): Likewise.
18681 (emit_insxl): Likewise.
18682 (alpha_arg_type): Likewise.
18683 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18684 (arc_preferred_simd_mode): Likewise.
18685 (arc_secondary_reload): Likewise.
18686 (get_arc_condition_code): Likewise.
18687 (arc_print_operand): Likewise.
18688 (arc_legitimate_constant_p): Likewise.
18689 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18690 * config/arc/arc.md (casesi_load): Likewise.
18691 (casesi_compact_jump): Likewise.
18692 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18693 (cc_use_register): Likewise.
18694 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18695 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18696 (arm_init_iwmmxt_builtins): Likewise.
18697 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18698 (neon_expand_vector_init): Likewise.
18699 (arm_attr_length_move_neon): Likewise.
18700 (maybe_get_arm_condition_code): Likewise.
18701 (arm_emit_vector_const): Likewise.
18702 (arm_preferred_simd_mode): Likewise.
18703 (arm_output_iwmmxt_tinsr): Likewise.
18704 (thumb1_output_casesi): Likewise.
18705 (thumb2_output_casesi): Likewise.
18706 (arm_emit_load_exclusive): Likewise.
18707 (arm_emit_store_exclusive): Likewise.
18708 (arm_expand_compare_and_swap): Likewise.
18709 (arm_evpc_neon_vuzp): Likewise.
18710 (arm_evpc_neon_vzip): Likewise.
18711 (arm_evpc_neon_vrev): Likewise.
18712 (arm_evpc_neon_vtrn): Likewise.
18713 (arm_evpc_neon_vext): Likewise.
18714 (arm_validize_comparison): Likewise.
18715 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18716 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18717 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18718 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18719 (c6x_preferred_simd_mode): Likewise.
18720 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18721 (epiphany_rtx_costs): Likewise.
18722 * config/epiphany/predicates.md (proper_comparison_operator):
18724 * config/frv/frv.c (condexec_memory_operand): Likewise.
18725 (frv_emit_move): Likewise.
18726 (output_move_single): Likewise.
18727 (output_condmove_single): Likewise.
18728 (frv_hard_regno_mode_ok): Likewise.
18729 (frv_matching_accg_mode): Likewise.
18730 * config/h8300/h8300.c (split_adds_subs): Likewise.
18731 (h8300_rtx_costs): Likewise.
18732 (h8300_print_operand): Likewise.
18733 (compute_mov_length): Likewise.
18734 (output_logical_op): Likewise.
18735 (compute_logical_op_length): Likewise.
18736 (compute_logical_op_cc): Likewise.
18737 (h8300_shift_needs_scratch_p): Likewise.
18738 (output_a_shift): Likewise.
18739 (compute_a_shift_length): Likewise.
18740 (compute_a_shift_cc): Likewise.
18741 (expand_a_rotate): Likewise.
18742 (output_a_rotate): Likewise.
18743 * config/i386/i386.c (classify_argument): Likewise.
18744 (function_arg_advance_32): Likewise.
18745 (function_arg_32): Likewise.
18746 (function_arg_64): Likewise.
18747 (function_value_64): Likewise.
18748 (ix86_gimplify_va_arg): Likewise.
18749 (ix86_legitimate_constant_p): Likewise.
18750 (put_condition_code): Likewise.
18751 (split_double_mode): Likewise.
18752 (ix86_avx256_split_vector_move_misalign): Likewise.
18753 (ix86_expand_vector_logical_operator): Likewise.
18754 (ix86_split_idivmod): Likewise.
18755 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18756 (ix86_build_const_vector): Likewise.
18757 (ix86_build_signbit_mask): Likewise.
18758 (ix86_match_ccmode): Likewise.
18759 (ix86_cc_modes_compatible): Likewise.
18760 (ix86_expand_branch): Likewise.
18761 (ix86_expand_sse_cmp): Likewise.
18762 (ix86_expand_sse_movcc): Likewise.
18763 (ix86_expand_int_sse_cmp): Likewise.
18764 (ix86_expand_vec_perm_vpermi2): Likewise.
18765 (ix86_expand_vec_perm): Likewise.
18766 (ix86_expand_sse_unpack): Likewise.
18767 (ix86_expand_int_addcc): Likewise.
18768 (ix86_split_to_parts): Likewise.
18769 (ix86_vectorize_builtin_gather): Likewise.
18770 (ix86_vectorize_builtin_scatter): Likewise.
18771 (avx_vpermilp_parallel): Likewise.
18772 (inline_memory_move_cost): Likewise.
18773 (ix86_tieable_integer_mode_p): Likewise.
18774 (x86_maybe_negate_const_int): Likewise.
18775 (ix86_expand_vector_init_duplicate): Likewise.
18776 (ix86_expand_vector_init_one_nonzero): Likewise.
18777 (ix86_expand_vector_init_one_var): Likewise.
18778 (ix86_expand_vector_init_concat): Likewise.
18779 (ix86_expand_vector_init_interleave): Likewise.
18780 (ix86_expand_vector_init_general): Likewise.
18781 (ix86_expand_vector_set): Likewise.
18782 (ix86_expand_vector_extract): Likewise.
18783 (emit_reduc_half): Likewise.
18784 (ix86_emit_i387_round): Likewise.
18785 (ix86_mangle_type): Likewise.
18786 (ix86_expand_round_sse4): Likewise.
18787 (expand_vec_perm_blend): Likewise.
18788 (canonicalize_vector_int_perm): Likewise.
18789 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18790 (expand_vec_perm_1): Likewise.
18791 (expand_vec_perm_interleave3): Likewise.
18792 (expand_vec_perm_even_odd_pack): Likewise.
18793 (expand_vec_perm_even_odd_1): Likewise.
18794 (expand_vec_perm_broadcast_1): Likewise.
18795 (ix86_vectorize_vec_perm_const_ok): Likewise.
18796 (ix86_expand_vecop_qihi): Likewise.
18797 (ix86_expand_mul_widen_hilo): Likewise.
18798 (ix86_expand_sse2_abs): Likewise.
18799 (ix86_expand_pextr): Likewise.
18800 (ix86_expand_pinsr): Likewise.
18801 (ix86_preferred_simd_mode): Likewise.
18802 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18803 * config/i386/sse.md (*andnot<mode>3): Likewise.
18804 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18805 (*<code><mode>3): Likewise.
18806 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18807 (ia64_expand_atomic_op): Likewise.
18808 (ia64_arg_type): Likewise.
18809 (ia64_mode_to_int): Likewise.
18810 (ia64_scalar_mode_supported_p): Likewise.
18811 (ia64_vector_mode_supported_p): Likewise.
18812 (expand_vec_perm_broadcast): Likewise.
18813 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18814 (iq2000_function_arg_advance): Likewise.
18815 (iq2000_function_arg): Likewise.
18816 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18817 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18818 (m68k_libcall_value): Likewise.
18819 (m68k_function_value): Likewise.
18820 (sched_attr_op_type): Likewise.
18821 * config/mcore/mcore.c (mcore_output_move): Likewise.
18822 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18824 (microblaze_function_arg): Likewise.
18825 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18826 (mips_print_operand): Likewise.
18827 (mips_mode_ok_for_mov_fmt_p): Likewise.
18828 (mips_vector_mode_supported_p): Likewise.
18829 (mips_preferred_simd_mode): Likewise.
18830 (mips_expand_vpc_loongson_even_odd): Likewise.
18831 (mips_expand_vec_unpack): Likewise.
18832 (mips_expand_vi_broadcast): Likewise.
18833 (mips_expand_vector_init): Likewise.
18834 (mips_expand_vec_reduc): Likewise.
18835 (mips_expand_msa_cmp): Likewise.
18836 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18837 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18838 (cc_flags_for_mode): Likewise.
18839 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18840 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18841 (nds32_output_casesi_pc_relative): Likewise.
18842 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18843 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18844 (nvptx_gen_unpack): Likewise.
18845 (nvptx_gen_pack): Likewise.
18846 (nvptx_gen_shuffle): Likewise.
18847 (nvptx_gen_wcast): Likewise.
18848 (nvptx_preferred_simd_mode): Likewise.
18849 * config/pa/pa.c (pa_secondary_reload): Likewise.
18850 * config/pa/predicates.md (base14_operand): Likewise.
18851 * config/powerpcspe/powerpcspe-c.c
18852 (altivec_resolve_overloaded_builtin): Likewise.
18853 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18855 (rs6000_preferred_simd_mode): Likewise.
18856 (output_vec_const_move): Likewise.
18857 (rs6000_expand_vector_extract): Likewise.
18858 (rs6000_split_vec_extract_var): Likewise.
18859 (reg_offset_addressing_ok_p): Likewise.
18860 (rs6000_legitimate_offset_address_p): Likewise.
18861 (rs6000_legitimize_address): Likewise.
18862 (rs6000_emit_set_const): Likewise.
18863 (rs6000_const_vec): Likewise.
18864 (rs6000_emit_move): Likewise.
18865 (spe_build_register_parallel): Likewise.
18866 (rs6000_darwin64_record_arg_recurse): Likewise.
18867 (swap_selector_for_mode): Likewise.
18868 (spe_init_builtins): Likewise.
18869 (paired_init_builtins): Likewise.
18870 (altivec_init_builtins): Likewise.
18871 (do_load_for_compare): Likewise.
18872 (rs6000_generate_compare): Likewise.
18873 (rs6000_expand_float128_convert): Likewise.
18874 (emit_load_locked): Likewise.
18875 (emit_store_conditional): Likewise.
18876 (rs6000_output_function_epilogue): Likewise.
18877 (rs6000_handle_altivec_attribute): Likewise.
18878 (rs6000_function_value): Likewise.
18879 (emit_fusion_gpr_load): Likewise.
18880 (emit_fusion_p9_load): Likewise.
18881 (emit_fusion_p9_store): Likewise.
18882 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
18883 (fusion_gpr_mem_load): Likewise.
18884 (fusion_addis_mem_combo_load): Likewise.
18885 (fusion_addis_mem_combo_store): Likewise.
18886 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
18887 (fusion_gpr_mem_load): Likewise.
18888 (fusion_addis_mem_combo_load): Likewise.
18889 (fusion_addis_mem_combo_store): Likewise.
18890 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18892 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
18893 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18894 (rs6000_preferred_simd_mode): Likewise.
18895 (output_vec_const_move): Likewise.
18896 (rs6000_expand_vector_extract): Likewise.
18897 (rs6000_split_vec_extract_var): Likewise.
18898 (reg_offset_addressing_ok_p): Likewise.
18899 (rs6000_legitimate_offset_address_p): Likewise.
18900 (rs6000_legitimize_address): Likewise.
18901 (rs6000_emit_set_const): Likewise.
18902 (rs6000_const_vec): Likewise.
18903 (rs6000_emit_move): Likewise.
18904 (rs6000_darwin64_record_arg_recurse): Likewise.
18905 (swap_selector_for_mode): Likewise.
18906 (paired_init_builtins): Likewise.
18907 (altivec_init_builtins): Likewise.
18908 (rs6000_expand_float128_convert): Likewise.
18909 (emit_load_locked): Likewise.
18910 (emit_store_conditional): Likewise.
18911 (rs6000_output_function_epilogue): Likewise.
18912 (rs6000_handle_altivec_attribute): Likewise.
18913 (rs6000_function_value): Likewise.
18914 (emit_fusion_gpr_load): Likewise.
18915 (emit_fusion_p9_load): Likewise.
18916 (emit_fusion_p9_store): Likewise.
18917 * config/rx/rx.c (rx_gen_move_template): Likewise.
18918 (flags_from_mode): Likewise.
18919 * config/s390/predicates.md (s390_alc_comparison): Likewise.
18920 (s390_slb_comparison): Likewise.
18921 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
18922 (s390_vector_mode_supported_p): Likewise.
18923 (s390_cc_modes_compatible): Likewise.
18924 (s390_match_ccmode_set): Likewise.
18925 (s390_canonicalize_comparison): Likewise.
18926 (s390_emit_compare_and_swap): Likewise.
18927 (s390_branch_condition_mask): Likewise.
18928 (s390_rtx_costs): Likewise.
18929 (s390_secondary_reload): Likewise.
18930 (__SECONDARY_RELOAD_CASE): Likewise.
18931 (s390_expand_cs): Likewise.
18932 (s390_preferred_simd_mode): Likewise.
18933 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
18934 * config/sh/sh.c (sh_print_operand): Likewise.
18935 (dump_table): Likewise.
18936 (sh_secondary_reload): Likewise.
18937 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18938 * config/sh/sh.md (casesi_worker_1): Likewise.
18939 (casesi_worker_2): Likewise.
18940 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
18941 (fcc_comparison_operator): Likewise.
18942 * config/sparc/sparc.c (sparc_expand_move): Likewise.
18943 (emit_soft_tfmode_cvt): Likewise.
18944 (sparc_preferred_simd_mode): Likewise.
18945 (output_cbranch): Likewise.
18946 (sparc_print_operand): Likewise.
18947 (sparc_expand_vec_perm_bmask): Likewise.
18948 (vector_init_bshuffle): Likewise.
18949 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18950 (spu_vector_mode_supported_p): Likewise.
18951 (spu_expand_insv): Likewise.
18952 (spu_emit_branch_or_set): Likewise.
18953 (spu_handle_vector_attribute): Likewise.
18954 (spu_builtin_splats): Likewise.
18955 (spu_builtin_extract): Likewise.
18956 (spu_builtin_promote): Likewise.
18957 (spu_expand_sign_extend): Likewise.
18958 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18959 (tilegx_simd_int): Likewise.
18960 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
18961 (tilepro_simd_int): Likewise.
18962 * config/v850/v850.c (const_double_split): Likewise.
18963 (v850_print_operand): Likewise.
18964 (ep_memory_offset): Likewise.
18965 * config/vax/vax.c (vax_rtx_costs): Likewise.
18966 (vax_output_int_move): Likewise.
18967 (vax_output_int_add): Likewise.
18968 (vax_output_int_subtract): Likewise.
18969 * config/visium/predicates.md (visium_branch_operator): Likewise.
18970 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
18971 (visium_print_operand_address): Likewise.
18972 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18973 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
18974 (xtensa_expand_conditional_branch): Likewise.
18975 (xtensa_copy_incoming_a7): Likewise.
18976 (xtensa_output_literal): Likewise.
18977 * dfp.c (decimal_real_maxval): Likewise.
18978 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18981 Alan Hayward <alan.hayward@arm.com>
18982 David Sherwood <david.sherwood@arm.com>
18984 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
18985 (mode_nunits_inline): Likewise.
18986 (mode_inner_inline): Likewise.
18987 (mode_unit_size_inline): Likewise.
18988 (mode_unit_precision_inline): Likewise.
18989 (emit_insn_modes_h): Likewise. Also emit a #define of the
18991 (emit_mode_wider): Add an E_ prefix to mode names.
18992 (emit_mode_complex): Likewise.
18993 (emit_mode_inner): Likewise.
18994 (emit_mode_adjustments): Likewise.
18995 (emit_mode_int_n): Likewise.
18996 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
18997 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
18998 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
18999 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19000 (CRC32_BUILTIN, ENTRY): Likewise.
19001 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19002 (aarch64_pop_regs): Likewise.
19003 (aarch64_process_components): Likewise.
19004 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19005 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19006 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19007 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19008 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19009 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19011 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19012 (pcmpistr): Likewise.
19013 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19014 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19015 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19017 * config/rl78/rl78.c (mduc_regs): Likewise.
19018 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19019 (htm_expand_builtin): Likewise.
19020 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19021 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19022 * config/xtensa/xtensa.c (print_operand): Likewise.
19023 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19024 (NUM_MODE_VECTOR_INT): Likewise.
19025 * genoutput.c (null_operand): Likewise.
19026 (output_operand_data): Likewise.
19027 * genrecog.c (print_parameter_value): Likewise.
19028 * lra.c (debug_operand_data): Likewise.
19030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19032 * dwarf2out.c (scompare_loc_descriptor_wide)
19033 (scompare_loc_descriptor_narrow): New functions, split out from...
19034 (scompare_loc_descriptor): ...here.
19035 * expmed.c (emit_store_flag_int): New function, split out from...
19036 (emit_store_flag): ...here.
19038 2017-08-30 Richard Biener <rguenther@suse.de>
19040 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19041 (dwarf2out_early_finish): Move setting of AT_pubnames from
19042 early debug output to early finish.
19044 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19046 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19047 and -mdata-region to the assembler.
19049 2017-08-30 Richard Biener <rguenther@suse.de>
19051 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19053 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19054 (dwarf2out_early_global_decl): Only generate a DIE for the
19055 abstract origin if it doesn't already exist or is a declaration DIE.
19056 (resolve_addr): Do not add the linkage name twice when
19057 generating a stub DIE for the DW_TAG_GNU_call_site target.
19059 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19061 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19062 Use machine_mode rather than int for arg1_mode.
19064 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19067 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19068 that the second argument of the built-in functions to unpack
19069 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19070 switch statement instead a lot of if statements.
19071 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19072 Allow 64-bit values to be in Altivec registers as well as
19073 traditional floating point registers.
19074 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19076 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19078 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19079 MAX_REGS_PER_ADDRESS == 1.
19081 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19083 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19084 * config/i386/i386.c (ix86_option_override_internal): Simplify
19085 setting of opts->x_flag_entry.
19087 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19088 Jakub Jelinek <jakub@redhat.com>
19089 Richard Biener <rguenther@suse.de>
19091 PR tree-optimization/81503
19092 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19093 folded constant fits in the target type; reorder tests for clarity.
19095 2017-08-29 Martin Liska <mliska@suse.cz>
19097 * passes.def: Include pass_lower_switch.
19098 * stmt.c (dump_case_nodes): Remove and move to
19099 tree-switch-conversion.
19100 (case_values_threshold): Likewise.
19101 (expand_switch_as_decision_tree_p): Likewise.
19102 (emit_case_decision_tree): Likewise.
19103 (expand_case): Likewise.
19104 (balance_case_nodes): Likewise.
19105 (node_has_low_bound): Likewise.
19106 (node_has_high_bound): Likewise.
19107 (node_is_bounded): Likewise.
19108 (emit_case_nodes): Likewise.
19109 (struct simple_case_node): New struct.
19110 (add_case_node): Remove.
19111 (emit_case_dispatch_table): Use vector instead of case_list.
19112 (reset_out_edges_aux): Remove.
19113 (compute_cases_per_edge): Likewise.
19114 (expand_case): Build list of simple_case_node.
19115 (expand_sjlj_dispatch_table): Use it.
19116 * tree-switch-conversion.c (struct case_node): Moved from
19117 stmt.c and adjusted.
19118 (emit_case_nodes): Likewise.
19119 (node_has_low_bound): Likewise.
19120 (node_has_high_bound): Likewise.
19121 (node_is_bounded): Likewise.
19122 (case_values_threshold): Likewise.
19123 (reset_out_edges_aux): Likewise.
19124 (compute_cases_per_edge): Likewise.
19125 (add_case_node): Likewise.
19126 (dump_case_nodes): Likewise.
19127 (balance_case_nodes): Likewise.
19128 (expand_switch_as_decision_tree_p): Likewise.
19129 (emit_jump): Likewise.
19130 (emit_case_decision_tree): Likewise.
19131 (try_switch_expansion): Likewise.
19132 (do_jump_if_equal): Likewise.
19133 (emit_cmp_and_jump_insns): Likewise.
19134 (fix_phi_operands_for_edge): New function.
19135 (record_phi_operand_mapping): Likewise.
19136 (class pass_lower_switch): New pass.
19137 (pass_lower_switch::execute): New function.
19138 (make_pass_lower_switch): Likewise.
19139 (conditional_probability):
19140 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19141 * tree-pass.h: Add make_pass_lower_switch.
19143 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19146 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19149 2017-08-29 Richard Biener <rguenther@suse.de>
19151 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19152 we do not add a DW_AT_inline attribute twice.
19153 (gen_subprogram_die): Remove code setting DW_AT_inline on
19154 DECL_ABSTRACT_P nodes.
19156 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19158 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19159 calls to internal functions.
19160 (gimplify_modify_expr): Likewise.
19161 * tree-call-cdce.c (use_internal_fn): Likewise.
19162 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19163 (convert_to_divmod): Set the nothrow flag.
19164 * tree-if-conv.c (predicate_mem_writes): Likewise.
19165 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19166 (vectorizable_call): Likewise.
19167 (vectorizable_store): Likewise.
19168 (vectorizable_load): Likewise.
19169 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19170 (vect_recog_mask_conversion_pattern): Likewise.
19172 2017-08-29 Martin Liska <mliska@suse.cz>
19175 * gcc.c (driver_handle_option): Add new argument.
19176 * opts-common.c (handle_option): Pass
19177 target_option_override_hook.
19178 * opts-global.c (lang_handle_option): Add new option.
19179 (set_default_handlers): Add new argument.
19180 (decode_options): Likewise.
19181 * opts.c (target_handle_option): Likewise.
19182 (common_handle_option): Call target_option_override_hook.
19183 * opts.h (struct cl_option_handler_func): Add hook for
19184 target option override.
19185 (struct cl_option_handlers): Likewise.
19186 (set_default_handlers): Add new argument.
19187 (decode_options): Likewise.
19188 (common_handle_option): Likewise.
19189 (target_handle_option): Likewise.
19190 * toplev.c (toplev::main): Pass targetm.target_option.override
19193 2017-08-29 Richard Biener <rguenther@suse.de>
19194 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19196 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19197 life to the active subtree.
19199 2017-08-28 Jeff Law <law@redhat.com>
19201 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19202 derive_equivalences.
19203 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19204 Code moved into....
19205 (edge_info::derive_equivalences): New private member function
19207 * tree-ssa-dom.c (class edge_info): Changed from a struct
19208 to a class. Add ctor/dtor, methods and data members.
19209 (edge_info::edge_info): Renamed from allocate_edge_info.
19210 Initialize additional members.
19211 (edge_info::~edge_info): New.
19212 (free_dom_edge_info): Delete the edge info.
19213 (record_edge_info): Use new class & associated member functions.
19214 Tighten forms for testing for edge equivalences.
19215 (record_temporary_equivalences): Iterate over the simple
19216 equivalences rather than assuming there's only one per edge.
19217 (cprop_into_successor_phis): Iterate over the simple
19218 equivalences rather than assuming there's only one per edge.
19219 (optimize_stmt): Use operand_equal_p rather than pointer
19220 equality for mini-DSE code.
19222 2017-08-28 Nathan Sidwell <nathan@acm.org>
19224 * gcc.c (execute): Fold SIGPIPE handling into switch
19225 statement. Adjust internal error message.
19227 2017-08-28 Richard Biener <rguenther@suse.de>
19230 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19231 Do nothing for removed DIEs.
19233 2017-08-28 Richard Biener <rguenther@suse.de>
19235 PR tree-optimization/81977
19236 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19239 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19242 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19243 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19244 using targetm.gen_mem_thread_fence.
19246 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19249 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19250 predicate to register_operand. Reorder operands.
19251 (*btr<mode>): Ditto.
19252 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19253 (*btr<mode>_mask): Ditto.
19255 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19257 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19258 * config/rs6000/xmmintrin.h: New file.
19259 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19261 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19264 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19265 parameter and_insn and return it.
19266 (recombine_lvx_pattern): Insert a copy to ensure availability of
19267 the base register of the copied masking operation at the point of
19268 the instruction replacement.
19269 (recombine_stvx_pattern): Likewise.
19271 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19273 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19274 undocumented switches.
19275 (-mpower9-dform-vector): Likewise.
19276 (-mpower9-dform): Likewise.
19277 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19278 comments to delete references to -mpower9-dform* switches.
19279 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19280 Delete reference to -mpower9-dform* switches, test for
19281 -mpower9-vector instead.
19282 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19283 (OTHER_P9_VECTOR_MASKS): Likewise.
19284 (POWERPC_MASKS): Likewise.
19285 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19286 tests against -mpower9-dform* to -mpower9-vector. Delete code
19287 that checked for -mpower9-dform* consistancy with other options.
19288 Add test for -mpower9-misc to enable other power9 switches.
19289 (rs6000_init_hard_regno_mode_ok): Likewise.
19290 (rs6000_option_override_internal): Likewise.
19291 (rs6000_emit_prologue): Likewise.
19292 (rs6000_emit_epilogue): Likewise.
19293 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19294 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19295 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19297 (emit_fusion_p9_store): Likewise.
19298 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19299 resetting these macros if the assembler does not support ISA 3.0
19301 (TARGET_P9_DFORM_VECTOR): Likewise.
19302 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19303 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19305 2017-08-25 Alan Modra <amodra@gmail.com>
19307 PR rtl-optimization/81747
19308 * cse.c (cse_extended_basic_block): Don't attempt to record
19309 equivalences for degenerate conditional branches that branch
19310 to their fall-through.
19312 2017-08-24 Martin Sebor <msebor@redhat.com>
19314 PR middle-end/81908
19315 * gimple-fold.c (size_must_be_zero_p): New function.
19316 (gimple_fold_builtin_memory_op): Call it.
19318 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19320 * config/rs6000/mm_malloc.h: New file.
19322 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19324 PR tree-optimization/81913
19325 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19326 analysis when either IVs in condition can wrap.
19328 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19330 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19331 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19333 2017-08-24 Richard Biener <rguenther@suse.de>
19336 * targhooks.c (default_target_can_inline_p): Properly
19337 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19338 is present and always compare.
19339 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19340 infer -mfpmath=sse from TARGET_SSE_P.
19341 (ix86_can_inline_p): Properly use target_option_default_node when
19342 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19344 2017-08-24 Richard Biener <rguenther@suse.de>
19347 * dwarf2out.c (output_die): Handle flag_generate_offload like
19349 (output_comp_unit): Likewise.
19350 (gen_array_type_die): Likewise.
19351 (dwarf2out_early_finish): Likewise.
19352 (note_variable_value_in_expr): Likewise.
19353 (dwarf2out_finish): Likewise. Adjust assert.
19354 * cgraphunit.c (symbol_table::compile): Move setting of
19355 flag_generate_offload earlier ...
19356 (symbol_table::finalize_compilation_unit): ... here, before
19357 early debug finalization.
19359 2017-08-24 Richard Biener <rguenther@suse.de>
19361 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19362 and ipa-fnsummary.h.
19363 (ix86_can_inline_p): When ix86_fpmath flags do not match
19364 check whether the callee uses FP math at all.
19366 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19368 PR middle-end/81931
19369 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19370 instead of TYPE_PRECISION.
19372 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19373 Alan Hayward <alan.hayward@arm.com>
19374 David Sherwood <david.sherwood@arm.com>
19376 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19378 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19380 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19382 (ix86_frame::stack_realign_allocate): New field.
19383 (struct machine_frame_state): Modify comments.
19384 (machine_frame_state::sp_realigned_fp_end): New field.
19385 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19386 layout calculation.
19387 (sp_valid_at): Add assertion to assure no attempt to access invalid
19388 offset of a realigned stack.
19389 (fp_valid_at): Likewise.
19390 (choose_baseaddr): Modify comments.
19391 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19392 ix86_expand_prologue.
19393 (ix86_expand_prologue): Modify stack realignment and allocation.
19394 (ix86_expand_epilogue): Modify comments.
19395 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19396 avx2_runtime, avx512f, and avx512f_runtime.
19398 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19400 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19401 (mstackrealign): Do not init to -1.
19402 * config/i386/i386.c (ix86_option_override_internal):
19403 Check opts_set, not opts when setting default value of
19404 opts->x_ix86_force_align_arg_pointer.
19406 2017-08-23 Richard Biener <rguenther@suse.de>
19408 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19409 lang_hooks.decl_printable_name.
19410 * print-rtl-function.c (print_rtx_function): Likewise.
19411 * tree-pretty-print.c (dump_function_header): Likewise.
19413 2017-08-23 Richard Biener <rguenther@suse.de>
19416 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19417 -g0 at compile-time.
19419 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19421 PR middle-end/19706
19422 * doc/sourcebuild.texi (Other hardware attributes):
19425 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19427 PR middle-end/19706
19428 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19429 Fix single-use check.
19431 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19433 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19435 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19437 * doc/install.texi: Modify to add more details on running selected
19440 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19442 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19443 is combined with -mabi=ms.
19444 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19447 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19449 PR tree-optimization/81488
19450 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19451 and cached_basis fields.
19452 (MAX_SPREAD): New constant.
19453 (alloc_cand_and_find_basis): Initialize new fields.
19454 (clear_visited): New function.
19455 (create_phi_basis_1): Rename from create_phi_basis, set visited
19456 and cached_basis fields.
19457 (create_phi_basis): New wrapper function.
19458 (phi_add_costs_1): Rename from phi_add_costs, add spread
19459 parameter, set visited field, short-circuit when limits reached.
19460 (phi_add_costs): New wrapper function.
19461 (record_phi_increments_1): Rename from record_phi_increments, set
19463 (record_phi_increments): New wrapper function.
19464 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19465 (phi_incr_cost): New wrapper function.
19466 (all_phi_incrs_profitable_1): Rename from
19467 all_phi_incrs_profitable, set visited field.
19468 (all_phi_incrs_profitable): New wrapper function.
19470 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19471 Alan Hayward <alan.hayward@arm.com>
19472 David Sherwood <david.sherwood@arm.com>
19474 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19475 that takes the outer and inner modes.
19476 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19477 comparison as the canonical test for a paradoxical subreg.
19478 * combine.c (simplify_set): Use paradoxical_subreg_p.
19479 (make_extraction): Likewise.
19480 (force_to_mode): Likewise.
19481 (rtx_equal_for_field_assignment_p): Likewise.
19482 (gen_lowpart_for_combine): Likewise.
19483 (simplify_comparison): Likewise.
19484 * cse.c (equiv_constant): Likewise.
19485 * expmed.c (store_bit_field_1): Likewise.
19486 * final.c (alter_subreg): Likewise.
19487 * fwprop.c (propagate_rtx): Likewise.
19488 (forward_propagate_subreg): Likewise.
19489 * ira-conflicts.c (ira_build_conflicts): Likewise.
19490 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19491 * lra-constraints.c (curr_insn_transform): Likewise.
19492 (split_reg): Likewise.
19493 * lra-eliminations.c (move_plus_up): Likewise.
19494 (lra_eliminate_regs_1): Likewise.
19495 * recog.c (general_operand): Likewise.
19496 * ree.c (combine_reaching_defs): Likewise.
19497 * reload.c (push_reload): Likewise.
19498 (find_reloads): Likewise.
19499 * reload1.c (elimination_effects): Likewise.
19500 (compute_reload_subreg_offset): Likewise.
19501 (choose_reload_regs): Likewise.
19502 * rtlanal.c (subreg_lsb_1): Likewise.
19503 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19504 (simplify_subreg): Likewise.
19505 * var-tracking.c (track_loc_p): Likewise.
19506 * emit-rtl.c (byte_lowpart_offset): Likewise.
19507 (paradoxical_subreg_p): Delete out-of-line definition.
19509 2017-08-22 Jeff Law <law@redhat.com>
19511 PR tree-optimization/81741
19512 PR tree-optimization/71947
19513 * tree-ssa-dom.c: Include tree-inline.h.
19514 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19515 equivalences if one is more expensive to compute than the other.
19516 * tree-ssa-scopedtables.h (class const_or_copies): Make
19517 record_const_or_copy_raw method private.
19518 (class avail_exprs_stack): New method simplify_binary_operation.
19519 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19520 avail_exprs_stack::simplify_binary_operation as needed.
19521 (avail_exprs_stack::simplify_binary_operation): New function.
19523 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19525 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19526 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19527 (DOT_SYMBOLS): Likewise.
19528 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19529 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19530 (RS6000_ABI_NAME): Likewise.
19531 (TARGET_CMODEL): Likewise.
19532 (TOC_SECTION_ASM_OP): Likewise.
19533 (SET_CMODEL): New macro.
19534 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19536 2017-08-22 Richard Biener <rguenther@suse.de>
19538 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19539 to allow for free-lang-data replacements similar to verify_type_variant.
19541 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19542 Alan Hayward <alan.hayward@arm.com>
19543 David Sherwood <david.sherwood@arm.com>
19545 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19546 VOIDmode for the LABEL_REF.
19548 2017-08-22 Richard Biener <rguenther@suse.de>
19550 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19552 2017-08-22 Richard Biener <rguenther@suse.de>
19554 * common.opt (feliminate-dwarf2-dups): Ignore.
19555 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19556 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19557 same_die_p_wrap, compute_section_prefix,
19558 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19559 (comdat_symbol_id, comdat_symbol_number): Likewise.
19560 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19562 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19563 (output_die): Mark unreachable path unreachable.
19564 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19565 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19566 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19567 (dwarf2out_early_finish): Likewise.
19569 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19571 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19573 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19576 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19577 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19578 (avr_attribute_table) <io, io_low, address>: Initialize
19579 .decl_required with true.
19581 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19583 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19584 undocumented debugging options.
19585 (-mvsx-scalar-double): Likewise.
19586 (-mallow-df-permute): Likewise.
19587 (-mvectorize-builtins): Likewise.
19588 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19589 (rs6000_builtin_vectorized_function): Likewise.
19590 (rs6000_builtin_md_vectorized_function): Likewise.
19591 (rs6000_opt_vars): Likewise.
19593 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19596 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19597 (*btrq_imm): Rename from *btrq.
19598 (*btcq_imm): Rename from *btcq.
19599 (btsc): New code attribute.
19600 (*<btsc><mode>): New insn pattern.
19601 (*btr<mode>): Ditto.
19602 (*<btsc><mode>_mask): New insn_and_split pattern.
19603 (*btr<mode>_mask): Ditto.
19605 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19606 Alan Hayward <alan.hayward@arm.com>
19607 David Sherwood <david.sherwood@arm.com>
19609 * function.c (pad_below): Simplify padding calculation.
19611 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19612 Alan Hayward <alan.hayward@arm.com>
19613 David Sherwood <david.sherwood@arm.com>
19615 * target.def (function_prologue): Remove frame size argument.
19616 (function_epilogue): Likewise.
19617 * doc/tm.texi: Regenerate.
19618 * final.c (final_start_function): Update call to function_prologue.
19619 (final_end_function): Update call to function_epilogue.
19620 (default_function_pro_epilogue): Remove frame size argument.
19621 * output.h (default_function_pro_epilogue): Likewise.
19622 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19623 (arm_output_function_prologue): Likewise.
19624 * config/frv/frv.c (frv_function_prologue): Likewise.
19625 (frv_function_epilogue): Likewise.
19626 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19627 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19628 (ia64_output_function_epilogue): Likewise.
19629 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19630 (m32r_output_function_epilogue): Likewise.
19631 * config/microblaze/microblaze.c (microblaze_function_prologue)
19632 (microblaze_function_epilogue): Likewise.
19633 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19634 (mips_output_function_epilogue): Likewise.
19635 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19636 (mmix_target_asm_function_epilogue): Likewise.
19637 * config/msp430/msp430.c (msp430_start_function): Likewise.
19638 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19639 (nds32_asm_function_epilogue): Likewise.
19640 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19641 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19642 (pa_output_function_epilogue): Likewise.
19643 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19644 (rs6000_output_function_epilogue): Likewise.
19645 * config/rl78/rl78.c (rl78_start_function): Likewise.
19646 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19647 (rs6000_output_function_epilogue): Likewise.
19648 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19649 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19650 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19651 (sparc_asm_function_epilogue): Likewise.
19653 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19655 * tree.h (type_has_mode_precision_p): New function.
19656 * convert.c (convert_to_integer_1): Use it.
19657 * expr.c (expand_expr_real_2): Likewise.
19658 (expand_expr_real_1): Likewise.
19659 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19660 * match.pd: Likewise.
19661 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19662 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19663 * tree-tailcall.c (process_assignment): Likewise.
19664 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19665 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19666 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19667 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19668 (vectorizable_assignment): Likewise.
19669 (vectorizable_shift): Likewise.
19670 (vectorizable_operation): Likewise.
19671 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19673 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19675 * match.pd: Add pow (C, x) simplification.
19677 2017-08-21 Richard Biener <rguenther@suse.de>
19679 PR tree-optimization/81900
19680 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19681 for blocks with abnormal predecessors.
19682 (compute_antic): Do not set visited flag prematurely.
19684 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19687 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19689 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19691 * stor-layout.h (vector_type_mode): Move to...
19692 * tree.h (vector_type_mode): ...here.
19693 * stor-layout.c (vector_type_mode): Move to...
19694 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19696 2017-08-21 Richard Biener <rguenther@suse.de>
19698 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19699 register_external_die hooks.
19700 (debug_false_tree_charstarstar_uhwistar): Declare.
19701 (debug_nothing_tree_charstar_uhwi): Likewise.
19702 * debug.c (do_nothing_debug_hooks): Adjust.
19703 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19704 (debug_nothing_tree_charstar_uhwi): Likewise.
19705 * dbxout.c (dbx_debug_hooks): Adjust.
19706 (xcoff_debug_hooks): Likewise.
19707 * sdbout.c (sdb_debug_hooks): Likewise.
19708 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19709 * dwarf2out.c (macinfo_label_base): New global.
19710 (dwarf2out_register_external_die): New function for the
19711 register_external_die hook.
19712 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19713 (dwarf2_debug_hooks): Use them.
19714 (dwarf2_lineno_debug_hooks): Adjust.
19715 (struct die_struct): Add with_offset flag.
19716 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19717 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19718 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19719 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19720 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19721 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19722 defining section names for the early LTO debug variants.
19723 (reset_indirect_string): New helper.
19724 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19725 (print_dw_val): Add support for offsetted symbol references.
19726 (get_ultimate_context): Split out from is_cxx.
19727 (is_cxx): Use get_ultimate_context.
19728 (is_fortran): Add decl overload.
19729 (compute_comp_unit_symbol): Split out worker from
19730 compute_section_prefix.
19731 (compute_section_prefix): Call compute_comp_unit_symbol and
19732 set comdat_type_p here.
19733 (output_die): Skip DIE symbol output for the LTO added one.
19734 Handle DIE symbol references with offset.
19735 (output_comp_unit): Guard section name mangling properly.
19736 For LTO debug sections emit a symbol at the section beginning
19737 which we use to refer to its DIEs.
19738 (add_abstract_origin_attribute): For DIEs registered via
19739 dwarf2out_register_external_die directly refer to the early
19740 DIE rather than indirectly through the shadow one we created.
19741 Remove obsolete call to dwarf2out_abstract_function for
19742 non-function/block origins.
19743 (gen_array_type_die): When generating early LTO debug do
19744 not emit DW_AT_string_length.
19745 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19746 late when in LTO. As suggested place a gcc_unreachable for
19747 the DECL_ABSTRACT_P case.
19748 (gen_subprogram_die): Avoid another specification DIE
19749 for early built declarations/definitions for the late LTO case.
19750 (gen_variable_die): Add type references for late duplicated VLA dies
19752 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19753 we have the abstract instance already.
19754 (process_scope_var): Adjust decl DIE contexts in LTO which
19755 first puts them in limbo.
19756 (gen_decl_die): Do not generate type DIEs late apart from
19757 types for VLAs or for decls we do not yet have a DIE. Do not
19758 call dwarf2out_abstract_function late.
19759 (dwarf2out_early_global_decl): Make sure to create DIEs
19760 for abstract instances of a decl first.
19761 (dwarf2out_late_global_decl): Adjust comment.
19762 (output_macinfo_op): With multiple macro sections use
19763 macinfo_label_base to distinguish labels.
19764 (output_macinfo): Likewise. Update macinfo_label_base.
19765 Pass in the line info label.
19766 (note_variable_value_in_expr): When generating LTO resolve
19767 all variable values here by generating DIEs as needed.
19768 (init_sections_and_labels): Add early LTO debug flag parameter
19769 and generate different sections and names if set. Add generation
19770 counter for the labels so we can have multiple of them.
19771 (reset_dies): Helper to allow DIEs to be output multiple times.
19772 (dwarf2out_finish): When outputting DIEs to the fat part of an
19773 LTO object first reset DIEs.
19774 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19775 (modified_type_die): Check for decl_ultimate_origin being self
19777 (gen_type_die_with_usage): Likewise.
19778 (gen_typedef_die): Allow decl_ultimate_origin being self.
19779 (set_decl_abstract_flags): Remove.
19780 (set_block_abstract_flags): Likewise.
19781 (dwarf2out_abstract_function): Treat the early generated DIEs
19782 as the abstract copy and only add DW_AT_inline and
19783 DW_AT_artificial here and call set_decl_origin_self.
19784 If the DIE has an abstract origin don't do anything.
19785 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19786 if we have none yet (Go fails to build one, PR78628).
19787 (variably_modified_type_p): Prevent endless recursion for Ada
19788 cyclic pointer types.
19789 * lto-streamer-in.c: Include debug.h.
19790 (dref_queue): New global.
19791 (lto_read_tree_1): Stream in DIE references.
19792 (lto_input_tree): Register DIE references.
19793 (input_function): Stream DECL_DEBUG_ARGS.
19794 * lto-streamer-out.c: Include debug.h.
19795 (lto_write_tree_1): Output DIE references.
19796 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19797 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19798 (output_function): Stream DECL_DEBUG_ARGS.
19799 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19800 Stream DECL_ABSTRACT_ORIGIN.
19801 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19802 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19803 DECL_CONTEXT for file-scope decls.
19804 * lto-streamer.h (struct dref_entry): Declare.
19805 (dref_queue): Likewise.
19806 * cfgexpand.c (pass_expand::execute): Do not call the
19807 outlining_inline_function hook here.
19808 * lto-wrapper.c (debug_obj): New global.
19809 (tool_cleanup): Unlink it if required.
19810 (debug_objcopy): New function.
19811 (run_gcc): Handle early debug sections in the IL files by
19812 extracting them to separate files, partially linkin them and
19813 feeding the result back as result to the linker.
19814 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19815 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19816 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19817 sections into a separate segment.
19818 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19820 (darwin_asm_dwarf_section): Likewise.
19821 (darwin_asm_output_dwarf_offset): Likewise.
19822 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19824 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19825 Alan Hayward <alan.hayward@arm.com>
19826 David Sherwood <david.sherwood@arm.com>
19828 * read-md.h (md_reader::record_potential_iterator_use): Replace
19829 pointer argument with an rtx and an index.
19830 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19831 (apply_mode_iterator): Likewise.
19832 (apply_code_iterator): Likewise.
19833 (apply_int_iterator): Likewise.
19834 (apply_subst_iterator): Likewise.
19835 (record_iterator_use): Likewise.
19836 (record_attribute_use): Likewise.
19837 (md_reader::record_potential_iterator_use): Likewise. Update calls
19838 to record_iterator_use and apply_iterator.
19839 (iterator_use): Replace ptr with x and index.
19840 (attribute_use): Likewise.
19841 (apply_attribute_uses): Update calls to apply_iterator.
19842 (apply_iterators): Likewise. Update initialization of iterator_use.
19843 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19844 and record_potential_iterator_use.
19845 (rtx_reader::read_rtx_operand): Likewise.
19847 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19848 Alan Hayward <alan.hayward@arm.com>
19849 David Sherwood <david.sherwood@arm.com>
19851 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19854 2017-08-21 Richard Biener <rguenther@suse.de>
19856 PR middle-end/81884
19857 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19858 at struct end conservatively when comparing common bases.
19860 2017-08-21 Richard Biener <rguenther@suse.de>
19862 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
19863 (mem_ref_in_stmt): Remove.
19864 (determine_max_movement): Use ref index to get at the reference.
19865 (invariantness_dom_walker::before_dom_children): Deal with
19866 lim data already initialized.
19867 (gather_mem_refs_stmt): Initialize lim data and record ref index.
19869 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19871 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
19872 (TARGET_ISA_ROUND): Ditto.
19873 (TARGET_ROUND): Ditto.
19874 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
19875 * config/i386/i386.md: Ditto.
19876 * config/i386/sse.md: Ditto.
19877 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
19878 with OPTION_MASK_ISA_SSE4_1.
19880 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19883 * doc/extend.texi (x86 Built-in Functions): Correct the name of
19884 __builtin_ia32_lzcnt_u16.
19886 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
19889 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
19890 (rs6000_set_current_function): Rewrite function to use it.
19892 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
19895 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
19896 and TYPE_WARN_IF_NOT_ALIGN.
19897 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
19898 (handle_warn_if_not_align): New.
19899 (place_union_field): Call handle_warn_if_not_align.
19900 (place_field): Call handle_warn_if_not_align.
19901 Copy TYPE_WARN_IF_NOT_ALIGN.
19902 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
19903 (layout_type): Likewise.
19904 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
19906 (tree_decl_common): Add warn_if_not_align.
19907 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
19908 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
19909 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
19910 (DECL_WARN_IF_NOT_ALIGN): Likewise.
19911 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
19912 * doc/extend.texi: Document warn_if_not_aligned attribute.
19913 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
19915 2017-08-17 Martin Liska <mliska@suse.cz>
19918 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
19919 (get_data_dependence): Use it as pointer type.
19920 (distribute_loop): Likewise.
19922 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19924 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
19925 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
19926 (p8_vmrgow_<mode>_direct): New define_insn.
19927 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
19928 handle endianness for vmrgew and vmrgow permute patterns.
19930 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19932 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
19933 * config/rs6000/rs6000-cpus.def: Remove comment.
19934 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
19935 (POWERPC_MASKS): Likewise.
19936 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
19937 use of TARGET_VSX_TIMODE.
19938 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
19939 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
19940 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
19941 (rs6000_option_override_internal): Remove dead code.
19942 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
19943 (rs6000_legitimize_reload_address): Likewise.
19944 (rs6000_legitimate_address_p): Likewise.
19945 (rs6000_opt_masks): Delete "vsx-timode".
19946 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
19947 from function comment.
19948 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
19949 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
19950 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
19952 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
19953 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
19954 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
19955 (VSX_TI): Likewise.
19957 (define_peephole2): Likewise.
19959 2017-08-17 Martin Sebor <msebor@redhat.com>
19962 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
19963 past the end of an array.
19964 (test_pp_format): Add test cases.
19966 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
19968 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
19969 missing ECF_NOTHROW flags.
19971 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19974 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
19975 operands residing in integer registers.
19976 (*vsx_le_perm_load_<mode>): Likewise.
19977 (*vsx_le_perm_store_<mode>): Likewise.
19978 (define_peephole2): Add peepholes to optimize the above.
19980 2017-08-17 Marek Polacek <polacek@redhat.com>
19982 PR middle-end/81814
19983 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
19984 to mimic what shorten_compare did. Change the return type to bool.
19985 (fold_cond_expr_with_comparison): Update call to
19986 operand_equal_for_comparison_p.
19987 (fold_ternary_loc): Likewise.
19989 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
19991 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
19993 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
19995 2017-08-17 Richard Biener <rguenther@suse.de>
19997 * tree-ssa-structalias.c (solve_graph): When propagating
19998 to successors update the graphs succ edges and avoid duplicate work.
20000 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20003 * config/i386/i386.c (ix86_option_override_internal): Save target
20004 specific options after ix86_stack_protector_guard_reg was changed.
20006 2017-08-17 Richard Biener <rguenther@suse.de>
20008 PR tree-optimization/81827
20009 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20010 (new_var_info): Initialize it conservatively.
20011 (get_call_vi): Mark register vars.
20012 (new_scalar_tmp_constraint_exp): Likewise.
20013 (handle_rhs_call): Likewise.
20014 (handle_const_call): Likewise.
20015 (create_function_info_for): Likewise.
20016 (solve_constraints): Sort varinfos to separate register from
20017 non-register vars to pack points-to solution bitmaps during
20020 2017-08-17 Marek Polacek <polacek@redhat.com>
20022 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20024 2017-08-17 Richard Biener <rguenther@suse.de>
20026 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20027 to true when overflow is undefined and we saturated the result.
20029 2017-08-17 Alan Modra <amodra@gmail.com>
20032 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20033 Don't use store multiple if only one reg needs saving.
20034 (interesting_frame_related_regno): New function.
20035 (rs6000_frame_related): Don't emit frame info for regs that
20037 (rs6000_emit_epilogue): Likewise.
20039 2017-08-16 Nathan Sidwell <nathan@acm.org>
20041 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20042 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20043 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20044 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20045 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20046 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20047 (verify_type): Adjust for TYPE_BINFO move.
20048 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20049 process TYPE_BINFO directly.
20050 (hash_tree): Likewise.
20051 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20053 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20056 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20058 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20060 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20063 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20064 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20065 Add ix86_binary_operator_ok to insn constraint.
20066 (*iordi_1_bts): Ditto.
20067 (*xordi_1_btc): Ditto.
20068 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20069 Update corresponding peephole2 pattern.
20073 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20075 PR tree-optimization/81832
20076 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20077 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20079 2017-08-16 Marek Polacek <polacek@redhat.com>
20082 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20083 regarding size_zero_node.
20085 2017-08-16 Martin Liska <mliska@suse.cz>
20088 * config.gcc: Respect previously set extra_objs in case
20091 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20093 PR tree-optimization/81835
20094 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20095 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20096 not depend on the phi.
20098 2017-08-16 Alan Modra <amodra@gmail.com>
20100 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20103 2017-08-16 Alan Modra <amodra@gmail.com>
20105 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20106 (save_reg_p): ..into this. Update all callers.
20107 (first_reg_to_save): Simplify.
20109 2017-08-16 Alan Modra <amodra@gmail.com>
20111 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20114 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20118 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20119 constant count if that count is less than 32.
20121 2017-08-15 Nathan Sidwell <nathan@acm.org>
20123 * gcc.c (execute): Emit friendlier message if inferior is killed
20124 by an external cause.
20126 2017-08-15 Richard Biener <rguenther@suse.de>
20128 PR tree-optimization/81790
20129 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20130 CONSTRUCTORs from simplifying and VN.
20132 2017-08-14 Martin Sebor <msebor@redhat.com>
20134 * builtin-attrs.def: Add comments.
20136 2017-08-14 Martin Sebor <msebor@redhat.com>
20139 * doc/extend.texi (attribute nonstring): Document new attribute.
20141 2017-08-14 Martin Sebor <msebor@redhat.com>
20144 * tree-diagnostic.c (default_tree_printer): Handle %G.
20145 * gimple-pretty-print.h (percent_G_format): Declare new function.
20146 * gimple-pretty-print.c (percent_G_format): Define.
20147 * tree-pretty-print.c (percent_K_format): Add argument.
20149 2017-08-14 Martin Sebor <msebor@redhat.com>
20151 PR translation/79998
20152 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20153 Remove a stray space.
20155 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20158 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20159 (*iordi_1_bts): Ditto.
20160 (*xordi_1_btc): Ditto.
20162 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20165 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20166 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20168 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20169 quoted strings, and make more translator-friendly.
20170 (darwin_rs6000_override_options): Likewise.
20171 (rs6000_option_override_internal): Likewise.
20172 (rs6000_return_in_memory): Fix overlong line.
20173 (init_cmulative_args): Use quoted strings, and make more
20174 translator-friendly.
20175 (rs6000_pass_by_reference): Fix overlong line.
20176 (def_builtin): Use quoted strings.
20177 (altivec_expand_predicate_builtin): Use quoted strings, and make
20178 more translator-friendly.
20179 (htm_expand_builtin): Use quoted strings.
20180 (cpu_expand_builtin): Use quoted strings, and make more
20181 translator-friendly.
20182 (altivec_expand_builtin): Likewise.
20183 (paired_expand_predicate_builtin): Likewise.
20184 (rs6000_invalid_builtin): Likewise.
20185 (builtin_function_type): Use quoted strings.
20186 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20187 more translator-friendly.
20188 (rs6000_trampoline_init): Likewise.
20189 (rs6000_handle_altivec_attribute): Likewise.
20190 (rs6000_inner_target_options): Use quoted strings.
20191 (rs6000_disable_incompatible_switches): Likewise.
20192 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20193 strings, and make more translator-friendly.
20194 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20196 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20198 PR tree-optimization/81799
20199 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20200 cond_expr to simple gimple operand.
20202 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20204 PR middle-end/46932
20205 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20207 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20211 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20212 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20213 TARGET_GASISR_PROLOGUES.
20214 * config/avr/avr.c (avr_option_override): Same.
20215 (avr_pass_pre_proep::execute): Same.
20217 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20220 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20221 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20223 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20225 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20226 define_insn_and_split. Split to a memory load from 0 in
20227 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20228 using PTR mode iterator.
20229 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20230 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20231 (*add_tp_<mode>): Redefine as define_insn_and_split.
20232 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20233 address space. Merge with *add_tp_x32 using PTR mode iterator.
20234 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20235 Split to an add with a memory load from 0 in
20236 DEFAULT_TLS_SEG_REG address space.
20238 2017-08-12 Andrew Pinski <apinski@cavium.com>
20240 * config/aarch64/aarch64-option-extensions.def (rdma):
20241 Fix feature string to what Linux prints out in /proc/cpuinfo.
20243 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20246 * dwarf2out.c (modified_type_die): For C typedef types that have
20247 an ultimate origin, process the ultimate origin instead of the
20249 (gen_typedef_die): Assert that input DECLs have no ultimate
20251 (gen_type_die_with_usage): For typedef variants that have an
20252 ultimate origin, just call gen_decl_die on the original DECL.
20253 (process_scope_var): Avoid creating DIEs for local typedefs and
20254 concrete static variables.
20256 2017-08-12 Alan Modra <amodra@gmail.com>
20260 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20261 match gnu-user.h startfile.
20262 (ENDFILE_LINUX_SPEC): Similarly.
20264 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20267 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20269 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20271 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20272 * config/aarch64/aarch64.md (mov<mode>): Change.
20273 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20274 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20275 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20277 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20279 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20280 for storage order barriers.
20282 2017-08-11 Martin Liska <mliska@suse.cz>
20285 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20286 variables of void type.
20288 2017-08-11 Martin Liska <mliska@suse.cz>
20290 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20291 TARGET_SUPPORTS_ALIASES.
20292 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20293 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20294 (optimize_weakref): Likewise.
20295 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20296 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20297 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20299 2017-08-11 Martin Liska <mliska@suse.cz>
20302 * config/i386/i386.c (make_resolver_func): Do complete
20303 refactoring of the function.
20305 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20308 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20309 * config/i386/i386.c (ix86_stack_protect_guard): Use
20310 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20311 * doc/invoke.texi (x86 Options): Document
20312 -mstack-protector-guard-symbol= option.
20314 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20316 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20317 * config/i386/i386.c (ix86_split_stack_guard): New function.
20318 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20319 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20320 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20321 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20322 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20323 (split_stack_space_check): Call ix86_split_stack_guard.
20325 2017-08-10 Martin Sebor <msebor@redhat.com>
20327 * print-tree.c (print_node): Print location using the established
20329 Replace spaces with colons.
20330 (debug_raw, debug): Ditto.
20332 2017-08-10 Martin Sebor <msebor@redhat.com>
20335 * pretty-print.c (pp_format): Correct the handling of %s precision.
20337 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20340 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20342 (ix86_finalize_stack_frame_flags): This. Also clear
20343 frame_pointer_needed if -fno-omit-frame-pointer is used without
20345 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20346 with ix86_finalize_stack_frame_flags.
20347 (ix86_expand_epilogue): Likewise.
20348 (ix86_expand_split_stack_prologue): Likewise.
20349 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20351 2017-08-10 Martin Liska <mliska@suse.cz>
20354 * c-attribs.c (handle_target_attribute):
20355 Report warning for an empty string argument of target attribute.
20357 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20360 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20362 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20363 or DECL_NONLOCAL labels.
20364 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20365 or DECL_NONLOCAL labels here.
20367 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20369 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20370 to indicate when early gimple folding has been disabled.
20371 (rs6000_gimple_fold_builtin): Add debug content.
20372 (rs6000_invalid_builtin): Fix whitespace.
20373 (rs6000_expand_builtin): Fix whitespace.
20374 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20376 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20379 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20380 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20382 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20384 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20385 (qdf24xx): Likewise.
20386 * config/aarch64/aarch64.md: Include falkor.md.
20387 * config/aarch64/falkor.md: New.
20389 2017-08-09 Marek Polacek <polacek@redhat.com>
20392 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20393 * diagnostic.c (emit_diagnostic): Add a comment.
20394 (emit_diagnostic_valist): New function.
20396 2017-08-09 Marek Polacek <polacek@redhat.com>
20399 * input.c (make_location): New overload.
20400 * input.h (make_location): Declare.
20402 2017-08-08 Alan Modra <amodra@gmail.com>
20403 H.J. Lu <hongjiu.lu@intel.com>
20406 * gcc.c (NO_PIE_SPEC): Delete.
20407 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20409 (LINK_PIE_SPEC): ..to here.
20410 (LINK_COMMAND_SPEC): Support -no-pie.
20411 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20412 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20414 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20415 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20416 (ENDFILE_CRTEND_SPEC): Similarly.
20418 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20421 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20422 (mstack-protector-guard-offset=): Ditto.
20423 * config/i386/i386.c (ix86_option_override): Handle
20424 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20426 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20427 ix86_stack_protect_guard_offset variables.
20428 (TARGET_STACK_PROTECT_GUARD): Always define.
20429 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20430 and -mstack-protector-guard-offset= options.
20432 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20434 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20435 boundary case for the last candidate.
20437 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20439 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20440 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20442 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20444 PR middle-end/19706
20445 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20446 * config/aarch64/aarch64-builtins.c
20447 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20448 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20449 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20451 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20452 Andrew Pinski <pinskia@gmail.com>
20454 PR middle-end/19706
20455 * internal-fn.def (XORSIGN): New.
20456 * optabs.def (xorsign_optab): New.
20457 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20458 (convert_expand_mult_copysign): New.
20459 (pass_optimize_widening_mul::execute): Call
20460 convert_expand_mult_copysign.
20462 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20464 PR tree-optimization/81354
20465 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20466 Insert on edges rather than explicitly creating landing pads.
20467 (analyze_candidates_and_replace): Commit edge inserts.
20469 2017-08-08 Richard Biener <rguenther@suse.de>
20471 PR middle-end/81719
20472 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20473 (expand_simple_operations): Also look through ADDR_EXPRs with
20474 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20476 2017-08-08 Richard Biener <rguenther@suse.de>
20478 PR tree-optimization/81723
20479 * tree-vect-slp.c (struct bst_traits): New hash traits.
20480 (bst_fail): New global.
20481 (vect_build_slp_tree_2): New worker, split out from ...
20482 (vect_build_slp_tree): ... this now wrapping it with using
20483 bst_fail set to cache SLP tree build fails. Properly handle
20485 (vect_analyze_slp_instance): Allocate and free bst_fail.
20487 2017-08-08 Martin Liska <mliska@suse.cz>
20490 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20491 LABEL_DECLs that can be from a different function.
20493 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20495 PR tree-optimization/81744
20496 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20497 loop's number of iterations.
20499 2017-08-08 Martin Liska <mliska@suse.cz>
20501 * asan.c: Include header files.
20502 * attribs.c (build_decl_attribute_variant): New function moved
20504 (build_type_attribute_qual_variant): Likewise.
20505 (cmp_attrib_identifiers): Likewise.
20506 (simple_cst_list_equal): Likewise.
20507 (omp_declare_simd_clauses_equal): Likewise.
20508 (attribute_value_equal): Likewise.
20509 (comp_type_attributes): Likewise.
20510 (build_type_attribute_variant): Likewise.
20511 (lookup_ident_attribute): Likewise.
20512 (remove_attribute): Likewise.
20513 (merge_attributes): Likewise.
20514 (merge_type_attributes): Likewise.
20515 (merge_decl_attributes): Likewise.
20516 (merge_dllimport_decl_attributes): Likewise.
20517 (handle_dll_attribute): Likewise.
20518 (attribute_list_equal): Likewise.
20519 (attribute_list_contained): Likewise.
20520 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20521 (lookup_attribute_by_prefix): Likewise.
20522 * bb-reorder.c: Include header files.
20523 * builtins.c: Likewise.
20524 * calls.c: Likewise.
20525 * cfgexpand.c: Likewise.
20526 * cgraph.c: Likewise.
20527 * cgraphunit.c: Likewise.
20528 * convert.c: Likewise.
20529 * dwarf2out.c: Likewise.
20530 * final.c: Likewise.
20531 * fold-const.c: Likewise.
20532 * function.c: Likewise.
20533 * gimple-expr.c: Likewise.
20534 * gimple-fold.c: Likewise.
20535 * gimple-pretty-print.c: Likewise.
20536 * gimple.c: Likewise.
20537 * gimplify.c: Likewise.
20538 * hsa-common.c: Likewise.
20539 * hsa-gen.c: Likewise.
20540 * internal-fn.c: Likewise.
20541 * ipa-chkp.c: Likewise.
20542 * ipa-cp.c: Likewise.
20543 * ipa-devirt.c: Likewise.
20544 * ipa-fnsummary.c: Likewise.
20545 * ipa-inline.c: Likewise.
20546 * ipa-visibility.c: Likewise.
20548 * lto-cgraph.c: Likewise.
20549 * omp-expand.c: Likewise.
20550 * omp-general.c: Likewise.
20551 * omp-low.c: Likewise.
20552 * omp-offload.c: Likewise.
20553 * omp-simd-clone.c: Likewise.
20554 * opts-global.c: Likewise.
20555 * passes.c: Likewise.
20556 * predict.c: Likewise.
20557 * sancov.c: Likewise.
20558 * sanopt.c: Likewise.
20559 * symtab.c: Likewise.
20560 * toplev.c: Likewise.
20561 * trans-mem.c: Likewise.
20562 * tree-chkp.c: Likewise.
20563 * tree-eh.c: Likewise.
20564 * tree-into-ssa.c: Likewise.
20565 * tree-object-size.c: Likewise.
20566 * tree-parloops.c: Likewise.
20567 * tree-profile.c: Likewise.
20568 * tree-ssa-ccp.c: Likewise.
20569 * tree-ssa-live.c: Likewise.
20570 * tree-ssa-loop.c: Likewise.
20571 * tree-ssa-sccvn.c: Likewise.
20572 * tree-ssa-structalias.c: Likewise.
20573 * tree-ssa.c: Likewise.
20574 * tree-streamer-in.c: Likewise.
20575 * tree-vectorizer.c: Likewise.
20576 * tree-vrp.c: Likewise.
20577 * tsan.c: Likewise.
20578 * ubsan.c: Likewise.
20579 * varasm.c: Likewise.
20580 * varpool.c: Likewise.
20581 * tree.c: Remove functions moved to attribs.[ch].
20582 * tree.h: Likewise.
20583 * config/aarch64/aarch64.c: Add attrs.h header file.
20584 * config/alpha/alpha.c: Likewise.
20585 * config/arc/arc.c: Likewise.
20586 * config/arm/arm.c: Likewise.
20587 * config/avr/avr.c: Likewise.
20588 * config/bfin/bfin.c: Likewise.
20589 * config/c6x/c6x.c: Likewise.
20590 * config/cr16/cr16.c: Likewise.
20591 * config/cris/cris.c: Likewise.
20592 * config/darwin.c: Likewise.
20593 * config/epiphany/epiphany.c: Likewise.
20594 * config/fr30/fr30.c: Likewise.
20595 * config/frv/frv.c: Likewise.
20596 * config/ft32/ft32.c: Likewise.
20597 * config/h8300/h8300.c: Likewise.
20598 * config/i386/winnt.c: Likewise.
20599 * config/ia64/ia64.c: Likewise.
20600 * config/iq2000/iq2000.c: Likewise.
20601 * config/lm32/lm32.c: Likewise.
20602 * config/m32c/m32c.c: Likewise.
20603 * config/m32r/m32r.c: Likewise.
20604 * config/m68k/m68k.c: Likewise.
20605 * config/mcore/mcore.c: Likewise.
20606 * config/microblaze/microblaze.c: Likewise.
20607 * config/mips/mips.c: Likewise.
20608 * config/mmix/mmix.c: Likewise.
20609 * config/mn10300/mn10300.c: Likewise.
20610 * config/moxie/moxie.c: Likewise.
20611 * config/msp430/msp430.c: Likewise.
20612 * config/nds32/nds32-isr.c: Likewise.
20613 * config/nds32/nds32.c: Likewise.
20614 * config/nios2/nios2.c: Likewise.
20615 * config/nvptx/nvptx.c: Likewise.
20616 * config/pa/pa.c: Likewise.
20617 * config/pdp11/pdp11.c: Likewise.
20618 * config/powerpcspe/powerpcspe.c: Likewise.
20619 * config/riscv/riscv.c: Likewise.
20620 * config/rl78/rl78.c: Likewise.
20621 * config/rx/rx.c: Likewise.
20622 * config/s390/s390.c: Likewise.
20623 * config/sh/sh.c: Likewise.
20624 * config/sol2.c: Likewise.
20625 * config/sparc/sparc.c: Likewise.
20626 * config/spu/spu.c: Likewise.
20627 * config/stormy16/stormy16.c: Likewise.
20628 * config/tilegx/tilegx.c: Likewise.
20629 * config/tilepro/tilepro.c: Likewise.
20630 * config/v850/v850.c: Likewise.
20631 * config/vax/vax.c: Likewise.
20632 * config/visium/visium.c: Likewise.
20633 * config/xtensa/xtensa.c: Likewise.
20635 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20638 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20639 constraints since the -mupper-regs-* switches have been
20641 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20642 into a vector from a double word element that was extracted from
20643 another vector, and eliminate extra XXPERMDI instructions.
20644 (vsx_concat_<mode>_2): Likewise.
20645 (vsx_concat_<mode>_3): Likewise.
20646 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20647 concat to allow optimizing inserts from previous extracts.
20649 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20651 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20652 memory reference to a SSP offset in TLS address space.
20653 (ix86_print_operand) <case '@'>: Remove.
20654 (ix86_print_operand_punct_valid_p): Remove '@' code.
20655 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20656 UNSPEC_SP_TLS_TEST.
20657 (stack_tls_protect_set_<mode>): Remove.
20658 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20659 (stack_tls_protect_test_<mode>): Remove.
20660 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20662 2017-08-07 Olivier Hainque <hainque@adacore.com>
20665 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20667 2017-08-07 Douglas Rupp <rupp@adacore.com>
20669 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20670 variable was referenced as multidir in command.
20672 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20675 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20678 2017-08-07 Martin Liska <mliska@suse.cz>
20680 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20681 * config/rl78/rl78.c: Add include of attribs.h.
20682 * config/sh/sh.c: Likewise.
20683 * config/v850/v850.c: Likewise.
20685 2017-08-07 Tom de Vries <tom@codesourcery.com>
20687 PR middle-end/78266
20688 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20690 2017-08-07 Martin Liska <mliska@suse.cz>
20692 * config/mips/mips.c: Include attribs.h.
20694 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20697 * doc/invoke.texi: Document change in behvaior for -Ofast for
20700 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20702 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20704 (aarch64_pop_regs): Likewise.
20705 (aarch64_gen_load_pair): Likewise.
20706 (aarch64_save_callee_saves): Likewise.
20707 (aarch64_restore_callee_saves): Likewise.
20709 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20711 * config/i386/i386.c: Revert the last change.
20713 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20716 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20718 (ix86_finalize_stack_frame_flags): This. Also clear
20719 frame_pointer_needed if -fno-omit-frame-pointer is used without
20721 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20722 with ix86_finalize_stack_frame_flags.
20723 (ix86_expand_epilogue): Likewise.
20724 (ix86_expand_split_stack_prologue): Likewise.
20726 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20729 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20730 to P_AES for Westmere.
20732 2017-08-07 Jonathan Yong <10walls@gmail.com>
20734 * config/i386/mingw.opt (fset-stack-executable): Removed.
20735 * config/i386/cygming.opt (fset-stack-executable): Moved
20737 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20739 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20741 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20743 2017-08-07 Marek Polacek <polacek@redhat.com>
20745 PR middle-end/81737
20746 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20748 2017-08-07 Martin Liska <mliska@suse.cz>
20750 * attribs.h (canonicalize_attr_name): New function.
20751 (cmp_attribs): Move from c-format.c and adjusted.
20752 (is_attribute_p): Moved from tree.h.
20753 * tree-inline.c: Add new includes.
20754 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20755 (private_is_attribute_p): Remove.
20756 (private_lookup_attribute): Likewise.
20757 (private_lookup_attribute_by_prefix): Simplify.
20758 (remove_attribute): Use is_attribute_p.
20759 * tree.h: Remove removed declarations.
20761 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20763 PR middle-end/81698
20764 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20765 instead of computing it in the function. Formatting fix.
20766 (expand_case): Don't rely on default_edge being the first edge,
20767 clear it if removing it, pass default_edge to
20768 emit_case_dispatch_table.
20769 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20772 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20774 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20775 insn in the function, emit NOP after the insn.
20777 2017-08-06 Tom de Vries <tom@codesourcery.com>
20779 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20782 2017-08-06 Tom de Vries <tom@codesourcery.com>
20784 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20787 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20789 PR tree-optimization/57371
20790 * match.pd: New pattern.
20792 2017-08-04 Marek Polacek <polacek@redhat.com>
20794 PR middle-end/81695
20795 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20796 perform the computation in offset_int.
20798 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20800 PR tree-optimization/81136
20801 * tree-vectorizer.h: Include tree-hash-traits.h.
20802 (vec_base_alignments): New typedef.
20803 (vec_info): Add a base_alignments field.
20804 (vect_record_base_alignments): Declare.
20805 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20807 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20808 (create_data_ref): Add an is_conditional_in_stmt argument.
20809 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20810 the is_conditional_in_stmt field.
20811 (data_ref_loc): Add an is_conditional_in_stmt field.
20812 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20813 (find_data_references_in_stmt): Update call to create_data_ref.
20814 (graphite_find_data_references_in_stmt): Likewise.
20815 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20816 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20817 (vect_record_base_alignment): New function.
20818 (vect_record_base_alignments): Likewise.
20819 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20820 for nested statements even if we fail to compute a misalignment.
20821 Use pooled base alignments for unconditional references.
20822 (vect_find_same_alignment_drs): Compare base addresses instead
20824 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20825 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20827 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20829 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20830 Add an explicit name for the enum. Use auto_vec for slp_instances
20831 and grouped_stores.
20832 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20834 (_bb_vec_info): Add a constructor and destructor.
20835 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20836 (destroy_loop_vec_info): Delete.
20837 (vect_destroy_datarefs): Likewise.
20838 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20839 (vec_info::vec_info): New function.
20840 (vec_info::~vec_info): Likewise.
20841 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20842 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20843 destroy_loop_vec_info.
20844 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20845 (_loop_vec_info::_loop_vec_info): ...this.
20846 (destroy_loop_vec_info): Replace with...
20847 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20848 the stmt_vec_infos. Leave handling of vec_info information to its
20849 destructor. Remove explicit vector releases.
20850 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20851 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20852 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20853 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20854 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20855 (destroy_bb_vec_info): Replace with...
20856 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20857 information to its destructor.
20858 (vect_slp_analyze_bb_1): Use new and delete instead of
20859 new_bb_vec_info and destroy_bb_vec_info.
20860 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
20863 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20865 * tree-data-ref.h (subscript): Add access_fn field.
20866 (data_dependence_relation): Add could_be_independent_p.
20867 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
20868 (same_access_functions): Move to tree-data-ref.c.
20869 * tree-data-ref.c (ref_contains_union_access_p): New function.
20870 (access_fn_component_p): Likewise.
20871 (access_fn_components_comparable_p): Likewise.
20872 (dr_analyze_indices): Add a reference to access_fn_component_p.
20873 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
20875 (constant_access_functions): Likewise.
20876 (add_other_self_distances): Likewise.
20877 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
20878 (initialize_data_dependence_relation): Use XCNEW and remove
20879 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
20880 of access functions that have the same type. Allow the
20881 subsequence to end with different bases in some circumstances.
20882 Record the chosen access functions in SUB_ACCESS_FN.
20883 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
20884 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
20885 (subscript_dependence_tester_1): Likewise dra and drb.
20886 (build_classic_dist_vector): Update calls accordingly.
20887 (subscript_dependence_tester): Likewise.
20888 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
20889 DDR_COULD_BE_INDEPENDENT_P.
20890 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
20891 comp_alias_ddrs instead of may_alias_ddrs.
20892 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
20894 (vect_analyze_data_ref_dependence): Use it if
20895 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
20896 distance vectors if that fails.
20897 (dependence_distance_ge_vf): New function.
20898 (vect_prune_runtime_alias_test_list): Use it. Don't clear
20899 LOOP_VINFO_MAY_ALIAS_DDRS.
20901 2017-08-04 Richard Biener <rguenther@suse.de>
20903 PR middle-end/81705
20904 * fold-const.c (fold_binary_loc): Properly restrict
20905 minus_var0 && minus_var1 case when associating undefined overflow
20908 2017-08-04 Tom de Vries <tom@codesourcery.com>
20910 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
20912 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20914 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20915 Don't start diagnostic messages with a capital letter.
20916 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20918 (rs6000_invalid_builtin): Likewise.
20919 (rs6000_trampoline_init): Likewise.
20921 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20924 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
20925 after setting changeable df flags.
20927 2017-08-03 Richard Biener <rguenther@suse.de>
20929 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
20930 up if the use is in USE - X.
20932 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
20934 * toplev.c (dumpfile.h): New include.
20935 (internal_error_reentered): New static function. Use it...
20936 (internal_error_function): ...here to handle reentered internal_error.
20938 2017-08-03 Richard Biener <rguenther@suse.de>
20940 PR middle-end/81148
20941 * fold-const.c (split_tree): Add minus_var and minus_con
20942 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
20943 here but always use minus_*.
20944 (associate_trees): Assert we never associate with MINUS_EXPR
20945 and NULL first operand. Do not recurse for PLUS_EXPR operands
20946 when associating as MINUS_EXPR either.
20947 (fold_binary_loc): Track minus_var and minus_con.
20949 2017-08-03 Tom de Vries <tom@codesourcery.com>
20952 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
20953 ACCEL_COMPILER, apply finish_options on
20954 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
20956 2017-08-03 Tom de Vries <tom@codesourcery.com>
20959 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
20960 function_entry_patch_area_size > 0.
20962 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20965 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
20966 instead of 10??LU, perform unit multiplication in wide_int,
20967 don't change alloc_object_size_limit if the limit is larger
20970 PR tree-optimization/81655
20971 PR tree-optimization/81588
20972 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
20973 the case when ranges[i].low and high are 1 for unsigned type with
20976 PR middle-end/81052
20977 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
20978 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
20980 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20982 * tree-vrp.h: Add include guard.
20984 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
20987 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
20988 (ud2): New insn pattern.
20989 * config/i386/i386.c (ix86_expand_epilogue):
20990 For naked functions, generate ud2 instead of trap insn.
20992 2017-08-02 Marek Polacek <polacek@redhat.com>
20995 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
20997 2017-08-02 Tom de Vries <tom@codesourcery.com>
20998 Cesar Philippidis <cesar@codesourcery.com>
21000 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21001 Add missing edge probabilities.
21003 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21005 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21006 Correct endianness.
21008 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21010 PR middle-end/79499
21011 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21012 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21013 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21014 prologue_seq sequences - if any.
21016 2017-08-02 Richard Biener <rguenther@suse.de>
21018 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21019 via vectors if supported, integer extracts via punning if supported
21020 or otherwise vector extracts.
21022 2017-08-02 Richard Biener <rguenther@suse.de>
21024 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21026 (bitmap_insert_into_set): ... this.
21028 2017-08-02 Richard Biener <rguenther@suse.de>
21030 PR tree-optimization/81633
21032 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21034 PR tree-optimization/71752
21035 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21037 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21039 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21040 (machine_function::call_ms2sysv_pad_out): Remove field.
21041 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21042 (ix86_compute_frame_layout): Likewise.
21044 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21047 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21048 attribute with interrupt attribute.
21050 2017-08-01 Andrew Pinski <apinski@cavium.com>
21052 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21053 BIT_INSERT_EXPR's operand 1
21054 to see if the types precision matches.
21056 2017-08-01 Martin Liska <mliska@suse.cz>
21058 PR middle-end/70140
21059 * builtins.c (expand_builtin_memcpy_args): Remove.
21060 (expand_builtin_memcpy): Call newly added function
21061 expand_builtin_memory_copy_args.
21062 (expand_builtin_memcpy_with_bounds): Likewise.
21063 (expand_builtin_mempcpy): Remove last argument.
21064 (expand_builtin_mempcpy_with_bounds): Likewise.
21065 (expand_builtin_memory_copy_args): New function created from
21066 expand_builtin_mempcpy_args with small modifications.
21067 (expand_builtin_mempcpy_args): Remove.
21068 (expand_builtin_stpcpy): Remove unused argument.
21069 (expand_builtin): Likewise.
21070 (expand_builtin_with_bounds): Likewise.
21072 2017-08-01 Martin Liska <mliska@suse.cz>
21075 Make mempcpy more optimal (PR middle-end/70140).
21077 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21080 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21081 __builtin_vec_cmpne verify both arguments are compatible vectors
21082 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21083 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21084 move computation of aligned to after checking the argument types.
21088 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21089 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21092 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21093 Doug Rupp <rupp@adacore.com>
21094 Olivier Hainque <hainque@adacore.com>
21096 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21097 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21099 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21100 for TARGET_OS_CPP_BUILTIN.
21101 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21102 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21104 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21105 passing required abi options to the assembler for EABI configurations.
21106 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21107 of .text.hot and .text.unlikely sections for kernel modules when
21108 using ARM style exceptions.
21109 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21110 options. Add EXTRA_CC1_SPEC.
21111 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21113 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21115 (ARM_TARGET2_DWARF_FORMAT): Define.
21116 * config/arm/t-vxworks: Adjust multilib control to removal of the
21117 Diab command line options.
21119 2017-08-01 Martin Liska <mliska@suse.cz>
21121 PR gcov-profile/81561
21122 * gcov.c (unblock): Make unblocking safe as we need to preserve
21123 index correspondence of blocks and block_lists.
21125 2017-08-01 Richard Biener <rguenther@suse.de>
21127 PR tree-optimization/81181
21128 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21129 (compute_antic): ... end of iteration here.
21131 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21133 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21134 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21135 (ftree-slp-vectorize): Likewise.
21136 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21137 can no longer be set independent of flag_tree_loop_vectorize.
21138 * omp-general.c (emp_max_vf): Likewise.
21139 * opts.c (enable_fdo_optimizations): Remove references to
21140 flag_tree_vectorize, these are now implicit.
21141 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21142 and leave it for the options machinery.
21144 2017-08-01 Martin Liska <mliska@suse.cz>
21146 PR middle-end/70140
21147 * builtins.c (expand_builtin_memcpy_args): Remove.
21148 (expand_builtin_memcpy): Call newly added function
21149 expand_builtin_memory_copy_args.
21150 (expand_builtin_memcpy_with_bounds): Likewise.
21151 (expand_builtin_mempcpy): Remove last argument.
21152 (expand_builtin_mempcpy_with_bounds): Likewise.
21153 (expand_builtin_memory_copy_args): New function created from
21154 expand_builtin_mempcpy_args with small modifications.
21155 (expand_builtin_mempcpy_args): Remove.
21156 (expand_builtin_stpcpy): Remove unused argument.
21157 (expand_builtin): Likewise.
21158 (expand_builtin_with_bounds): Likewise.
21160 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21163 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21164 print "ds:" only for immediates in generic address space.
21166 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21169 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21170 (ix86_function_ok_for_sibcall): Return false for naked functions.
21172 2017-08-01 Richard Biener <rguenther@suse.de>
21174 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21175 (compute_antic): Seed worklist with exit block predecessors.
21176 * cfganal.c (dfs_find_deadend): For a cycle return the source
21177 of the edge closing it.
21179 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21181 * config/aarch64/aarch64.c
21182 (aarch64_can_const_movi_rtx_p): Move 0 check.
21184 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21186 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21187 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21189 * match.pd: Ditto in address comparison pattern.
21191 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21193 PR tree-optimization/81627
21194 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21195 closed ssa form for store-store chain.
21197 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21199 PR tree-optimization/81620
21200 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21201 for store-store chain.
21203 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21205 PR tree-optimization/81588
21206 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21207 ranges[i].in_p, invert comparison code ccode. For >/>=,
21208 swap rhs1 and rhs2 and comparison code unconditionally,
21209 for </<= don't do that. Don't swap rhs1/rhs2 again if
21210 ranges[i].in_p, instead invert comparison code ccode if
21211 opcode or oe->rank is BIT_IOR_EXPR.
21214 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21215 a direct optab to conversion optab.
21216 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21217 with GET_MODE_INNER as last argument instead of optab_handler.
21218 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21219 vector extraction if possible and optab is available.
21220 * expr.c (store_constructor): Use convert_optab_handler instead
21221 of optab_handler. Use vector initialization from smaller
21222 vectors if possible and optab is available.
21223 * tree-vect-stmts.c (vectorizable_load): Likewise.
21224 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21225 now have two modes.
21226 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21227 of vec_init from half-sized vectors with the same element mode.
21228 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21229 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21230 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21231 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21232 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21233 after mode in gen_vec_extract* calls.
21234 (vec_extract<mode>): Renamed to ...
21235 (vec_extract<mode><ssescalarmodelower>): ... this.
21236 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21237 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21238 element mode after mode in gen_vec_init* calls.
21239 (VEC_INIT_HALF_MODE): New mode iterator.
21240 (vec_init<mode>): Renamed to ...
21241 (vec_init<mode><ssescalarmodelower>): ... this.
21242 (vec_init<mode><ssehalfvecmodelower>): New expander.
21243 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21244 (vec_extractv2sfsf): ... this.
21245 (vec_initv2sf): Renamed to ...
21246 (vec_initv2sfsf): ... this.
21247 (vec_extractv2si): Renamed to ...
21248 (vec_extractv2sisi): ... this.
21249 (vec_initv2si): Renamed to ...
21250 (vec_initv2sisi): ... this.
21251 (vec_extractv4hi): Renamed to ...
21252 (vec_extractv4hihi): ... this.
21253 (vec_initv4hi): Renamed to ...
21254 (vec_initv4hihi): ... this.
21255 (vec_extractv8qi): Renamed to ...
21256 (vec_extractv8qiqi): ... this.
21257 (vec_initv8qi): Renamed to ...
21258 (vec_initv8qiqi): ... this.
21259 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21260 (vec_init<mode>): Renamed to ...
21261 (vec_init<mode><VEC_base_l>): ... this.
21262 (vec_extract<mode>): Renamed to ...
21263 (vec_extract<mode><VEC_base_l>): ... this.
21264 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21265 (vec_initv2sfsf): ... this.
21266 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21267 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21268 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21269 element mode after mode in gen_vec_init* calls.
21270 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21271 (vec_init<mode><Vel>): ... this.
21272 (vec_extract<mode>): Renamed to ...
21273 (vec_extract<mode><Vel>): ... this.
21274 * config/aarch64/iterators.md (Vel): New mode attribute.
21275 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21276 Add element mode after mode in gen_vec_extract* calls.
21277 * config/s390/vector.md (non_vec_l): New mode attribute.
21278 (vec_extract<mode>): Renamed to ...
21279 (vec_extract<mode><non_vec_l>): ... this.
21280 (vec_init<mode>): Renamed to ...
21281 (vec_init<mode><non_vec_l>): ... this.
21282 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21283 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21285 * config/arm/iterators.md (V_elem_l): New mode attribute.
21286 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21287 (vec_extract<mode><V_elem_l>): ... this.
21288 (vec_extractv2di): Renamed to ...
21289 (vec_extractv2didi): ... this.
21290 (vec_init<mode>): Renamed to ...
21291 (vec_init<mode><V_elem_l>): ... this.
21292 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21293 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21294 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21295 Add element mode after gen_vec_extract* calls.
21296 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21297 (vec_init<mode><unitmode>): ... this.
21298 (vec_extract<mode>): Renamed to ...
21299 (vec_extract<mode><unitmode>): ... this.
21300 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21301 (vec_init<mode><unitmode>): ... this.
21302 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21303 (vec_initv2sfsf): ... this.
21304 (vec_extractv2sf): Renamed to ...
21305 (vec_extractv2sfsf): ... this.
21306 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21307 Add element mode after gen_vec_extract* calls.
21308 * config/mips/mips.md (unitmode): New mode iterator.
21309 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21310 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21311 * config/spu/spu.md (inner_l): New mode attribute.
21312 (vec_init<mode>): Renamed to ...
21313 (vec_init<mode><inner_l>): ... this.
21314 (vec_extract<mode>): Renamed to ...
21315 (vec_extract<mode><inner_l>): ... this.
21316 * config/sparc/sparc.md (veltmode): New mode iterator.
21317 (vec_init<VMALL:mode>): Renamed to ...
21318 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21319 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21320 (vec_initv2sisi): ... this.
21321 (vec_initv2sf): Renamed to ...
21322 (vec_initv2sfsf): ... this.
21323 (vec_extractv2sf): Renamed to ...
21324 (vec_extractv2sfsf): ... this.
21325 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21326 (vec_init<mode>): Renamed to ...
21327 (vec_init<mode><VEC_base_l>): ... this.
21328 (vec_extract<mode>): Renamed to ...
21329 (vec_extract<mode><VEC_base_l>): ... this.
21330 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21331 (vec_initv2sfsf): ... this.
21332 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21333 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21334 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21335 gen_vec_init* calls.
21337 2017-08-01 Richard Biener <rguenther@suse.de>
21339 PR tree-optimization/81297
21340 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21341 TREE_OVERFLOW from INTEGER_CSTs.
21343 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21345 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21347 2017-07-31 Carl Love <cel@us.ibm.com>
21349 * config/rs6000/rs6000-c: Add support for built-in functions
21350 vector signed char vec_xl_be (signed long long, signed char *);
21351 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21352 vector signed int vec_xl_be (signed long long, signed int *);
21353 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21354 vector signed long long vec_xl_be (signed long long, signed long long *);
21355 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21356 vector signed short vec_xl_be (signed long long, signed short *);
21357 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21358 vector double vec_xl_be (signed long long, double *);
21359 vector float vec_xl_be (signed long long, float *);
21360 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21361 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21362 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21364 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21365 (altivec_expand_builtin): Add switch statement to call
21366 altivec_expand_xl_be for each builtin.
21367 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21368 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21369 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21370 __builtin_vsx_le_be_v16qi.
21371 * doc/extend.texi: Update the built-in documentation file for the
21372 new built-in functions.
21374 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21377 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21379 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21381 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21383 * config.gcc: Add z14.
21384 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21385 CPU model numbers for z13s and z14.
21386 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21388 * config/s390/s390-opts.h (enum processor_type): Rename
21389 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21390 * config/s390/s390.c (processor_table): Add field for CPU name to
21391 be passed to Binutils.
21392 (s390_asm_output_machine_for_arch): Use the new field in
21393 processor_table for Binutils.
21394 (s390_expand_builtin): Replace arch12 with z14.
21395 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21396 (s390_get_sched_attrmask): Likewise.
21397 (s390_get_unit_mask): Likewise.
21398 * config/s390/s390.opt: Add z14 to processor_type enum.
21400 2017-07-31 Martin Jambor <mjambor@suse.cz>
21403 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21404 regardless of optimization level.
21406 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21407 Martin Liska <mliska@suse.cz>
21409 * predict.def: Remove old comment and adjust probability.
21410 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21411 PREDICT statements.
21413 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21416 * config/i386/i386.c (ix86_function_naked): New function.
21417 (ix86_can_use_return_insn_p): Return false for naked functions.
21418 (ix86_expand_prologue): Skip prologue for naked functions.
21419 (ix86_expand_epilogue): Skip epilogue for naked functions
21420 and emit trap instruction.
21421 (ix86_warn_func_return): New function.
21422 (ix86_attribute_table): Add "naked" attribute specification.
21423 (TARGET_WARN_FUNC_RETURN): Define.
21424 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21426 2017-07-31 Martin Liska <mliska@suse.cz>
21428 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21429 (dump_gimple_bb_header): Always dump BB info.
21430 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21432 2017-07-31 Martin Liska <mliska@suse.cz>
21435 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21436 also with current_function_decl non-null equality.
21438 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21441 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21442 change type to the element type, instead add eltype variable and
21443 use it where we are interested in the element type.
21445 PR tree-optimization/81603
21446 * ipa-polymorphic-call.c
21447 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21448 offset arithmetic in offset_int, bail out if the resulting bit offset
21449 doesn't fit into shwi.
21451 2017-07-31 Martin Liska <mliska@suse.cz>
21453 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21454 (gimplify_save_expr): Fix comment.
21456 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21459 * config/i386/i386.c (ix86_function_arg): Update arguments for
21461 (ix86_compute_frame_layout): Set the initial stack offset to
21462 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21463 INCOMING_FRAME_SP_OFFSET.
21464 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21465 stack before exception handler returns.
21466 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21467 the 'ERROR_CODE' for exception handler.
21469 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21471 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21472 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21473 (ASM_OUTPUT_REG_POP): Ditto.
21474 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21475 instead of asm_fprintf to output pure string.
21477 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21479 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21480 to imported_module_or_decl hook.
21481 (debug_nothing_tree_tree_tree_bool): Remove.
21482 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21483 * debug.c (do_nothing_debug_hooks): Use
21484 debug_nothing_tree_tree_tree_bool_bool instead of
21485 debug_nothing_tree_tree_tree_bool.
21486 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21487 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21488 * sdbout.c (sdb_debug_hooks): Likewise.
21489 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21490 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21492 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21493 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21494 attribute, don't add anything.
21496 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21498 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21499 (fold_build2_stat_loc): Likewise.
21500 (fold_build3_stat_loc): Likewise.
21501 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21502 (fold_build1_loc): Remove macro.
21503 (fold_build2_loc): Likewise.
21504 (fold_build3_loc): Likewise.
21506 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21508 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21509 (gimple_build_debug_bind_source_stat): Likewise.
21510 * gimple.h (gimple_build_debug_bind): Remove macro.
21511 (gimple_build_debug_bind_source): Likewise.
21513 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21515 * bitmap.c (bitmap_alloc): Adjust.
21516 (bitmap_gc_alloc): Likewise.
21517 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21519 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21521 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21522 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21523 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21524 (bitmap_gc_alloc_stat): Likewise.
21525 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21527 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21529 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21530 * rtl.h (shallow_copy_rtx): Remove macro.
21532 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21534 * emit-rtl.c (gen_raw_REG): Adjust.
21535 * gengenrtl.c (gendef): Likewise.
21536 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21537 * rtl.h (rtx_alloc): Remove macro.
21539 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21541 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21542 (build_tree_list_stat): Likewise.
21543 * tree.h (build_tree_list): Remove macro.
21544 (build_tree_list_vec): Likewise.
21546 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21548 * tree.c (make_vector_stat): Remove _stat from name.
21549 (build_vector_stat): Likewise.
21550 * tree.h (make_vector_stat): Remove macro.
21551 (build_vector_stat): Likewise.
21553 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21555 * tree.h (build_var_debug_value): Remove prototype.
21557 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21559 * tree.c (tree_cons_stat): Remove _stat from name.
21560 * tree.h (tree_cons): Remove macro.
21562 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21564 * tree.c (build_vl_exp_stat): Remove _stat from name.
21565 * tree.h (build_vl_exp): Remove macro.
21567 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21569 * tree.c (build_decl_stat): Remove _stat from name.
21570 * tree.h (build_decl): Remove macro.
21572 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21574 * gimple.c (gimple_build_with_ops_stat): Adjust.
21575 (gimple_alloc_stat): Remove _stat from name.
21576 * gimple.h (gimple_alloc): Remove macro.
21578 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21580 * tree.c (make_tree_vec_stat): Remove _stat from name.
21581 (grow_tree_vec_stat): Likewise.
21582 * tree.h (make_tree_vec_stat): Adjust prototype.
21583 (grow_tree_vec_stat): Likewise.
21584 (make_tree_vec): Remove macro.
21585 (grow_tree_vec): Likewise.
21587 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21589 * fold-const.c (fold_build1_stat_loc): Adjust.
21590 (fold_build2_stat_loc): Likewise.
21591 (fold_build3_stat_loc): Likewise.
21592 * tree.c (build0_stat): Remove _stat from name.
21593 (build1_stat): Likewise.
21594 (build2_stat): Likewise.
21595 (build3_stat): Likewise.
21596 (build4_stat): Likewise.
21597 (build5_stat): Likewise.
21598 * tree.h (build1_loc): Remove macro, and rename _stat function
21600 (build2_loc): Likewise.
21601 (build3_loc): Likewise.
21602 (build4_loc): Likewise.
21603 (build5_loc): Likewise.
21605 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21607 * tree.c (make_int_cst_stat): Remove _stat from name.
21608 * tree.h (make_int_cst_stat): Adjust prototype.
21609 (make_int_cst): Remove macro.
21611 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21613 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21614 * tree.h (make_tree_binfo_stat): Adjust prototype.
21615 (make_tree_binfo): Remove.
21617 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21619 * tree.c (copy_node_stat): Rename to copy_node.
21620 (build_distinct_type_copy): Adjust.
21621 * tree.h (copy_node_stat): Adjust prototype.
21622 (copy_node): Remove macro.
21624 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21626 * tree.c (make_node_stat): rename to make_node.
21627 (build_tree_list_stat): Adjust.
21628 (build0_stat): Likewise.
21629 (build2_stat): Likewise.
21630 (build3_stat): Likewise.
21631 (build4_stat): Likewise.
21632 (build5_stat): Likewise.
21633 (build_decl_stat): Likewise.
21634 * tree.h (make_node_stat): Adjust prototype.
21635 (make_node): remove macro.
21637 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21639 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21640 (PPC_FEATURE2_SCV): Likewise.
21641 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21643 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21645 * config/aarch64/aarch64.c
21646 (aarch64_internal_mov_immediate): Add new special pattern.
21647 * config/aarch64/aarch64.md (*movdi_aarch64):
21648 Add reg/32bit const mov case.
21650 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21651 Richard Sandiford <richard.sandiford@linaro.org>
21653 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21654 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21655 Add integer and movi cases.
21656 (movi-split-hf-df-sf split, fp16): New.
21657 (enabled): Added TARGET_FP_F16INST.
21658 * config/aarch64/iterators.md (GPF_HF): New.
21659 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21661 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21663 * config/aarch64/aarch64.c
21664 (aarch64_simd_container_mode): Add prototype.
21665 (aarch64_expand_mov_immediate): Add HI support.
21666 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21667 (aarch64_can_const_movi_rtx_p): New.
21668 (aarch64_preferred_reload_class):
21669 Remove restrictions of using FP registers for certain SIMD operations.
21670 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21671 (aarch64_valid_floating_const): Add integer move validation.
21672 (aarch64_simd_imm_scalar_p): Remove.
21673 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21674 (aarch64_legitimate_constant_p): Expand list of supported cases.
21675 * config/aarch64/aarch64-protos.h
21676 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21677 (aarch64_reinterpret_float_as_int): New.
21678 (aarch64_simd_imm_scalar_p): Remove.
21679 * config/aarch64/constraints.md (Uvi): New.
21680 (Dd): Split into Ds and new Dd.
21681 * config/aarch64/aarch64.md (*movsi_aarch64):
21683 (*movdi_aarch64): Add SIMD mov case.
21685 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21687 * tree-predcom.c: (struct chain): Handle store-store chain in which
21688 stores for elimination only store loop invariant values.
21689 (execute_pred_commoning_chain): Ditto.
21690 (prepare_initializers_chain_store_elim): Ditto.
21691 (prepare_finalizers): Ditto.
21692 (is_inv_store_elimination_chain): New function.
21693 (initialize_root_vars_store_elim_1): New function.
21695 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21697 * tree-predcom.c: Revise general description of the pass.
21698 (enum chain_type): New enum type for store elimination.
21699 (struct chain): New field supporting store elimination.
21700 (struct component): Ditto.
21701 (dump_chain): Dump store-stores chain.
21702 (release_chain): Release resources.
21703 (split_data_refs_to_components): Compute and create component
21704 contains only stores for elimination.
21705 (get_chain_last_ref_at): New function.
21706 (make_invariant_chain): Initialization.
21707 (make_rooted_chain): Specify chain type in parameter and record it.
21708 (add_looparound_copies): Skip for store-stores chain.
21709 (determine_roots_comp): Compute type of chain and pass it to
21711 (initialize_root_vars_store_elim_2): New function.
21712 (finalize_eliminated_stores): New function.
21713 (remove_stmt): Handle store for elimination.
21714 (execute_pred_commoning_chain): Execute predictive commoning on
21715 store-store chains.
21716 (determine_unroll_factor): Skip unroll for store-stores chain.
21717 (prepare_initializers_chain_store_elim): New function.
21718 (prepare_initializers_chain): Hanlde store-store chain.
21719 (prepare_finalizers_chain, prepare_finalizers): New function.
21720 (tree_predictive_commoning_loop): Return integer value indicating
21721 if loop is unrolled or lcssa form is corrupted.
21722 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21724 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21726 * tree-predcom.c (initialize_root): Delete.
21727 (execute_pred_commoning_chain): Initialize root vars and replace
21728 reference of non-combined chain directly, rather than call above
21731 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21733 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21734 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21736 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21738 * tree-predcom.c (struct chain): New field init_seq.
21739 (release_chain): Release init_seq.
21740 (prepare_initializers_chain): Record intialization stmts in above
21742 (insert_init_seqs): New function.
21743 (tree_predictive_commoning_loop): Call insert_init_seqs.
21745 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21747 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21749 2017-07-28 Richard Biener <rguenther@suse.de>
21751 * match.pd: Remove superfluous :c.
21752 * genmatch.c (simplify::id): Add member.
21753 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21755 (current_id): New global.
21756 (dt_node::parent): Move from ...
21757 (dt_operand::parent): ... here. Add for_id member.
21758 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21759 (decision_tree::find_node): Relax order requirement when
21760 merging DT_TRUE nodes to ones inbetween the current simplify
21761 and the one we try to merge with. Add diagnostic whenever
21762 we need to enforce pattern order by not merging.
21763 (decision_tree::insert): Set current_id.
21764 (decision_tree::print_node): Dump parent node and for_id.
21765 (parser::last_id): Add member.
21766 (parser::push_simplify): Assign unique id.
21767 (parser::parser): Initialize last_id.
21769 2017-07-28 Martin Liska <mliska@suse.cz>
21772 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21773 gimple_build_debug_bind.
21775 2017-07-28 Richard Biener <rguenther@suse.de>
21777 PR tree-optimization/81502
21778 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21780 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21782 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21783 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21784 for BIT_FIELD_REF args.
21785 * fold-const.c (make_bit_field_ref): Likewise.
21786 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21788 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21791 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21792 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21793 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21794 Or it into SANITIZER_UNDEFINED.
21795 * ubsan.c: Include gimple-fold.h and varasm.h.
21796 (ubsan_expand_ptr_ifn): New function.
21797 (instrument_pointer_overflow): New function.
21798 (maybe_instrument_pointer_overflow): New function.
21799 (instrument_object_size): Formatting fix.
21800 (pass_ubsan::execute): Call instrument_pointer_overflow
21801 and maybe_instrument_pointer_overflow.
21802 * internal-fn.c (expand_UBSAN_PTR): New function.
21803 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21804 * sanitizer.def (__ubsan_handle_pointer_overflow,
21805 __ubsan_handle_pointer_overflow_abort): New builtins.
21806 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21807 * internal-fn.def (UBSAN_PTR): New internal function.
21808 * opts.c (sanitizer_opts): Add pointer-overflow.
21809 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21810 * fold-const.c (build_range_check): Compute pointer range check in
21811 integral type if pointer arithmetics would be needed. Formatting
21814 2017-07-28 Martin Liska <mliska@suse.cz>
21817 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21818 parameters that are of a variable-length.
21820 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21822 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21824 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21825 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21826 (CRT_CALL_STATIC_FUNCTION): Likewise.
21827 (ASM_DEFAULT_SPEC): New define.
21828 (ASM_SPEC32): Likewise.
21829 (ASM_SPEC64): Likewise.
21830 (ASM_SPEC_COMMON): Likewise.
21831 (ASM_SPEC): Likewise.
21832 (INVALID_64BIT): Likewise.
21833 (LINK_OS_DEFAULT_SPEC): Likewise.
21834 (LINK_OS_SPEC32): Likewise.
21835 (LINK_OS_SPEC64): Likewise.
21836 (POWERPC_LINUX): Likewise.
21837 (PTRDIFF_TYPE): Likewise.
21838 (RESTORE_FP_PREFIX): Likewise.
21839 (RESTORE_FP_SUFFIX): Likewise.
21840 (SAVE_FP_PREFIX): Likewise.
21841 (SAVE_FP_SUFFIX): Likewise.
21842 (SIZE_TYPE): Likewise.
21843 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21844 (TARGET_64BIT): Likewise.
21845 (TARGET_64BIT): Likewise.
21846 (TARGET_AIX): Likewise.
21847 (WCHAR_TYPE_SIZE): Likewise.
21848 (WCHAR_TYPE): Undefine.
21849 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21850 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21851 (CPP_OS_RTEMS_SPEC): Delete.
21852 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21853 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21855 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21857 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21859 PR tree-optimization/81578
21860 * tree-parloops.c (build_new_reduction): Bail out if
21861 reduction_code isn't one of the standard OpenMP reductions.
21862 Move the details printing after that decision.
21864 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21866 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
21867 related to reload_in_progress.
21868 (splat_input_operand): Likewise.
21869 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
21871 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
21873 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
21874 (TARGET_INSTANTIATE_DECLS): Likewise.
21875 (legitimate_indexed_address_p): Delete reload_in_progress code.
21876 (rs6000_debug_legitimate_address_p): Likewise.
21877 (rs6000_eliminate_indexed_memrefs): Likewise.
21878 (rs6000_emit_le_vsx_store): Likewise.
21879 (rs6000_emit_move_si_sf_subreg): Likewise.
21880 (rs6000_emit_move): Likewise.
21881 (register_to_reg_type): Likewise.
21882 (rs6000_pre_atomic_barrier): Likewise.
21883 (rs6000_machopic_legitimize_pic_address): Likewise.
21884 (rs6000_allocate_stack_temp): Likewise.
21885 (rs6000_address_for_fpconvert): Likewise.
21886 (rs6000_address_for_altivec): Likewise.
21887 (rs6000_secondary_memory_needed_rtx): Delete function.
21888 (rs6000_check_sdmode): Likewise.
21889 (rs6000_alloc_sdmode_stack_slot): Likewise.
21890 (rs6000_instantiate_decls): Likewise.
21891 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
21892 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
21893 Delete reload_in_progress.
21894 (*vec_reload_and_plus_<mptrsize>): Likewise.
21895 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
21896 (vsx_div_v2di): Likewise.
21897 (vsx_udiv_v2di): Likewise.
21899 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21901 * config/rs6000/rs6000.opt (mlra): Replace with stub.
21902 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
21903 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
21904 (rs6000_debug_reg_global): Delete print of LRA status.
21905 (rs6000_option_override_internal): Delete dead LRA related code.
21906 (rs6000_lra_p): Delete function.
21907 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
21909 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
21911 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
21912 * config/riscv/rtems.h: New file.
21914 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21915 Sudakshina Das <sudi.das@arm.com>
21917 * config/aarch64/aarch64.md
21918 (define_split for and<mode>3nr_compare): Move
21919 non aarch64_logical_operand to a register.
21920 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
21921 register immediate operand to a register.
21922 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
21924 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21926 PR middle-end/81564
21927 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
21929 2017-07-27 Richard Biener <rguenther@suse.de>
21931 PR tree-optimization/81573
21932 PR tree-optimization/81494
21933 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
21934 multi defuse cycle case.
21936 2017-07-27 Richard Biener <rguenther@suse.de>
21938 PR tree-optimization/81571
21939 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
21942 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
21944 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
21945 earlier and only if MASK_FPU is set. Adjust formatting.
21947 2017-07-27 Martin Liska <mliska@suse.cz>
21949 * opt-functions.awk: Add validation of value of Init.
21950 * optc-gen.awk: Pass new argument.
21952 2017-07-27 Martin Liska <mliska@suse.cz>
21954 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
21955 Fix wrong condition.
21957 2017-07-27 Martin Liska <mliska@suse.cz>
21959 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
21960 BBs and edges seen by autoFDO.
21962 2017-07-27 Richard Biener <rguenther@suse.de>
21964 PR tree-optimization/81502
21965 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
21966 with incompatible but same sized type.
21967 (execute_update_addresses_taken): Likewise.
21969 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
21971 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
21972 flag_tree_loop_vectorize rather than flag_tree_vectorize.
21974 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21977 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
21978 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
21979 Change s_operand to memory_operand.
21981 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
21983 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
21984 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
21985 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
21986 Emit instructions rather than returning an expression. Handle TFmode
21987 and KFmode by casting to TImode.
21988 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
21989 (rs6000_emit_le_vsx_store): Likewise.
21990 * config/rs6000/vsx.md (VSX_TI): New iterator.
21991 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
21992 (*vsx_le_undo_permute_<mode>): Likewise.
21993 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
21994 emit the split sequence.
21995 (*vsx_le_perm_store_<mode>): Likewise.
21997 2017-07-27 Jakub Jelinek <jakub@redhat.com>
21999 PR tree-optimization/81555
22000 PR tree-optimization/81556
22001 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22002 if true, force CHANGED for the recursive invocation.
22003 (reassociate_bb): Remember original length of ops array, pass
22004 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22006 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22007 attributes for noipa attribute. For naked attribute use
22008 lookup_attribute first before lookup_attribute_spec.
22009 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22011 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22012 for functions with noipa attribute.
22013 (cgraph_externally_visible_p): Return true for functions with noipa
22015 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22016 for functions with noipa attribute.
22017 * doc/extend.texi: Document noipa function attribute.
22018 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22019 also for functions with noipa attribute.
22020 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22022 2017-07-26 Andrew Pinski <apinski@cavium.com>
22024 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22025 vec_unalign_load_cost and vec_unalign_store_cost.
22027 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22029 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22030 -mvsx-small-integer option.
22031 (ISA_3_0_MASKS_IEEE): Likewise.
22032 (OTHER_VSX_VECTOR_MASKS): Likewise.
22033 (POWERPC_MASKS): Likewise.
22034 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22035 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22036 code, only testing for DImode being allowed in non-VSX floating
22038 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22039 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22041 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22042 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22043 TARGET_P8_VECTOR test.
22044 (rs6000_secondary_reload_simple_move): Likewise.
22045 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22046 since TARGET_P9_VECTOR was already tested.
22047 (rs6000_opt_masks): Remove -mvsx-small-integer.
22048 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22049 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22051 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22052 test for TARGET_VEXTRACTUB was used, and that uses
22054 (p9 extract splitter): Likewise.
22055 (vsx_extract_<mode>_di_p9): Likewise.
22056 (vsx_extract_<mode>_store_p9): Likewise.
22057 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22058 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22059 the elimination of TARGET_VSX_SMALL_INTEGER.
22060 (vsx_extract_<mode>_p8): Likewise.
22061 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22062 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22063 (vsx_set_<mode>_p9): Likewise.
22064 (vsx_set_v4sf_p9): Likewise.
22065 (vsx_set_v4sf_p9_zero): Likewise.
22066 (vsx_insert_extract_v4sf_p9): Likewise.
22067 (vsx_insert_extract_v4sf_p9_2): Likewise.
22068 * config/rs6000/rs6000.md (sign extend splitter): Change
22069 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22070 (floatsi<mode>2_lfiwax_mem): Likewise.
22071 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22072 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22073 since a test for TARGET_P9_VECTOR was used.
22074 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22075 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22076 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22077 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22078 TARGET_P8_VECTOR test.
22079 (fix_trunc<mode>si2_stfiwx): Likewise.
22080 (fix_trunc<mode>si2_internal): Likewise.
22081 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22082 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22084 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22085 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22086 TARGET_P8_VECTOR test.
22087 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22088 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22089 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22091 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22092 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22093 since a test for TARGET_P9_VECTOR was used.
22094 (splitter for loading small constants): Likewise.
22096 2017-07-26 Andrew Pinski <apinski@cavium.com>
22098 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22101 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22104 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22105 (fp_valid_at): Likewise.
22107 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22109 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22110 (qdf24xx_addrcost_table): Likewise.
22111 (cortexa57_tunings): Update to use generic_branch_cost.
22112 (cortexa72_tunings): Likewise.
22113 (cortexa73_tunings): Likewise.
22114 (qdf24xx_tunings): Likewise.
22116 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22118 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22119 (thunderx2t99_branch_cost): Likewise.
22120 (cortexa35_tunings): Update to use generic_branch_cost.
22121 (cortexa53_tunings): Likewise.
22122 (cortexa57_tunings): Likewise.
22123 (cortexa72_tunings): Likewise.
22124 (cortexa73_tunings): Likewise.
22125 (thunderx2t99_tunings): Likewise.
22127 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22129 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22130 (sparc_option_override): Honour MASK_FSMULD.
22131 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22132 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22133 * config/sparc/sparc.opt (mfsmuld): New option.
22134 * doc/invoke.texi (mfsmuld): Document option.
22136 2017-07-26 Marek Polacek <polacek@redhat.com>
22138 PR middle-end/70992
22139 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22141 2017-07-26 Richard Biener <rguenther@suse.de>
22143 * gimple-match-head.c (do_valueize): Return OP if valueize
22145 (get_def): New helper to get at the def stmt of a SSA name
22146 if valueize allows.
22147 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22148 do_valueize to get at the def stmt.
22149 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22151 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22153 PR middle-end/46932
22154 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22156 2017-07-26 Martin Liska <mliska@suse.cz>
22159 * function.c (expand_function_start): Make expansion of
22160 nonlocal_goto_save_area after parm_birth_insn.
22162 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22164 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22165 from all CPU target flags enable members.
22167 2017-07-26 Richard Biener <rguenther@suse.de>
22169 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22170 (decision_tree::gen): Make 'type' const.
22171 (write_predicate): Likewise.
22173 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22175 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22176 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22177 (rs6000_option_override_internal): Likewise.
22178 (rs6000_expand_vector_set): Likewise.
22179 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22180 (TARGET_UPPER_REGS_SF): Likewise.
22181 (TARGET_UPPER_REGS_DI): Likewise.
22182 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22183 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22184 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22185 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22186 (Splitters for DI constants in Altivec registers): Likewise.
22187 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22188 (vsx_set_v4sf_p9): Likewise.
22189 (vsx_set_v4sf_p9_zero): Likewise.
22190 (vsx_insert_extract_v4sf_p9): Likewise.
22191 (vsx_insert_extract_v4sf_p9_2): Likewise.
22193 2017-07-25 Carl Love <cel@us.ibm.com>
22195 * doc/extend.texi: Update the built-in documentation file for the
22196 existing built-in functions
22197 vector signed char vec_cnttz (vector signed char);
22198 vector unsigned char vec_cnttz (vector unsigned char);
22199 vector signed short vec_cnttz (vector signed short);
22200 vector unsigned short vec_cnttz (vector unsigned short);
22201 vector signed int vec_cnttz (vector signed int);
22202 vector unsigned int vec_cnttz (vector unsigned int);
22203 vector signed long long vec_cnttz (vector signed long long);
22204 vector unsigned long long vec_cnttz (vector unsigned long long);
22206 2017-07-25 Andrew Pinski <apinski@cavium.com>
22208 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22209 accesses where the use is for the first operand of a BIT_INSERT.
22211 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22214 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22215 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22217 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22219 * config/i386/gstabs.h: Delete.
22220 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22222 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22224 * config/i386/i386.c (ix86_decompose_address): Do not check for
22225 register RTX when looking at index_reg or base_reg.
22226 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22228 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22230 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22231 to update EH info here.
22233 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22235 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22237 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22239 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22241 2017-07-25 Torsten Duwe <duwe@suse.de>
22243 * common.opt: Introduce -fpatchable-function-entry
22244 command line option, and its variables function_entry_patch_area_size
22245 and function_entry_patch_area_start.
22246 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22247 including a two-value parser.
22248 * target.def (print_patchable_function_entry): New target hook.
22249 * targhooks.h (default_print_patchable_function_entry): New function.
22250 * targhooks.c (default_print_patchable_function_entry): Likewise.
22251 * toplev.c (process_options): Switch off IPA-RA if
22252 patchable function entries are being generated.
22253 * varasm.c (assemble_start_function): Look at the
22254 patchable-function-entry command line switch and current
22255 function attributes and maybe generate NOP instructions by
22256 calling the print_patchable_function_entry hook.
22257 * doc/extend.texi: Document patchable_function_entry attribute.
22258 * doc/invoke.texi: Document -fpatchable_function_entry
22259 command line option.
22260 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22262 * doc/tm.texi: Re-generate.
22264 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22267 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22268 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22270 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22272 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22274 2017-07-25 Richard Biener <rguenther@suse.de>
22276 PR tree-optimization/81455
22277 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22278 not walk in cycles when looking for guards.
22280 2017-07-25 Richard Biener <rguenther@suse.de>
22282 PR tree-optimization/81529
22283 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22284 when optimizing backedge uses.
22286 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22288 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22290 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22291 to dl_section_ref. On AIX, append an expression to subtract
22292 the size of the section length to dl_section_ref.
22294 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22296 * configure.ac: If any of the config.* scripts fail, exit 1.
22297 * configure: Regenerate.
22299 2017-07-25 Richard Biener <rguenther@suse.de>
22301 PR middle-end/81546
22302 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22303 of immediate uses, be more verbose on errors.
22305 2017-07-25 Richard Biener <rguenther@suse.de>
22307 PR tree-optimization/81510
22308 * tree-vect-loop.c (vect_is_simple_reduction): When the
22309 reduction stmt is not inside the loop bail out.
22311 2017-07-25 Richard Biener <rguenther@suse.de>
22313 PR tree-optimization/81303
22314 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22315 profitability check against LOOP_VINFO_NITERSM1.
22317 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22319 * domwalk.c (cmp_bb_postorder): Simplify.
22320 (sort_bbs_postorder): New function. Use it...
22321 (dom_walker::walk): ...here to optimize common cases.
22323 2017-07-25 Martin Liska <mliska@suse.cz>
22326 * ipa-visibility.c (function_and_variable_visibility): Make the
22327 redirection just on target that supports aliasing.
22328 Fix GNU coding style.
22330 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22333 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22335 * config/arm/rtems.h: Likewise.
22336 * config/bfin/rtems.h: Likewise.
22337 * config/i386/rtemself.h: Likewise.
22338 * config/lm32/rtems.h: Likewise.
22339 * config/m32c/rtems.h: Likewise.
22340 * config/m68k/rtemself.h: Likewise.
22341 * config/microblaze/rtems.h: Likewise.
22342 * config/mips/rtems.h: Likewise.
22343 * config/moxie/rtems.h: Likewise.
22344 * config/nios2/rtems.h: Likewise.
22345 * config/powerpcspe/rtems.h: Likewise.
22346 * config/rs6000/rtems.h: Likewise.
22347 * config/rtems.h: Likewise.
22348 * config/sh/rtems.h: Likewise.
22349 * config/sh/rtemself.h: Likewise.
22350 * config/sparc/rtemself.h: Likewise.
22352 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22355 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22356 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22357 * tree-ssa-structalias.c (alias_get_name): Same.
22359 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22362 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22363 instructions if no du chain is found.
22365 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22367 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22369 2017-07-25 Richard Biener <rguenther@suse.de>
22371 PR middle-end/81505
22372 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22375 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22377 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22378 upper-regs options.
22379 (ISA_2_7_MASKS_SERVER): Likewise.
22380 (ISA_3_0_MASKS_IEEE): Likewise.
22381 (OTHER_P8_VECTOR_MASKS): Likewise.
22382 (OTHER_VSX_VECTOR_MASKS): Likewise.
22383 (POWERPC_MASKS): Likewise.
22384 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22385 duplicate list of options.
22386 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22387 explicit -mupper-regs options.
22388 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22389 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22390 alias for -mupper-regs-df.
22391 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22392 (rs6000_init_hard_regno_mode_ok): Likewise.
22393 (rs6000_option_override_internal): Likewise.
22394 (rs6000_opt_masks): Likewise.
22395 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22396 options in terms of whether -mvsx or -mpower8-vector was used.
22397 (TARGET_UPPER_REGS_DI): Likewise.
22398 (TARGET_UPPER_REGS_SF): Likewise.
22399 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22400 -mupper-regs-* options.
22402 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22404 * passes.c (emergency_dump_function): Print some empty lines and a
22405 header before the RTL dump.
22407 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22409 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22411 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22414 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22415 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22417 2017-07-24 Carl Love <cel@us.ibm.com>
22419 * config/rs6000/rs6000-c.c: Add support for built-in functions
22420 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22421 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22422 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22423 vec_extract_fp_from_shortl): Add defines for the two builtins.
22424 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22425 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22427 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22428 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22429 * doc/extend.texi: Update the built-in documentation file for the
22430 new built-in function.
22432 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22435 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22437 * doc/generic.texi: Likewise.
22438 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22439 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22441 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22443 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22444 (aarch64_mls_elt_merge<mode>): Likewise.
22446 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22448 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22449 having __cxa_atexit.
22451 2017-07-23 Michael Collison <michael.collison@arm.com>
22453 * config/arm/arm.c (arm_option_override): Deprecate
22454 use of -mstructure-size-boundary.
22455 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22456 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22458 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22461 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22462 Reduce cost estimate for direct moves.
22464 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22467 * config/i386/i386.c (ix86_option_override_internal): Disable
22468 BMI, BMI2 and TBM instructions for -m16.
22470 2017-07-21 Carl Love <cel@us.ibm.com>
22472 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22473 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22474 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22475 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22476 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22477 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22478 VMULOSW): New enum "unspec" values.
22479 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22480 altivec_vmulosw): New patterns.
22481 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22482 VMULOSW): Add definitions.
22484 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22486 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22487 (qdf24xx): Likewise.
22488 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22489 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22490 (AARCH64_FL_V8_1): Renumber.
22491 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22492 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22493 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22494 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22495 rdma to feature modifiers list.
22497 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22499 PR middle-end/56727
22500 * ipa-visibility (function_and_variable_visibility): Convert
22501 recursive PLT call to direct call if appropriate.
22503 2017-07-21 Andrew Pinski <apinski@cavium.com>
22505 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22506 operand 1 to see if the types precision matches.
22507 * fold-const.c (operand_equal_p): Likewise.
22509 2017-07-21 Richard Biener <rguenther@suse.de>
22511 PR tree-optimization/81303
22512 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22513 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22514 (vect_peeling_hash_get_lowest_cost): Adjust.
22515 (vect_enhance_data_refs_alignment): Likewise. Use
22516 vect_get_peeling_costs_all_drs to compute the penalty for no
22517 peeling to match up costs.
22519 2017-07-21 Richard Biener <rguenther@suse.de>
22521 PR tree-optimization/81500
22522 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22523 we didn't identify a reduction path.
22525 2017-07-21 Tom de Vries <tom@codesourcery.com>
22526 Cesar Philippidis <cesar@codesourcery.com>
22528 PR gcov-profile/81442
22529 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22532 2017-07-21 Tom de Vries <tom@codesourcery.com>
22535 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22537 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22538 nvptx_override_options_after_change.
22540 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22542 * dwarf2out.c (output_file_names): Avoid double testing for
22543 dwarf_version >= 5.
22545 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22547 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22549 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22551 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22553 (try_crossjump_to_edge): Do not punt on partitioned functions.
22555 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22557 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22558 Put all BBs reachable only via paths crossing cold region to cold
22560 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22562 2016-07-21 Richard Biener <rguenther@suse.de>
22564 PR tree-optimization/81303
22565 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22566 into account prologue and epilogue iterations when raising
22567 min_profitable_iters to sth at least covering one vector iteration.
22569 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22571 * config/arm/arm.c (arm_test_cpu_arch_dat):
22574 2017-07-20 Nathan Sidwell <nathan@acm.org>
22576 Remove TYPE_METHODS.
22577 * tree.h (TYPE_METHODS): Delete.
22578 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22579 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22580 (dbxout_type_methods): Scan TYPE_FIELDS.
22581 (dbxout_type): Don't check TYPE_METHODS here.
22582 * function.c (use_register_for_decl): Always ignore register for
22583 class types when not optimizing.
22584 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22585 * tree.c (free_lang_data_in_type): Stitch out member functions and
22586 templates from TYPE_FIELDS.
22587 (build_distinct_type_copy, verify_type_variant,
22588 verify_type): Member fns are on TYPE_FIELDS.
22589 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22590 * tree-pretty-print.c (dump_generic_node): Likewise.
22592 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22595 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22596 V2TImode and V4TImode.
22597 (ix86_expand_vector_extract): Likewise.
22598 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22599 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22600 (ssescalarmode): Handle V4TImode and V2TImode.
22601 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22602 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22603 (VEXTRACTI128_MODE): New mode iterator.
22604 (splitter for *vec_extractv?ti first element): New.
22605 (VEC_INIT_MODE): New mode iterator.
22606 (vec_init<mode>): Consolidate 3 expanders into one using
22607 VEC_INIT_MODE mode iterator.
22609 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22611 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22612 non_spilled_static_chain_regno_p.
22614 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22616 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22618 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22620 * bb-reorder.c (connect_traces): Allow copying of blocks within
22623 2017-07-20 Richard Biener <rguenther@suse.de>
22625 * gimple.h (gimple_phi_result): Add gphi * overload.
22626 (gimple_phi_result_ptr): Likewise.
22627 (gimple_phi_arg): Likewise. Adjust index assert to only
22628 allow actual argument accesses rather than all slots available
22630 (gimple_phi_arg_def): Add gphi * overload.
22631 * tree-phinodes.c (make_phi_node): Initialize only actual
22633 (resize_phi_node): Clear memory not covered by old node,
22634 do not initialize excess argument slots.
22635 (reserve_phi_args_for_new_edge): Initialize new argument slot
22638 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22640 PR tree-optimization/81388
22642 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22644 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22645 by removing computation of may_be_zero.
22647 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22648 Tom de Vries <tom@codesourcery.com>
22650 PR middle-end/81030
22651 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22652 when gimple level profile disagrees with what RTL expander did.
22654 2017-07-20 Richard Biener <rguenther@suse.de>
22656 PR tree-optimization/61171
22657 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22658 (vect_analyze_stmt): Add slp instance parameter.
22659 (vectorizable_reduction): Likewise.
22660 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22661 (vect_is_simple_reduction): Deal with chains not detected
22662 as SLP reduction chain, specifically not properly associated
22663 chains containing a mix of plus/minus.
22664 (get_reduction_op): Remove.
22665 (get_initial_defs_for_reduction): Simplify, pass in whether
22666 this is a reduction chain, pass in the SLP node for the PHIs.
22667 (vect_create_epilog_for_reduction): Get the SLP instance as
22669 (vectorizable_reduction): Get the SLP instance as arg.
22670 During analysis remember the SLP node with the PHIs in the
22671 instance. Simplify getting at the vectorized reduction PHIs.
22672 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22673 through SLP instance.
22674 (vect_slp_analyze_operations): Likewise.
22675 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22676 (vect_transform_stmt): Likewise.
22678 2017-07-20 Tom de Vries <tom@codesourcery.com>
22680 PR tree-optimization/81489
22681 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22682 read of phi arg location to before loop that modifies phi.
22684 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22686 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22689 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22691 PR middle-end/81331
22692 * except.c (execute): Fix ordering issue.
22694 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22696 PR rtl-optimization/81423
22697 * combine.c (make_compound_operation_int): Don't try to optimize
22698 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22700 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22702 PR rtl-optimization/81423
22703 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22704 with a constant that is -1 in the truncated to mode.
22706 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22708 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22709 (determine_unlikely_bbs): ... here.
22710 * predict.h (propagate_unlikely_bbs_forward): Declare.
22711 * cfgexpand.c (pass_expand::execute): Use it.
22712 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22714 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22715 propagate_unlikely_bbs_forward.
22717 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22719 PR middle-end/81331
22720 * except.c (maybe_add_nop_after_section_switch): New function.
22723 2017-07-19 Tom de Vries <tom@codesourcery.com>
22725 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22727 2017-07-19 Tom de Vries <tom@codesourcery.com>
22729 * gimple.h (gimple_phi_arg): Make assert more strict.
22731 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22733 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22734 * config/rs6000/mmintrin.h: New file.
22735 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22737 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22739 PR tree-optimization/81346
22740 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22742 2017-07-19 Tom de Vries <tom@codesourcery.com>
22744 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22746 2017-07-19 Tom de Vries <tom@codesourcery.com>
22748 * config/nvptx/nvptx-modes.def: Add V2DImode.
22749 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22750 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22751 (nvptx_output_mov_insn): Handle lack of mov.b128.
22752 (nvptx_print_operand): Handle 'H' and 'L' codes.
22753 (nvptx_vector_mode_supported): Allow V2DImode.
22754 (nvptx_preferred_simd_mode): New function.
22755 (nvptx_data_alignment): New function.
22756 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22757 nvptx_preferred_simd_mode.
22758 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22760 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22762 2017-07-19 Tom de Vries <tom@codesourcery.com>
22764 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22765 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22766 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22767 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22768 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22769 (mov<VECIM>_insn): New define_insn.
22770 (define_expand "mov<VECIM>): New define_expand.
22772 2017-07-19 Tom de Vries <tom@codesourcery.com>
22774 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22776 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22778 PR tree-optimization/81346
22779 * fold-const.h (fold_div_compare, range_check_type): Declare.
22780 * fold-const.c (range_check_type): New function.
22781 (build_range_check): Use range_check_type.
22782 (fold_div_compare): No longer static, rewritten into
22783 a match.pd helper function.
22784 (fold_comparison): Don't call fold_div_compare here.
22785 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22786 as helper function.
22788 2017-07-19 Nathan Sidwell <nathan@acm.org>
22790 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22791 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22792 * tree.c (find_decls_types_r, verify_type): Use
22793 TYPE_{MIN,MAX}_VALUE_RAW.
22794 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22795 (hash_tree): Likewise.
22796 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22798 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22801 2017-07-18 Tom de Vries <tom@codesourcery.com>
22803 PR middle-end/81464
22804 * omp-expand.c (expand_omp_for_static_chunk): Handle
22805 equal-argument loop exit phi.
22807 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22810 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22811 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22812 operand 2 predicate.
22813 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22814 operand 2 predicate.
22815 (ror,rol -> rorx splitters): Use const_int_operand as
22816 operand 2 predicate.
22818 2017-06-18 Richard Biener <rguenther@suse.de>
22820 PR tree-optimization/81410
22821 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22822 the gap in the ! slp_perm SLP case after each group.
22824 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22826 PR middle-end/81463
22827 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22830 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22832 PR middle-end/81462
22833 * predict.c (set_even_probabilities): Cleanup; do not affect
22834 probabilities that are already known.
22835 (combine_predictions_for_bb): Call even when count is set.
22837 2017-07-18 Nathan Sidwell <nathan@acm.org>
22839 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22842 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22845 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22846 optimization for loop niter analysis.
22848 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22851 * config/avr/avr.c (avr_optimize_casesi): Don't use
22852 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22854 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22856 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22857 body_cost_vec from _vect_peel_extended_info.
22858 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
22859 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
22862 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22864 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
22866 2017-07-18 Richard Biener <rguenther@suse.de>
22868 PR tree-optimization/80620
22869 PR tree-optimization/81403
22870 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
22871 info when re-using a VN table entry.
22873 2017-07-18 Richard Biener <rguenther@suse.de>
22875 PR tree-optimization/81418
22876 * tree-vect-loop.c (vectorizable_reduction): Properly compute
22877 vectype_in. Verify that with lane-reducing reduction operations
22878 we have a single def-use cycle.
22880 2017-07-17 Carl Love <cel@us.ibm.com>
22882 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
22884 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22885 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22886 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22887 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22888 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22889 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22890 VMULOSW): New enum "unspec" values.
22891 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22892 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22893 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22894 altivec_vmulosw): New patterns.
22895 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22896 VMULOSW): Add definitions.
22898 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
22900 * config/alpha/alpha.c: Include predict.h.
22902 2017-07-17 Yury Gribov <tetra2005@gmail.com>
22904 * tree-vrp.c (compare_assert_loc): Fix comparison function
22905 to return predictable results.
22907 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22909 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
22911 (subdi3): Likewise.
22912 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
22913 * doc/invoke.texi (mexpand-adddi): Update text.
22915 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22917 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
22918 that also clobbers the CC register. The old expand code is moved
22920 (*arc_clzsi2): ... here.
22921 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
22922 the CC register. The old expand code is moved to ...
22923 (arc_ctzsi2): ... here.
22925 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22927 * config/arc/arc.opt (mindexed-loads): Use initial value
22928 TARGET_INDEXED_LOADS_DEFAULT.
22929 (mauto-modify-reg): Use initial value
22930 TARGET_AUTO_MODIFY_REG_DEFAULT.
22931 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22932 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22933 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22934 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22936 2017-07-17 Martin Liska <mliska@suse.cz>
22939 * opts.c (finish_options): Do not allow -fgnu-tm
22940 w/ -fsanitize={kernel-,}address. Say sorry.
22942 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22945 * tree-loop-distribution.c (classify_partition): Only assert on
22946 numer of iterations.
22947 (merge_dep_scc_partitions): Delete prameter. Update function call.
22948 (distribute_loop): Remove code handling loop with unknown niters.
22949 (pass_loop_distribution::execute): Skip loop with unknown niters.
22951 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22954 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
22955 function sort_partitions_by_post_order.
22957 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22959 PR tree-optimization/81374
22960 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
22961 the max index of basic blocks, rather than number of basic blocks.
22963 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22965 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
22967 (arc_legitimate_pic_operand_p): Likewise.
22968 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
22970 (arc_needs_pcl_p): Likewise.
22971 (arc_legitimate_pc_offset_p): Likewise.
22972 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
22973 function is also used in constrains.md.
22974 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
22975 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
22976 PLUS. Only return true/false in known cases, otherwise assert.
22977 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
22978 is already called in arc_legitimate_constant_p.
22979 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
22981 (LEGITIMATE_PIC_OPERAND_P): Use
22982 arc_raw_symbolic_reference_mentioned_p function.
22983 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
22988 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22989 Andrew Burgess <andrew.burgess@embecosm.com>
22991 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
22992 (arc_return_address_register): New function.
22993 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
22994 (arc_handle_fndecl_attribute): Add naked attribute.
22995 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22996 (TARGET_WARN_FUNC_RETURN): Likewise.
22997 (arc_allocate_stack_slots_for_args): New function.
22998 (arc_warn_func_return): Likewise.
22999 (machine_function): Change type fn_type.
23000 (arc_compute_function_type): Consider new naked function type,
23001 change function return type.
23002 (arc_must_save_register): Adapt to handle new
23003 arc_compute_function_type's return type.
23004 (arc_expand_prologue): Likewise.
23005 (arc_expand_epilogue): Likewise.
23006 (arc_return_address_regs): Delete.
23007 (arc_return_address_register): New function.
23008 (arc_epilogue_uses): Use above function.
23009 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23010 (arc_function_type): Change encoding, add naked type.
23011 (ARC_INTERRUPT_P): Change to handle the new encoding.
23012 (ARC_FAST_INTERRUPT_P): Likewise.
23013 (ARC_NORMAL_P): Define.
23014 (ARC_NAKED_P): Likewise.
23015 (arc_compute_function_type): Delete prototype.
23016 * config/arc/arc.md (in_ret_delay_slot): Use
23017 arc_return_address_register function.
23018 (simple_return): Likewise.
23019 (p_return_i): Likewise.
23021 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23023 PR tree-optimization/81428
23024 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23025 can't be built for those types.
23027 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23029 Remove stuff dead since r239246.
23031 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23032 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23033 (avr_inform_devices): Remove dead stuff.
23035 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23037 * config/arm/arm_neon.h: Fix softp typo.
23039 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23041 PR tree-optimization/81365
23042 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23043 aggregate moves onto bb predecessor edges, make sure there are no
23044 loads that could alias the lhs in between the start of bb and the
23047 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23050 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23051 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23052 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23054 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23056 PR tree-optimization/81396
23057 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23058 (init_symbolic_number): Initialize it to 1.
23059 (perform_symbolic_merge): Add n_ops from both operands into the new
23061 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23062 without base_addr as useless if they need more than one operation.
23063 (bswap_replace): Handle !bswap case for NULL base_addr.
23065 2017-07-17 Tom de Vries <tom@codesourcery.com>
23068 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23071 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23073 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23074 conditional builtin define __FIX_LEON3FT_B2BST.
23076 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23078 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23079 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23082 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23084 PR rtl-optimization/81424
23085 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23086 to remove potential trapping from operands if -fnon-call-exceptions.
23088 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23090 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23091 profile_proability for scalling.
23092 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23094 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23096 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23098 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23100 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23101 fixpoint arithmetics.
23103 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23105 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23106 fixpoint arithmetics.
23108 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23110 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23111 fixpoint arithmetics.
23113 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23115 * profile-count.h (profile_probability::from_reg_br_prob_note,
23116 profile_probability::to_reg_br_prob_note): New functions.
23117 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23118 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23119 * predict.c (probability_reliable_p): Update.
23120 (edge_probability_reliable_p): Update.
23121 (br_prob_note_reliable_p): Update.
23122 (invert_br_probabilities): Update.
23123 (add_reg_br_prob_note): New function.
23124 (combine_predictions_for_insn): Update.
23125 * asan.c (asan_clear_shadow): Update.
23126 * cfgbuild.c (compute_outgoing_frequencies): Update.
23127 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23128 (update_br_prob_note): Update.
23129 (rtl_verify_edges): Update.
23130 (purge_dead_edges): Update.
23131 (fixup_reorder_chain): Update.
23132 * emit-rtl.c (try_split): Update.
23133 * ifcvt.c (cond_exec_process_insns): Update.
23134 (cond_exec_process_if_block): Update.
23135 (dead_or_predicable): Update.
23136 * internal-fn.c (expand_addsub_overflow): Update.
23137 (expand_neg_overflow): Update.
23138 (expand_mul_overflow): Update.
23139 * loop-doloop.c (doloop_modify): Update.
23140 * loop-unroll.c (compare_and_jump_seq): Update.
23141 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23142 * predict.h: Update.
23143 * reorg.c (mostly_true_jump): Update.
23145 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23146 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23147 * config/arc/arc.c: (emit_unlikely_jump): Update.
23148 * config/arm/arm.c: (emit_unlikely_jump): Update.
23149 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23150 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23151 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23152 (ix86_print_operand): Update.
23153 (ix86_split_fp_branch): Update.
23154 (predict_jump): Update.
23155 * config/ia64/ia64.c (ia64_print_operand): Update.
23156 * config/mmix/mmix.c (mmix_print_operand): Update.
23157 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23158 (rs6000_expand_split_stack_prologue): Update.
23159 * config/rs6000/rs6000.c: Update.
23160 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23161 (s390_expand_vec_movstr): Update.
23162 (s390_expand_cs_tdsi): Update.
23163 (s390_expand_split_stack_prologue): Update.
23164 * config/sh/sh.c (sh_print_operand): Update.
23165 (expand_cbranchsi4): Update.
23166 (expand_cbranchdi4): Update.
23167 * config/sparc/sparc.c (output_v9branch): Update.
23168 * config/spu/spu.c (get_branch_target): Update.
23169 (ea_load_store_inline): Update.
23170 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23171 * config/tilepro/tilepro.c: Update.
23173 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23175 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23176 (gimplify_save_expr): Likewise.
23178 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23180 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23182 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23184 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23186 * timevar.def (TV_IPA_FNSUMMARY): Define.
23188 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23190 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23191 to back store errata sensitive sequence from being generated.
23192 (sqrtdf2_fix): Likewise.
23194 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23196 * tree-ssa-threadupdate.c (compute_path_counts,
23197 update_joiner_offpath_counts): Use profile_probability.
23199 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23202 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23204 * config/arm/arm-c.c (arm_cpu_builtins): Define
23205 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23207 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23209 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23210 array entries to represent __ieee128 versions of the
23211 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23212 scalar_extract_sig, and scalar_insert_exp built-in functions.
23213 (altivec_resolve_overloaded_builtin): Add special case handling
23214 for the __builtin_scalar_insert_exp function, as represented by
23215 the P9V_BUILTIN_VEC_VSIEDP constant.
23216 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23217 exponent support for __ieee128 argument.
23218 (VSESQP): Add scalar extract signature support for __ieee128
23220 (VSTDCNQP): Add scalar test negative support for __ieee128
23222 (VSIEQP): Add scalar insert exponent support for __int128 argument
23223 with __ieee128 result.
23224 (VSIEQPF): Add scalar insert exponent support for __ieee128
23225 argument with __ieee128 result.
23226 (VSTDCQP): Add scalar test data class support for __ieee128
23228 (VSTDCNQP): Add overload support for scalar test negative with
23229 __ieee128 argument.
23230 (VSTDCQP): Add overload support for scalar test data class
23231 __ieee128 argument.
23232 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23233 UNSPEC_VSX_SXSIGDP.
23234 (UNSPEC_VSX_SIEXPQP): New constant.
23235 (xsxexpqp): New insn for VSX scalar extract exponent quad
23237 (xsxsigqp): New insn for VSX scalar extract significand quad
23239 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23240 precision with floating point argument.
23241 (xststdcqp): New expand for VSX scalar test data class quad
23243 (xststdcnegqp): New expand for VSX scalar test negative quad
23245 (xststdcqp): New insn to match expansions for VSX scalar test data
23246 class quad precision and VSX scalar test negative quad precision.
23247 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23248 special case operand checking to enforce that second operand of
23249 VSX scalar test data class with quad precision argument is a 7-bit
23251 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23252 prototypes and descriptions of __ieee128 versions of
23253 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23254 scalar_test_data_class, and scalar_test_neg built-in functions.
23256 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23258 PR tree-optimization/81162
23259 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23260 replace a negate with an add.
23262 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23264 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23266 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23268 * config/arm/arm-c.c (arm_cpu_builtins): Define
23269 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23271 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23273 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23274 (armv8-r): Set ARM Cortex-R52 as default CPU.
23275 * config/arm/arm-tables.opt: Regenerate.
23276 * config/arm/arm-tune.md: Regenerate.
23277 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23279 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23280 extension for -mcpu=cortex-r52.
23282 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23284 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23285 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23286 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23287 (fp-armv8): Define it as FP_ARMv8 only.
23288 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23289 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23291 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23292 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23293 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23294 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23295 than TARGET_FPU_ARMV8.
23296 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23297 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23298 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23300 * config/arm/neon.md (neon_vrint): Likewise.
23301 (neon_vcvt): Likewise.
23302 (neon_<fmaxmin_op><mode>): Likewise.
23303 (<fmaxmin><mode>3): Likewise.
23304 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23305 * config/arm/predicates.md (arm_cond_move_operator): Check against
23306 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23308 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23310 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23311 to top of function.
23313 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23315 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23316 loop in comment with memset.
23318 2017-07-14 Martin Liska <mliska@suse.cz>
23320 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23321 * dwarf2out.c (is_java): Remove the function.
23322 (output_pubname): Remove usage of the function.
23323 (lower_bound_default): Remove usage of DW_LANG_Java.
23324 (gen_compile_unit_die): Likewise.
23325 * gcc.c: Remove compiler defaults for .java and .zip files.
23326 * gimple-expr.c (remove_suffix): Change as there's no longer
23327 extension than 4-letter one.
23328 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23329 (gimplify_save_expr): Likewise.
23330 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23331 as it's possible even for other languages than Java.
23332 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23333 * lto-opts.c (lto_write_options): Remove reference to Java.
23334 * opts.c (strip_off_ending): Update file extension handling.
23335 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23336 * tree-eh.c (lower_resx): Likewise.
23337 * tree.c (free_lang_data_in_type): Remove dead code.
23338 (find_decls_types_r): Likewise.
23339 (build_common_builtin_nodes): Remove Java from a comment.
23340 (verify_type): Remove dead code.
23341 * varasm.c (assemble_external): Remove Java from a comment.
23343 2017-07-14 Martin Liska <mliska@suse.cz>
23345 * opts.c (finish_options): Add quotes.
23346 (common_handle_option): Likewise.
23348 2017-07-14 Martin Liska <mliska@suse.cz>
23350 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23351 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23352 Remove N_SO_PASCAL.
23353 * dwarf2out.c (lower_bound_default): Do not handle
23355 (gen_compile_unit_die): Likewise.
23356 * gcc.c: Remove default extension binding for GNU Pascal.
23357 * stmt.c: Remove Pascal language from a comment.
23358 * xcoffout.c: Likewise.
23360 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23363 * diagnostic-show-locus.c (fixit_cmp): New function.
23364 (layout::layout): Sort m_fixit_hints.
23365 (column_range::column_range): Assert that the values are valid.
23366 (struct char_span): New struct.
23367 (correction::overwrite): New method.
23368 (struct source_line): New struct.
23369 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23370 calls in terms of classes source_line and char_span, and
23371 correction::overwrite.
23372 (selftest::test_overlapped_fixit_printing_2): New function.
23373 (selftest::diagnostic_show_locus_c_tests): Call it.
23375 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23377 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23378 early if there is no lhs.
23380 2017-07-13 Martin Liska <mliska@suse.cz>
23382 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23383 (gen_reference_type_die): Likewise.
23384 * stor-layout.c: Remove Pascal-related comment.
23386 2017-07-13 Martin Liska <mliska@suse.cz>
23388 * opts.c (finish_options): Add quotes to error messages.
23389 (parse_sanitizer_options): Likewise.
23391 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23393 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23395 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23397 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23399 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23401 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23403 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23405 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23408 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23409 provides the hardware capability bits, define the macro
23410 __BUILTIN_CPU_SUPPORTS__.
23411 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23412 if GLIBC does not provide the hardware capability bits. Add a
23413 gcc_unreachable call if the built-in cpu function is neither
23414 __builtin_cpu_is nor __builtin_cpu_supports.
23415 (rs6000_get_function_versions_dispatcher): Change the warning
23416 that an old GLIBC is used which does not export the capability
23417 bits to be an error.
23418 * doc/extend.texi (target_clones attribute): Document the
23419 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23420 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23421 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23422 the macros defined by GCC if the newer GLIBC is available.
23424 2017-07-12 Jeff Law <law@redhat.com>
23426 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23427 remaining includes slightly.
23428 * config/riscv/riscv-builtins.c: Include profile-count.h.
23430 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23433 * config/avr/avr.c (avr_set_current_function): In diagnostic
23434 messages: Quote keywords and (parts of) identifiers.
23435 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23438 2017-07-12 Carl Love <cel@us.ibm.com>
23440 * config/rs6000/rs6000-c.c: Add support for built-in functions
23441 vector bool char vec_revb (vector bool char);
23442 vector bool short vec_revb (vector short char);
23443 vector bool int vec_revb (vector bool int);
23444 vector bool long long vec_revb (vector bool long long);
23445 * doc/extend.texi: Update the built-in documentation file for the
23446 new built-in functions.
23448 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23450 * config/s390/s390.md: Remove movcc splitter.
23452 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23454 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23455 load/store on condition.
23457 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23460 * config/avr/avr.c (avr_encode_section_info)
23461 [progmem && !TREE_READONLY]: Error if progmem object needs
23464 2017-07-11 Michael Collison <michael.collison@arm.com>
23466 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23469 2017-07-11 Carl Love <cel@us.ibm.com>
23471 * config/rs6000/rs6000-c.c: Add support for builtins
23472 vector unsigned int vec_parity_lsbb (vector signed int);
23473 vector unsigned int vec_parity_lsbb (vector unsigned int);
23474 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23475 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23476 vector unsigned long long vec_parity_lsbb (vector signed long long);
23477 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23478 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23479 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23480 * doc/extend.texi: Update the built-in documentation file for the
23481 new built-in functions.
23483 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23485 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23486 (layout::m_primary_loc): New field.
23487 (layout::layout): Initialize new field. Move location filtering
23488 logic from here to...
23489 (layout::maybe_add_location_range): ...this new method. Add
23490 support for filtering to just the lines already specified by other
23492 (layout::will_show_line_p): New method.
23493 (gcc_rich_location::add_location_if_nearby): New method.
23494 (selftest::test_add_location_if_nearby): New test function.
23495 (selftest::diagnostic_show_locus_c_tests): Call it.
23496 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23499 2017-07-11 Tom de Vries <tom@codesourcery.com>
23501 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23502 (bb_first_real_insn): New function.
23503 (nvptx_single): Add extra initialization of broadcasted condition
23506 2017-07-11 Nathan Sidwell <nathan@acm.org>
23508 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23510 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23512 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23513 Binutils doc as TEXI will mess them up.
23514 * doc/invoke.texi (AVR Options): Same here.
23516 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23518 * config/sparc/sparc.opt (mfix-ut700): New option.
23519 (mfix-gr712rc): Likewise.
23520 (sparc_fix_b2bst): New variable.
23521 * doc/invoke.texi (SPARC options): Document them.
23522 (ARM options): Fix warnings.
23523 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23524 instructions to prevent sequences that can trigger the store-store
23525 errata for certain LEON3FT processors.
23526 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23527 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23528 * config/sparc/sparc.md (fix_b2bst): New attribute.
23529 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23531 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23534 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23536 (*rsqrtsf2_sse): Ditto.
23538 (div<mode>3): Macroize insn from divdf3 and divsf3
23539 using MODEF mode iterator.
23541 2017-07-10 Martin Sebor <msebor@redhat.com>
23543 PR tree-optimization/80397
23544 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23545 instead of testing for equality to INTEGER_TYPE.
23547 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23549 * config.gcc: Remove uclibc from arc target spec.
23551 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23553 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23555 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23558 * lto-wrapper.c (remove_option): New function.
23559 (merge_and_complain): Merge PIC/PIE options more realistically.
23561 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23563 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23567 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23568 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23569 * config.in: Regenerate.
23570 * configure: Regenerate.
23571 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23572 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23573 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23574 (TARGET_GASISR_PROLOGUES): ...target mask.
23575 * common/config/avr/avr-common.c
23576 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23577 Set -mgas-isr-prologues.
23578 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23579 INSERT_PASS_BEFORE for it.
23580 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23581 * config/avr/avr.c (avr_option_override)
23582 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23583 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23584 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23585 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23586 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23588 (make_avr_pass_pre_proep): New function.
23589 (emit_push_sfr) <treg>: Add argument to function and use it
23590 instead of TMP_REG.
23591 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23592 and set machine->gasisr.yes.
23593 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23594 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23595 __gcc_isr.n_pushed to .L__stack_usage.
23596 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23597 (avr_asm_final_postscan_insn): ...this new static function.
23598 * config/avr/avr.h (machine_function)
23599 <is_no_gccisr, use_L__stack_usage>: New fields.
23600 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23601 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23602 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23603 (gasisr, *gasisr): New expander and insn.
23604 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23605 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23606 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23608 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23610 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23613 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23615 Move jump-tables out of .text again.
23618 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23619 (ASM_OUTPUT_ADDR_VEC): New function.
23620 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23621 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23622 INSN_ADDRESSes as asm comment.
23623 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23624 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23625 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23626 * config/avr/avr.md (*tablejump): Adjust comment.
23627 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23628 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23630 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23631 (avr_output_addr_vec): New proto.
23632 (avr_log_t) <insn_addresses>: New field.
23634 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23637 * config/i386/i386.c (ix86_function_arg_advance): Set
23638 outgoing_args_on_stack to true if there are outgoing arguments
23640 (ix86_function_arg): Likewise.
23641 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23642 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23643 * config/i386/i386.h (machine_function): Add
23644 outgoing_args_on_stack.
23646 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23648 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23649 supporting pthreds.
23650 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23652 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23654 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23655 (REAL_H): Remove $(MACHMODE_H).
23656 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23658 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23659 $(MACHMODE_H) and double-int.h.
23660 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23662 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23665 2017-07-07 Andrew Pinski <apinski@cavium.com>
23667 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23668 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23670 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23672 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23673 Add warning if GCC was not configured to link against a GLIBC that
23674 exports the hardware capability bits.
23675 (make_resolver_func): Make resolver function private and not a
23676 COMDAT function. Create the name with clone_function_name instead
23677 of make_unique_name.
23680 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23681 correct operand in doing the split.
23683 2017-07-07 Carl Love <cel@us.ibm.com>
23685 * config/rs6000/rs6000-c: Add support for built-in function
23686 vector unsigned short vec_pack_to_short_fp32 (vector float,
23688 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23689 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23690 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23691 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23692 (convert_4f32_8i16): Add define_expand.
23693 * doc/extend.texi: Update the built-in documentation file for the
23694 new built-in function.
23696 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23698 * config/sparc/m8.md: New file.
23699 * config/sparc/sparc.md: Include m8.md.
23701 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23703 * config/sparc/sparc.opt: New option -mvis4b.
23704 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23705 (sparc_option_override): Handle VIS4B.
23706 (enum sparc_builtins): Define
23707 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23708 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23709 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23710 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23711 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23712 (check_constant_argument): New function.
23713 (sparc_vis_init_builtins): Define builtins
23714 __builtin_vis_dictunpack{8,16,32},
23715 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23716 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23717 __builtin_vis_fpcmpde{8,16,32}shl and
23718 __builtin_vis_fpcmpur{8,16,32}shl.
23719 (sparc_expand_builtin): Check that the constant operands to
23720 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23721 constant and in range.
23722 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23724 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23725 (SPARC_IMM5_P): Likewise.
23726 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23727 (enabled): Handle vis4b.
23728 (UNSPEC_DICTUNPACK): New unspec.
23729 (UNSPEC_FPCMPSHL): Likewise.
23730 (UNSPEC_FPUCMPSHL): Likewise.
23731 (UNSPEC_FPCMPDESHL): Likewise.
23732 (UNSPEC_FPCMPURSHL): Likewise.
23733 (cpu_feature): New CPU feature `vis4b'.
23734 (dictunpack{8,16,32}): New insns.
23735 (FPCSMODE): New mode iterator.
23736 (fpcscond): New code iterator.
23737 (fpcsucond): Likewise.
23738 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23739 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23740 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23741 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23742 * config/sparc/constraints.md: Define constraints `q' for unsigned
23743 2-bit integer constants and `t' for unsigned 5-bit integer
23745 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23747 (imm5_operand_dictunpack16): Likewise.
23748 (imm5_operand_dictunpack32): Likewise.
23749 (imm2_operand): Likewise.
23750 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23751 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23752 ditunpack* and fpcmp*shl builtins.
23754 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23756 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23757 * config.in: Add HAVE_AS_SPARC6 define.
23758 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23760 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23762 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23763 (CPP_CPU_SPEC): Handle m8.
23764 (ASM_CPU_SPEC): Likewise.
23765 * config/sparc/sparc-opts.h (enum processor_type): Add
23767 * config/sparc/sparc.c (m8_costs): New struct.
23768 (sparc_option_override): Handle TARGET_CPU_m8.
23769 (sparc32_initialize_trampoline): Likewise.
23770 (sparc64_initialize_trampoline): Likewise.
23771 (sparc_issue_rate): Likewise.
23772 (sparc_register_move_cost): Likewise.
23773 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23774 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23775 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23776 (CPP_CPU_SPEC): Handle M8.
23777 (ASM_CPU_SPEC): Likewise.
23778 (AS_M8_FLAG): Define.
23779 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23780 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23781 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23783 * configure: Regenerate.
23784 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23787 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23789 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23791 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23792 ("*movdi_insn_sp32"): Do not set v3pipe.
23793 ("*movsi_insn"): Likewise.
23794 ("*movdi_insn_sp64"): Likewise.
23795 ("*movsf_insn"): Likewise.
23796 ("*movdf_insn_sp32"): Likewise.
23797 ("*movdf_insn_sp64"): Likewise.
23798 ("*zero_extendsidi2_insn_sp64"): Likewise.
23799 ("*sign_extendsidi2_insn"): Likewise.
23800 ("*mov<VM32:mode>_insn"): Likewise.
23801 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23802 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23803 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23804 ("<vlop:code><VL:mode>3"): Likewise.
23805 ("*not_<vlop:code><VL:mode>3"): Likewise.
23806 ("*nand<VL:mode>_vis"): Likewise.
23807 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23808 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23809 ("one_cmpl<VL:mode>2"): Likewise.
23810 ("faligndata<VM64:mode>_vis"): Likewise.
23811 ("alignaddrsi_vis"): Likewise.
23812 ("alignaddrdi_vis"): Likweise.
23813 ("alignaddrlsi_vis"): Likewise.
23814 ("alignaddrldi_vis"): Likewise.
23815 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23816 ("bmaskdi_vis"): Likewise.
23817 ("bmasksi_vis"): Likewise.
23818 ("bshuffle<VM64:mode>_vis"): Likewise.
23819 ("cmask8<P:mode>_vis"): Likewise.
23820 ("cmask16<P:mode>_vis"): Likewise.
23821 ("cmask32<P:mode>_vis"): Likewise.
23822 ("pdistn<P:mode>_vis"): Likewise.
23823 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23825 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23827 * config/sparc/sparc.md ("subtype"): New insn attribute.
23828 ("*wrgsr_sp64"): Set insn subtype.
23829 ("*rdgsr_sp64"): Likewise.
23830 ("alignaddrsi_vis"): Likewise.
23831 ("alignaddrdi_vis"): Likewise.
23832 ("alignaddrlsi_vis"): Likewise.
23833 ("alignaddrldi_vis"): Likewise.
23834 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23835 ("fexpand_vis"): Likewise.
23836 ("fpmerge_vis"): Likewise.
23837 ("faligndata<VM64:mode>_vis"): Likewise.
23838 ("bshuffle<VM64:mode>_vis"): Likewise.
23839 ("cmask8<P:mode>_vis"): Likewise.
23840 ("cmask16<P:mode>_vis"): Likewise.
23841 ("cmask32<P:mode>_vis"): Likewise.
23842 ("fchksm16_vis"): Likewise.
23843 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23844 ("fmean16_vis"): Likewise.
23845 ("fp<plusminus_insn>64_vis"): Likewise.
23846 ("<plusminus_insn>v8qi3"): Likewise.
23847 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23848 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23849 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23850 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23851 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23852 ("*movqi_insn"): Likewise.
23853 ("*movhi_insn"): Likewise.
23854 ("*movsi_insn"): Likewise.
23855 ("movsi_pic_gotdata_op"): Likewise.
23856 ("*movdi_insn_sp32"): Likewise.
23857 ("*movdi_insn_sp64"): Likewise.
23858 ("movdi_pic_gotdata_op"): Likewise.
23859 ("*movsf_insn"): Likewise.
23860 ("*movdf_insn_sp32"): Likewise.
23861 ("*movdf_insn_sp64"): Likewise.
23862 ("*zero_extendhisi2_insn"): Likewise.
23863 ("*zero_extendqihi2_insn"): Likewise.
23864 ("*zero_extendqisi2_insn"): Likewise.
23865 ("*zero_extendqidi2_insn"): Likewise.
23866 ("*zero_extendhidi2_insn"): Likewise.
23867 ("*zero_extendsidi2_insn_sp64"): Likewise.
23868 ("ldfsr"): Likewise.
23869 ("prefetch_64"): Likewise.
23870 ("prefetch_32"): Likewise.
23871 ("tie_ld32"): Likewise.
23872 ("tie_ld64"): Likewise.
23873 ("*tldo_ldub_sp32"): Likewise.
23874 ("*tldo_ldub1_sp32"): Likewise.
23875 ("*tldo_ldub2_sp32"): Likewise.
23876 ("*tldo_ldub_sp64"): Likewise.
23877 ("*tldo_ldub1_sp64"): Likewise.
23878 ("*tldo_ldub2_sp64"): Likewise.
23879 ("*tldo_ldub3_sp64"): Likewise.
23880 ("*tldo_lduh_sp32"): Likewise.
23881 ("*tldo_lduh1_sp32"): Likewise.
23882 ("*tldo_lduh_sp64"): Likewise.
23883 ("*tldo_lduh1_sp64"): Likewise.
23884 ("*tldo_lduh2_sp64"): Likewise.
23885 ("*tldo_lduw_sp32"): Likewise.
23886 ("*tldo_lduw_sp64"): Likewise.
23887 ("*tldo_lduw1_sp64"): Likewise.
23888 ("*tldo_ldx_sp64"): Likewise.
23889 ("*mov<VM32:mode>_insn"): Likewise.
23890 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23891 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23893 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23895 * config/sparc/sparc.md ("type"): New insn type viscmp.
23896 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
23898 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
23899 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
23900 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23901 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
23903 ("n7_vis_logical_11cycle"): Likewise.
23904 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
23905 * config/sparc/niagara2.md ("niag3_vis": Likewise.
23906 * config/sparc/niagara.md ("niag_vis"): Likewise.
23907 * config/sparc/ultra3.md ("us3_fga"): Likewise.
23908 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
23910 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23912 * config/sparc/sparc.md: New instruction type `bmask'.
23913 (bmaskdi_vis): Use the `bmask' type.
23914 (bmasksi_vis): Likewise.
23915 * config/sparc/ultra3.md (us3_array): Likewise.
23916 * config/sparc/niagara7.md (n7_array): Likewise.
23917 * config/sparc/niagara4.md (n4_array): Likewise.
23918 * config/sparc/niagara2.md (niag2_vis): Likewise.
23919 (niag3_vis): Likewise.
23920 * config/sparc/niagara.md (niag_vis): Likewise.
23922 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23924 * ipa-comdats.c: Remove optimize check from gate.
23925 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
23926 for functions not optimized.
23927 (ipa_fn_summary_read): Skip optimize check.
23928 (ipa_fn_summary_write): Likewise.
23929 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
23931 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
23933 (can_inline_edge_p): Check flag_pcc_struct_return for match.
23934 (check_callers): Give up on caller which is not optimized.
23935 (inline_small_functions): Likewise.
23936 (ipa_inline): Do not give up when not optimizing.
23937 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
23938 away unoptimizes cdtors.
23939 (whole_program_function_and_variable_visibility): Do
23940 ipa_discover_readonly_nonaddressable_vars in LTO mode.
23941 * ipa.c (process_references): Do not check optimize.
23942 (symbol_table::remove_unreachable_nodes): Update optimize check.
23943 (set_writeonly_bit): Update optimize check.
23944 (pass_ipa_cdtor_merge::gate): Do not check optimize.
23945 (pass_ipa_single_use::gate): Remove.
23947 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23949 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
23950 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
23951 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
23952 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
23953 permute_load, permute_store, adjust_extract, adjust_splat,
23954 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
23955 replace_swap_with_copy, dump_swap_insn_table,
23956 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
23957 recombine_lvx_pattern, recombine_stvx_pattern,
23958 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
23959 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
23960 to file rs6000-p8swap.c.
23961 * config/rs6000/rs6000-p8swap.c: New file.
23962 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
23963 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
23964 and rs6000*-*-* targets.
23966 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23968 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
23970 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23972 * lto-wrapper.c (merge_and_complain): Do not merge
23973 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
23974 fsigned_zeros, ftrapping_math, fwrapv.
23975 (append_compiler_options): Do not track these options.
23976 (append_linker_options): Likewie
23978 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23980 * cgraphunit.c (cgraph_node::finalize_function): When
23981 !flag_toplevel_reorde set no_reorder flag.
23982 (varpool_node::finalize_decl): Likewise.
23983 (symbol_table::compile): Drop no toplevel reorder path.
23985 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23987 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
23988 edges; zero probability is not better than uninitialized.
23990 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23992 * asan.h (asan_sanitize_allocas_p): Declare.
23993 * asan.c (asan_sanitize_allocas_p): New function.
23994 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
23995 (handle_builtin_alloca): Likewise.
23996 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
23997 if !asan_sanitize_allocas_p.
23998 * params.def (asan-instrument-allocas): Add new option.
23999 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24000 * opts.c (common_handle_option): Disable allocas sanitization for
24003 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24005 * asan.c: Include gimple-fold.h.
24006 (get_last_alloca_addr): New function.
24007 (handle_builtin_stackrestore): Likewise.
24008 (handle_builtin_alloca): Likewise.
24009 (asan_emit_allocas_unpoison): Likewise.
24010 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24011 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24012 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24013 (instrument_builtin_call): Pass gimple iterator to
24014 get_mem_refs_of_builtin_call.
24015 (last_alloca_addr): New global.
24016 * asan.h (asan_emit_allocas_unpoison): Declare.
24017 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24018 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24019 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24020 if function calls alloca.
24021 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24022 * gimple-fold.h (replace_call_with_value): Declare.
24023 * internal-fn.c: Include asan.h.
24024 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24025 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24027 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24029 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24030 (C_SELFTEST_FLAGS): New.
24031 (CPP_SELFTEST_FLAGS): New.
24032 (SELFTEST_DEPS): New, from deps of s-selftest.
24033 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24034 (CPP_SELFTEST_DEPS): New.
24035 (selftest): Add dependency on s-selftest-c++.
24036 (s-selftest): Rename to...
24037 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24038 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24039 than SELFTEST_FLAGS.
24040 (selftest-gdb): Rename to...
24041 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24043 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24044 (selftest-valgrind): Rename to...
24045 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24047 (selftest-valgrind): Reintroduce as an alias for
24048 selftest-c-valgrind.
24049 (s-selftest-c++): New.
24050 (selftest-c++-gdb): New.
24051 (selftest-c++-valgrind): New.
24053 2017-07-06 Olivier Hainque <hainque@adacore.com>
24055 * gcc.c (process_command): When deciding if undefined variables
24056 should be ignored when processing specs, accept "gcc -v" as well.
24058 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24060 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24061 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24063 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24065 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24066 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24067 * config/arm/arm-tables.opt: Regenerate.
24068 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24070 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24072 2017-07-06 Carl Love <cel@us.ibm.com>
24074 * ChangeLog: Clean up from mid air collision
24076 2017-07-06 Carl Love <cel@us.ibm.com>
24078 * config/rs6000/rs6000-c.c: Add support for built-in functions
24079 vector signed int vec_subc (vector signed int, vector signed int);
24080 vector signed __int128 vec_subc (vector signed __int128,
24081 vector signed __int128);
24082 vector unsigned __int128 vec_subc (vector unsigned __int128,
24083 vector unsigned __int128);
24084 vector signed int vec_sube (vector signed int, vector signed int,
24085 vector signed int);
24086 vector unsigned int vec_sube (vector unsigned int,
24087 vector unsigned int,
24088 vector unsigned int);
24089 vector signed __int128 vec_sube (vector signed __int128,
24090 vector signed __int128,
24091 vector signed__int128);
24092 vector unsigned __int128 vec_sube (vector unsigned __int128,
24093 vector unsigned __int128,
24094 vector unsigned __int128);
24095 vector signed int vec_subec (vector signed int, vector signed int,
24096 vector signed int);
24097 vector unsigned int vec_subec (vector unsigned int,
24098 vector unsigned int,
24099 vector unsigned int);
24100 vector signed __int128 vec_subec (vector signed __int128,
24101 vector signed __int128,
24102 vector signed__int128);
24103 vector unsigned __int128 vec_subec (vector unsigned __int128,
24104 vector unsigned __int128,
24105 vector unsigned __int128);
24106 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24107 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24108 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24109 BU_ALTIVEC_OVERLOAD_X definitions.
24110 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24111 * doc/extend.texi: Update the built-in documentation file for the new
24112 built-in functions.
24114 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24117 * diagnostic-show-locus.c (layout::layout): Use start and finish
24118 spelling location for the start and finish of each range.
24119 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24120 Add unused aspect param.
24121 * input.c (expand_location_1): Add "aspect" param, and use it
24122 to access the correct part of the location.
24123 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24125 (expand_location_to_spelling_point): Likewise.
24126 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24127 param, and pass it to expand_location_1.
24129 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24131 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24132 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24133 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24134 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24135 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24136 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24137 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24138 _mm_maskz_getmant_ss): New intrinsics.
24139 (__builtin_ia32_getexpss128_mask): Changed to ...
24140 __builtin_ia32_getexpss128_round ... this.
24141 (__builtin_ia32_getexpsd128_mask): Changed to ...
24142 __builtin_ia32_getexpsd128_round ... this.
24143 * config/i386/i386-builtin-types.def
24144 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24145 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24146 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24147 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24148 __builtin_ia32_getmantss_mask_round): New builtins.
24149 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24150 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24151 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24152 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24153 * config/i386/sse.md
24154 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24155 avx512f_sgetexp<mode><mask_scalar_name>
24156 <round_saeonly_scalar_name> ... this.
24157 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24158 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24159 vgetexp<ssescalarmodesuffix>
24160 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24161 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24162 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24163 avx512f_vgetmant<mode><mask_scalar_name>
24164 <round_saeonly_scalar_name> ... this.
24165 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24166 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24167 vgetmant<ssescalarmodesuffix>
24168 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24169 %0<mask_scalar_operand4>, %1, %2
24170 <round_saeonly_scalar_mask_op4>, %3} ... this.
24171 * config/i386/subst.md (mask_scalar_operand4,
24172 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24173 round_saeonly_scalar_nimm_predicate): New subst attributes.
24175 2017-07-06 Julia Koval <julia.koval@intel.com>
24177 * config/i386/i386.c (ix86_erase_embedded_rounding):
24178 Remove code for old rounding pattern.
24180 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24182 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24184 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24186 * doc/sourcebuild.texi (Test Directives, Variants of
24187 dg-require-support): Add documentation for dg-require-stack-check.
24189 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24191 * config/i386/subst.md (mask_scalar, round_scalar,
24192 round_saeonly_scalar): New meta-templates.
24193 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24194 round_scalar_mask_operand3, round_scalar_mask_op3,
24195 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24196 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24197 round_saeonly_scalar_constraint,
24198 round_saeonly_scalar_prefix): New subst attribute.
24199 * config/i386/sse.md
24200 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24201 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24202 <round_scalar_name> ... this.
24203 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24204 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24205 <round_scalar_name> ... this.
24206 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24207 <sse>_vm<code><mode>3<mask_scalar_name>
24208 <round_saeonly_scalar_name> ... this.
24209 (v<plusminus_mnemonic><ssescalarmodesuffix>
24210 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24211 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24212 v<plusminus_mnemonic><ssescalarmodesuffix>
24213 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24214 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24215 (v<multdiv_mnemonic><ssescalarmodesuffix>
24216 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24217 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24218 v<multdiv_mnemonic><ssescalarmodesuffix>
24219 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24220 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24221 (v<maxmin_float><ssescalarmodesuffix>
24222 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24223 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24224 v<maxmin_float><ssescalarmodesuffix>
24225 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24226 %0<mask_scalar_operand3>, %1, %<iptr>2
24227 <round_saeonly_scalar_mask_op3>} ... this.
24229 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24231 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24232 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24234 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24235 Alan Hayward <alan.hayward@arm.com>
24236 David Sherwood <david.sherwood@arm.com>
24238 * combine.c (simplify_if_then_else): Remove "enum" before
24240 * compare-elim.c (can_eliminate_compare): Likewise.
24241 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24243 (aarch64_lookup_simd_builtin_type): Likewise.
24244 (aarch64_simd_builtin_type): Likewise.
24245 (aarch64_init_simd_builtin_types): Likewise.
24246 (aarch64_simd_expand_args): Likewise.
24247 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24249 (aarch64_reverse_mask): Likewise.
24250 (aarch64_simd_emit_reg_reg_move): Likewise.
24251 (aarch64_gen_adjusted_ldpstp): Likewise.
24252 (aarch64_ccmp_mode_to_code): Likewise.
24253 (aarch64_operands_ok_for_ldpstp): Likewise.
24254 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24255 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24257 (aarch64_min_divisions_for_recip_mul): Likewise.
24258 (aarch64_reassociation_width): Likewise.
24259 (aarch64_get_condition_code_1): Likewise.
24260 (aarch64_simd_emit_reg_reg_move): Likewise.
24261 (aarch64_simd_attr_length_rglist): Likewise.
24262 (aarch64_reverse_mask): Likewise.
24263 (aarch64_operands_ok_for_ldpstp): Likewise.
24264 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24265 (aarch64_gen_adjusted_ldpstp): Likewise.
24266 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24268 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24269 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24270 (arm_lookup_simd_builtin_type): Likewise.
24271 (arm_simd_builtin_type): Likewise.
24272 (arm_init_simd_builtin_types): Likewise.
24273 (arm_expand_builtin_args): Likewise.
24274 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24275 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24276 (ft32_setup_incoming_varargs): Likewise.
24277 (ft32_function_arg): Likewise.
24278 (ft32_function_arg_advance): Likewise.
24279 (ft32_pass_by_reference): Likewise.
24280 (ft32_arg_partial_bytes): Likewise.
24281 (ft32_valid_pointer_mode): Likewise.
24282 (ft32_addr_space_pointer_mode): Likewise.
24283 (ft32_addr_space_legitimate_address_p): Likewise.
24284 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24286 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24287 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24288 (iamcu_alignment): Likewise.
24289 (canonicalize_vector_int_perm): Likewise.
24290 (ix86_noce_conversion_profitable_p): Likewise.
24291 (ix86_mpx_bound_mode): Likewise.
24292 (ix86_operands_ok_for_move_multiple): Likewise.
24293 * config/microblaze/microblaze-protos.h
24294 (microblaze_expand_conditional_branch_reg): Likewise.
24295 * config/microblaze/microblaze.c
24296 (microblaze_expand_conditional_branch_reg): Likewise.
24297 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24299 (rs6000_reassociation_width): Likewise.
24300 (rs6000_invalid_binary_op): Likewise.
24301 (fusion_p9_p): Likewise.
24302 (emit_fusion_p9_load): Likewise.
24303 (emit_fusion_p9_store): Likewise.
24304 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24306 (riscv_hard_regno_mode_ok_p): Likewise.
24307 (riscv_address_insns): Likewise.
24308 (riscv_split_symbol): Likewise.
24309 (riscv_legitimize_move): Likewise.
24310 (riscv_function_value): Likewise.
24311 (riscv_hard_regno_nregs): Likewise.
24312 (riscv_expand_builtin): Likewise.
24313 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24314 (riscv_build_integer): Likewise.
24315 (riscv_split_integer): Likewise.
24316 (riscv_legitimate_constant_p): Likewise.
24317 (riscv_cannot_force_const_mem): Likewise.
24318 (riscv_regno_mode_ok_for_base_p): Likewise.
24319 (riscv_valid_base_register_p): Likewise.
24320 (riscv_valid_offset_p): Likewise.
24321 (riscv_valid_lo_sum_p): Likewise.
24322 (riscv_classify_address): Likewise.
24323 (riscv_legitimate_address_p): Likewise.
24324 (riscv_address_insns): Likewise.
24325 (riscv_load_store_insns): Likewise.
24326 (riscv_force_binary): Likewise.
24327 (riscv_split_symbol): Likewise.
24328 (riscv_force_address): Likewise.
24329 (riscv_legitimize_address): Likewise.
24330 (riscv_move_integer): Likewise.
24331 (riscv_legitimize_const_move): Likewise.
24332 (riscv_legitimize_move): Likewise.
24333 (riscv_address_cost): Likewise.
24334 (riscv_subword): Likewise.
24335 (riscv_output_move): Likewise.
24336 (riscv_canonicalize_int_order_test): Likewise.
24337 (riscv_emit_int_order_test): Likewise.
24338 (riscv_function_arg_boundary): Likewise.
24339 (riscv_pass_mode_in_fpr_p): Likewise.
24340 (riscv_pass_fpr_single): Likewise.
24341 (riscv_pass_fpr_pair): Likewise.
24342 (riscv_get_arg_info): Likewise.
24343 (riscv_function_arg): Likewise.
24344 (riscv_function_arg_advance): Likewise.
24345 (riscv_arg_partial_bytes): Likewise.
24346 (riscv_function_value): Likewise.
24347 (riscv_pass_by_reference): Likewise.
24348 (riscv_setup_incoming_varargs): Likewise.
24349 (riscv_print_operand): Likewise.
24350 (riscv_elf_select_rtx_section): Likewise.
24351 (riscv_save_restore_reg): Likewise.
24352 (riscv_for_each_saved_reg): Likewise.
24353 (riscv_register_move_cost): Likewise.
24354 (riscv_hard_regno_mode_ok_p): Likewise.
24355 (riscv_hard_regno_nregs): Likewise.
24356 (riscv_class_max_nregs): Likewise.
24357 (riscv_memory_move_cost): Likewise.
24358 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24359 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24360 (rl78_addr_space_address_mode): Likewise.
24361 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24363 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24364 (rs6000_reassociation_width): Likewise.
24365 (rs6000_invalid_binary_op): Likewise.
24366 (fusion_p9_p): Likewise.
24367 (emit_fusion_p9_load): Likewise.
24368 (emit_fusion_p9_store): Likewise.
24369 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24370 (ok_for_simple_move_operands): Likewise.
24371 (ok_for_simple_move_strict_operands): Likewise.
24372 (ok_for_simple_arith_logic_operands): Likewise.
24373 (visium_legitimize_reload_address): Likewise.
24374 (visium_select_cc_mode): Likewise.
24375 (output_cbranch): Likewise.
24376 (visium_split_double_move): Likewise.
24377 (visium_expand_copysign): Likewise.
24378 (visium_expand_int_cstore): Likewise.
24379 (visium_expand_fp_cstore): Likewise.
24380 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24381 (visium_function_arg): Likewise.
24382 (visium_function_arg_advance): Likewise.
24383 (visium_libcall_value): Likewise.
24384 (visium_setup_incoming_varargs): Likewise.
24385 (visium_legitimate_constant_p): Likewise.
24386 (visium_legitimate_address_p): Likewise.
24387 (visium_legitimize_address): Likewise.
24388 (visium_secondary_reload): Likewise.
24389 (visium_register_move_cost): Likewise.
24390 (visium_memory_move_cost): Likewise.
24391 (prepare_move_operands): Likewise.
24392 (ok_for_simple_move_operands): Likewise.
24393 (ok_for_simple_move_strict_operands): Likewise.
24394 (ok_for_simple_arith_logic_operands): Likewise.
24395 (visium_function_value_1): Likewise.
24396 (rtx_ok_for_offset_p): Likewise.
24397 (visium_legitimize_reload_address): Likewise.
24398 (visium_split_double_move): Likewise.
24399 (visium_expand_copysign): Likewise.
24400 (visium_expand_int_cstore): Likewise.
24401 (visium_expand_fp_cstore): Likewise.
24402 (visium_split_cstore): Likewise.
24403 (visium_select_cc_mode): Likewise.
24404 (visium_split_cbranch): Likewise.
24405 (output_cbranch): Likewise.
24406 (visium_print_operand_address): Likewise.
24407 * expmed.c (flip_storage_order): Likewise.
24408 * expmed.h (emit_cstore): Likewise.
24409 (flip_storage_order): Likewise.
24410 * genrecog.c (validate_pattern): Likewise.
24411 * hsa-gen.c (gen_hsa_addr): Likewise.
24412 * internal-fn.c (expand_arith_overflow): Likewise.
24413 * ira-color.c (allocno_copy_cost_saving): Likewise.
24414 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24415 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24416 (process_invariant_for_inheritance): Likewise.
24417 * lra-eliminations.c (move_plus_up): Likewise.
24418 * omp-low.c (lower_oacc_reductions): Likewise.
24419 * simplify-rtx.c (simplify_subreg): Likewise.
24420 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24421 (TARGET_CHKP_BOUND_MODE): Likewise..
24422 * targhooks.c (default_chkp_bound_mode): Likewise.
24423 (default_setup_incoming_vararg_bounds): Likewise.
24424 * targhooks.h (default_chkp_bound_mode): Likewise.
24425 (default_setup_incoming_vararg_bounds): Likewise.
24426 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24427 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24428 (have_whole_vector_shift): Likewise.
24429 * tree-vect-stmts.c (vectorizable_load): Likewise.
24430 * doc/tm.texi: Regenerate.
24432 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24434 Graceful degrade if Binutils PR21472 is not available.
24437 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24438 .rodata in flash test fails.
24439 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24440 * confgure: Regenerate.
24441 * config.in: Regenerate.
24442 * config/avr/avr.c (avr_asm_named_section)
24443 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24444 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24445 (avr_asm_init_sections): Same.
24447 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24449 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24450 (fma<VH:mode>4_intrinsic): Likewise.
24451 (*fmsub<VCVTF:mode>4): Likewise.
24452 (*fmsub<VH:mode>4_intrinsic): Likewise.
24454 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24457 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24458 Don't depend on "optimize > 0".
24459 (out_movhi_r_mr, out_movqi_mr_r): Same.
24460 (out_movhi_mr_r, out_movqi_r_mr): Same.
24461 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24462 io_address_operand on "optimize > 0".
24464 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24466 * tree-loop-distribution.c: Add general explanantion on the pass.
24467 (generate_loops_for_partition): Mark distributed loop.
24468 (pg_add_dependence_edges): New parameter. Handle alias data
24469 dependence specially and record it in the parameter if asked.
24470 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24471 (init_partition_graph_vertices, add_partition_graph_edge): New.
24472 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24473 (free_partition_graph_vdata, build_partition_graph): New.
24474 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24475 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24476 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24477 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24478 (version_for_distribution_p, finalize_partitions): New.
24479 (distribute_loop): Handle alias data dependence specially. Factor
24480 out loop fusion code as functions and call these functions.
24482 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24484 * tree-loop-distribution.c (classify_partition): New parameter and
24485 better handle reduction statement.
24486 (rdg_build_partitions): Revise comment.
24487 (distribute_loop): Compute statements in all partitions and pass it
24488 to classify_partition.
24490 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24492 * tree-loop-distribution.c (enum partition_type): New.
24493 (struct partition): New field type.
24494 (partition_merge_into): Add parameter. Update partition type.
24495 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24496 (build_rdg_partition_for_vertex): Compute partition type.
24497 (rdg_build_partitions): Dump partition type.
24498 (distribute_loop): Update calls to partition_merge_into.
24500 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24502 * tree-loop-distribution.c (struct ddr_hasher): New.
24503 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24505 (classify_partition): Call get_data_dependence.
24506 (pg_add_dependence_edges): Ditto.
24507 (distribute_loop): Release data dependence hash table.
24509 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24511 * tree-loop-distribution.c (ref_base_address): Delete.
24512 (similar_memory_accesses): Rename ...
24513 (share_memory_accesses): ... to this. Check if partitions access
24514 the same memory reference.
24515 (distribute_loop): Call share_memory_accesses.
24517 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24519 * tree-loop-distribution.c (struct partition): New field recording
24520 its data reference.
24521 (partition_alloc, partition_free): Init and release data refs.
24522 (partition_merge_into): Merge data refs.
24523 (build_rdg_partition_for_vertex): Collect data refs for partition.
24524 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24526 (distribute_loop): Remve data refs from vertice data of partition
24529 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24531 * tree-loop-distribution.c (params.h): Include header file.
24532 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24533 (datarefs_vec): New global var.
24534 (create_rdg_vertices): Use datarefs_vec directly.
24535 (free_rdg): Don't free data references.
24536 (build_rdg): Update use. Don't free data references.
24537 (distribute_loop): Compute global variable for data references.
24538 Bail out if there are too many data references.
24540 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24542 * tree-loop-distribution.c (loop_nest): New global var.
24543 (build_rdg): Use loop directly, rather than loop nest.
24544 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24546 (distribute_loop): Compute global variable loop nest. Update use.
24548 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24550 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24551 (partition_merge_into): New parameter. Dump reason for fusion.
24552 (distribute_loop): Update use of partition_merge_into.
24554 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24556 * tree-loop-distribution.c (bb_top_order_index): New.
24557 (bb_top_order_index_size, bb_top_order_cmp): New.
24558 (stmts_from_loop): Use topological order.
24559 (pass_loop_distribution::execute): Compute and release topological
24560 order for basic blocks.
24562 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24564 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24567 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24569 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24570 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24571 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24572 * internal-fn.def (LOOP_DIST_ALIAS): New.
24573 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24574 (fold_loop_internal_call): ... this.
24575 (vect_loop_dist_alias_call): New function.
24576 (set_uid_loop_bbs): Call fold_loop_internal_call.
24577 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24580 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24583 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24584 Require dead FLAGS_REG at the beginning of a peephole.
24586 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24589 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24590 arguments in the call to __builtin_ia32_sbb_u32.
24591 (_subborrow_u64): Swap _X and _Y arguments in the call to
24592 __builtin_ia32_sbb_u64.
24594 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24597 * tree-vrp.c (compare_assert_loc): Turn into a function template
24598 with stable template parameter. Only test if a->e is NULL,
24599 !a->e == !b->e has been verified already. Use e == NULL or
24600 e != NULL instead of e or ! e tests. If stable is true, don't use
24601 iterative_hash_expr, on the other side allow a or b or both NULL
24602 and sort the NULLs last.
24603 (process_assert_insertions): Sort using compare_assert_loc<false>
24604 instead of compare_assert_loc, later sort using
24605 compare_assert_loc<true> before calling process_assert_insertions_for
24606 in a loop. Use break instead of continue once seen NULL pointer.
24608 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24610 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24611 Cortex-R7 and Cortex-R8 processors.
24613 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24615 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24616 uninitialized while src is not.
24618 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24620 * common/config/arm/arm-common.c: Adjust include path for
24622 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24623 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24624 (arm-cpu-data.h): Likewise.
24625 (arm-cpu-cdata.h): Likewise.
24626 * config/arm/arm-cpu.h: Delete.
24627 * config/arm/arm-cpu-cdata.h: Delete.
24628 * config/arm/arm-cpu-data.h: Delete.
24630 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24632 * config/arm/arm-cpus.in (cortex-a55): New.
24633 (cortex-a75): Likewise.
24634 (cortex-a75.cortex-a55): Likewise.
24635 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24637 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24638 * config/arm/arm-cpu-cdata.h: Regenerate.
24639 * config/arm/arm-cpu-data.h: Regenerate.
24640 * config/arm/arm-cpu.h: Regenerate.
24641 * config/arm/arm-tables.opt: Regenerate.
24642 * config/arm/arm-tune.md: Regenerate.
24644 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24646 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24648 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24650 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24653 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24655 PR tree-optimization/81292
24656 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24657 full_string_p, also call adjust_related_strinfos if the adjustment
24658 is simple, otherwise invalidate related strinfos.
24660 2017-07-04 Martin Liska <mliska@suse.cz>
24663 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24664 newly created variable as DECL_IGNORED_P.
24666 2017-07-04 Martin Liska <mliska@suse.cz>
24669 * ipa-inline.c (inline_small_functions):
24670 Use xstrdup_for_dump.
24672 2017-07-04 Tom de Vries <tom@codesourcery.com>
24674 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24676 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24679 * config/darwin.c (darwin_function_switched_text_sections):
24682 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24684 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24686 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24688 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24690 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24692 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24693 min_profitable_iters, and th as inclusive lower bounds.
24694 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24695 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24696 for min_profitable_iters and min_profitable_estimate.
24697 (vect_transform_loop): Treat th as an inclusive lower bound.
24698 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24700 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24703 * config/darwin.c (darwin_function_switched_text_sections):
24704 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24705 in two pieces, and suppress the use of buf.
24707 2017-07-03 Nathan Sidwell <nathan@acm.org>
24709 * hash-table.h (hash_table_mod1): Fix indentation.
24711 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24713 PR middle-end/81290
24714 * predict.c (force_edge_cold): Be more careful about propagation
24716 * profile-count.h (profile_probability::guessed,
24717 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24719 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24721 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24723 * doc/invoke.texi (rcpc architecture extension): Document it.
24725 2017-07-03 Richard Biener <rguenther@suse.de>
24727 PR tree-optimization/60510
24728 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24729 the scalar reduction PHI and use it.
24730 (vectorizable_reduction): Properly guard the single_defuse_cycle
24731 path for non-SLP reduction chains where we cannot use it.
24732 Rework reduc_def/index and vector type deduction. Rework
24733 vector operand gathering during reduction op code-gen.
24734 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24735 chains dissolve the chain and leave it to non-SLP reduction
24738 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24740 * tree-data-ref.h (dr_alignment): Declare.
24741 * tree-data-ref.c (dr_alignment): New function.
24742 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24743 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24745 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24747 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24749 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24750 and base_misalignment fields.
24751 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24752 * tree-data-ref.c: Include builtins.h.
24753 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24754 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24755 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24756 * tree-vect-data-refs.c: Include tree-cfg.h.
24757 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24758 fields instead of calculating an alignment here.
24759 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24760 innermost_loop_behavior fields.
24762 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24764 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24766 (DR_STEP_ALIGNMENT): New macro.
24767 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24768 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24769 (create_data_ref): Print it.
24770 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24771 to tell whether the step preserves vector (mis)alignment.
24772 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24773 Move the check for an integer step and generalise to all INTEGER_CST.
24774 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24775 Print the outer step alignment.
24777 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24779 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24780 with offset_alignment.
24781 (DR_ALIGNED_TO): Delete.
24782 (DR_OFFSET_ALIGNMENT): New macro.
24783 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24784 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24785 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24786 (create_data_ref): Likewise.
24787 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24788 (vect_analyze_data_refs): Likewise.
24789 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24790 creating dummy innermost behavior.
24792 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24794 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24795 with a "innermost_loop_behavior *" and refeence tree.
24796 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24797 (create_data_ref): Update call accordingly.
24798 * tree-predcom.c (find_looparound_phi): Likewise.
24800 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24802 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24803 fields with dr_wrt_vec_loop.
24804 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24805 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24806 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24807 (vect_dr_behavior): New function.
24808 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24809 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24810 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24811 track whether the step preserves the misalignment.
24812 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24813 Use vect_dr_behavior.
24814 (vect_setup_realignment): Update call accordingly.
24815 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24816 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24817 call to vect_create_addr_base_for_vector_ref.
24818 (vect_create_cond_for_align_checks): Likewise.
24819 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24820 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24821 (vect_recog_mask_conversion_pattern): Likewise.
24822 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24823 (new_stmt_vec_info): Remove redundant zeroing.
24825 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24827 * common/config/arm/arm-common.c (arm_be8_option): New function.
24828 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24829 (ISA_ARMv6): Add isa_bit_be8.
24830 * config/arm/arm.h (arm_be8_option): Add prototype.
24831 (BE8_SPEC_FUNCTION): New define.
24832 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24833 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24834 (mlittle-endian): Similarly.
24835 (mbe8, mbe32): New options.
24836 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24837 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24839 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24841 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24843 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24845 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24846 (cleanup_tree_cfg_bb): Use it.
24847 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24849 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24851 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24854 * loop-doloop.c (add_test): Update profile.
24856 2017-07-03 Martin Liska <mliska@suse.cz>
24859 * sanopt.c (rewrite_usage_of_param): New function.
24860 (sanitize_rewrite_addressable_params): Likewise.
24861 (pass_sanopt::execute): Call rewrite_usage_of_param.
24863 2017-07-03 Richard Biener <rguenther@suse.de>
24865 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
24866 back to using VIEW_CONVERT_EXPR.
24868 2017-07-03 Martin Liska <mliska@suse.cz>
24871 * doc/extend.texi: Document when a resolver function is
24872 generated for target_clones.
24874 2017-07-03 Martin Liska <mliska@suse.cz>
24876 * asan.c (asan_emit_stack_protection): Unpoison just red zones
24877 and shadow memory of auto variables which are subject of
24878 use-after-scope sanitization.
24879 (asan_expand_mark_ifn): Add do set only when is_poison.
24881 2016-07-03 Richard Biener <rguenther@suse.de>
24883 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
24885 (vect_force_simple_reduction): Record reduction def -> phi mapping.
24886 (vectorizable_reduction): Perform reduction PHI creation when
24887 visiting a reduction PHI and adjust and simplify code generation
24888 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
24889 (vect_transform_loop): Visit reduction PHIs.
24890 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
24891 defs into the SLP tree.
24892 (vect_build_slp_tree): Reduction defs terminate the recursion.
24893 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
24895 (vect_get_vec_defs_for_stmt_copy): Export.
24896 (vect_get_vec_defs): Likewise.
24897 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
24899 (vect_get_vec_defs_for_stmt_copy): Declare.
24900 (vect_get_vec_defs): Likewise.
24902 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24904 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
24905 parameter with a "loop" parameter and use it instead of the
24906 loop containing DR_STMT. Don't check simple_iv when doing
24907 BB analysis. Describe the two analysis modes in the comment.
24909 2017-07-03 Tom de Vries <tom@codesourcery.com>
24911 PR tree-optimization/69468
24912 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
24913 (find_same_succ_bb): Handle ignore_edge_flags.
24915 2017-07-03 Tom de Vries <tom@codesourcery.com>
24917 PR tree-optimization/81192
24918 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
24920 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
24922 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
24924 2017-07-03 Tom de Vries <tom@codesourcery.com>
24926 PR tree-optimization/81192
24927 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
24928 BB_SAME_SUCC (bb) == NULL.
24930 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24932 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
24935 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24937 * dumpfile.c: Include profile-count.h
24938 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
24940 (insert_cond_bb): Update profile.
24941 * tree-cfg.h (insert_cond_bb): Update prototype.
24942 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
24943 * tree-dump.c: Do not include tree-cfg.
24945 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24947 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
24949 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24951 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
24954 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24956 * tree-complex.c (expand_complex_div_wide): update profile.
24958 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24959 Alan Hayward <alan.hayward@arm.com>
24960 David Sherwood <david.sherwood@arm.com>
24962 * Makefile.in (MACHMODE_H): Remove insn-modes.h
24963 (CORETYPES_H): New define.
24964 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
24965 (insn-modes-inline.h, s-modes-inline-h): New rules.
24966 (generated_files): Add insn-modes-inline.h.
24967 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
24968 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
24969 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
24970 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
24971 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
24972 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
24973 (build/gencodes.o, build/genconditions.o): Likewise.
24974 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
24975 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
24976 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
24977 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
24978 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
24979 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
24980 * coretypes.h: Include everything up to real.h for generators.
24981 Include insn-modes.h first. Include wide-int-print.h after
24982 wide-int.h. Include insn-modes-inline.h and then machmode.h.
24983 * machmode.h: Don't include insn-modes.h here.
24984 * function-tests.c: Remove includes of signop.h, machmode.h,
24985 double-int.h and wide-int.h.
24987 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
24989 * optc-save-gen.awk: Likewise.
24990 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
24991 * godump.c: Remove include of wide-int-print.h.
24992 * pretty-print.h: Likewise.
24993 * wide-int-print.cc: Likewise.
24994 * wide-int.cc: Likewise.
24995 * hash-map-tests.c: Remove include of signop.h.
24996 * hash-set-tests.c: Likewise.
24997 * rtl-tests.c: Likewise.
24998 * mkconfig.sh: Remove include of machmode.h.
24999 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25001 (emit_insn_modes_inline_h): ...this new function. Emit the code
25002 into an insn-modes-inline.h header file, adding appropriate
25003 include guards and end comments.
25004 (emit_insn_modes_c_header): Remove include of machmode.h.
25005 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25007 (main): Handle -i flag and call emit_insn_modes_inline_h when
25010 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25012 * tree-ssa-strlen.c (strinfo): Rename the length field to
25013 nonzero_chars. Add a full_string_p field.
25014 (compare_nonzero_chars, zero_length_string_p): New functions.
25015 (get_addr_stridx): Add an offset_out parameter.
25016 Use compare_nonzero_chars.
25017 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25018 (new_strinfo): Update after above changes to strinfo.
25019 (set_endptr_and_length): Set full_string_p.
25020 (get_string_length): Update after above changes to strinfo.
25021 (unshare_strinfo): Update call to new_strinfo.
25022 (maybe_invalidate): Likewise.
25023 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25024 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25025 as a uhwi instead of an shwi. Update after above changes to
25026 strinfo and new_strinfo.
25027 (zero_length_string): Assert that chainsi contains full strings.
25028 Use zero_length_string_p. Update call to new_strinfo.
25029 (adjust_related_strinfos): Update after above changes to strinfo.
25030 Copy full_string_p from origsi.
25031 (adjust_last_stmt): Use zero_length_string_p.
25032 (handle_builtin_strlen): Update after above changes to strinfo and
25033 new_strinfo. Install the lhs as the string length if the previous
25034 entry didn't describe a full string.
25035 (handle_builtin_strchr): Update after above changes to strinfo
25037 (handle_builtin_strcpy): Likewise.
25038 (handle_builtin_strcat): Likewise.
25039 (handle_builtin_malloc): Likewise.
25040 (handle_pointer_plus): Likewise.
25041 (handle_builtin_memcpy): Likewise. Track nonzero characters
25042 that aren't necessarily followed by a nul terminator.
25043 (handle_char_store): Likewise.
25045 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25047 PR tree-optimization/80769
25048 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25049 for malloc and calloc. Document the new invariant that all related
25050 strinfos have delayed lengths or none do.
25051 (verify_related_strinfos): Move earlier in file.
25052 (set_endptr_and_length): New function, split out from...
25053 (get_string_length): ...here. Also set the lengths of related
25055 (zero_length_string): Assert that chainsi has known (rather than
25057 (adjust_related_strinfos): Likewise.
25059 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25061 PR tree-optimization/81136
25062 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25063 assert that two references with the same misalignment have the same
25064 compile-time misalignment if those compile-time misalignments
25067 2017-07-01 Andi Kleen <ak@linux.intel.com>
25069 * print-tree.c (print_node): Print all attributes.
25071 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25073 * cfg.c (scale_bbs_frequencies): New function.
25074 * cfg.h (scale_bbs_frequencies): Declare it.
25075 * cfgloopanal.c (single_likely_exit): Cleanup.
25076 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25078 (scale_loop_profile): Likewise.
25079 (loop_version): Likewise.
25080 (create_empty_loop_on_edge): Update.
25081 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25082 scale_loop_frequencies, scale_loop_profile, loopify,
25083 loop_version): Update prototypes.
25084 * modulo-sched.c (sms_schedule): Update.
25085 * predict.c (unlikely_executed_edge_p): Also check probability.
25086 (probably_never_executed_edge_p): Fix typo.
25087 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25088 * tree-parloops.c (gen_parallel_loop): Update.
25089 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25090 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25091 * tree-ssa-loop-split.c (split_loop): Update.
25092 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25093 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25094 (vect_loop_versioning): Update.
25095 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25097 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25099 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25101 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25103 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25104 to keep profile consistent.
25106 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25108 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25109 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25110 * profile-count.h (max_safe_multiplier): Make unsigned.
25111 (profile_count::guessed_zero): New.
25113 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25115 * bb-reorder.c (fix_up_crossing_landing_pad,
25116 fix_crossing_conditional_branches): Use make_single_succ_edge
25117 to keep profile consistent.
25119 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25121 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25124 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25127 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25128 the right scopes, make sure cond_jump isn't preserved between multiple
25129 iterations. Search for fallthru edge whenever there are 3+ edges and
25130 use find_fallthru_edge for it.
25132 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25134 Patch by Alexander Monakov <amonakov@ispras.ru>
25135 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25136 probabilities consistently.
25138 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25140 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25141 emit_cmp_and_jump_insns.
25143 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25146 * tree-inline.c (expand_call_inline): Combine profile statuses.
25148 2017-06-30 Andrew Pinski <apinski@cavium.com>
25150 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25151 fold_stmt returned true.
25153 2017-06-30 Nathan Sidwell <nathan@acm.org>
25155 * ggc.h (empty_string): Delete.
25156 * cfgexpand.c (expand_asm_stmt): Use plain "".
25157 * optabs.c (expand_asm_memory_barrier): Likewise.
25158 * stringpool.c (empty_string): Delete.
25159 (digit_vector, digit_string): Delete.
25160 (ggc_alloc_string): Use plain "", don't optimize single digit
25161 strings. Use ggc_alloc_atomic.
25163 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25165 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25166 comparison set and one other set, use the cost of the non-comparison
25169 2017-06-30 Nathan Sidwell <nathan@acm.org>
25171 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25174 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25176 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25177 loops. Remove now unneeded calls to gimple_switch_set_label() that
25178 just set removed labels to NULL_TREE.
25180 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25182 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25183 (set_range_info): ...here. Only call set_range_info_raw if domain
25185 (set_nonzero_bits): Call set_range_info_raw.
25186 * tree-ssanames.h (set_range_info_raw): New.
25188 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25191 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25192 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25193 of nonimmediate_operand and <store_mask_constraint> instead of m
25194 for the input operand. For V8FI iterator, always split if input
25195 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25196 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25197 <store_mask_predicate> instead of register_operand and
25198 <store_mask_constraint> instead of v for the input operand. Make
25199 sure both operands aren't MEMs for if not <mask_applied>.
25201 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25203 * lto-wrapper.c (copy_file) Close both file descriptors before
25206 2017-06-30 Martin Liska <mliska@suse.cz>
25209 * multiple_target.c (create_dispatcher_calls): Make ifunc
25210 also for function that don't have calls or are not referenced.
25212 2017-06-30 Richard Biener <rguenther@suse.de>
25214 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25215 analyze the first scalar stmt. Move vector type computation
25216 for the BB case here from ...
25217 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25218 live operation processing in the SLP case properly.
25220 2017-06-30 Richard Biener <rguenther@suse.de>
25222 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25224 2017-06-30 Martin Liska <mliska@suse.cz>
25227 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25228 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25230 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25232 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25235 2017-06-30 Richard Biener <rguenther@suse.de>
25237 PR tree-optimization/81249
25238 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25239 condition reduction result to original scalar type.
25241 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25243 * profile-count.h (enum profile_quality): Fix typos and whitespace
25246 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25248 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25249 type for branch probabilities.
25251 2017-06-29 Julian Brown <julian@codesourcery.com>
25252 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25254 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25255 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25256 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25257 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25259 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25261 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25262 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25263 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25264 CC usage from generic code to here.
25265 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25266 CC usage into the target macros.
25268 2017-06-29 Maya Rashish <coypu@sdf.org>
25270 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25273 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25275 * arm/arm-builtins.c: Include profile-count.h
25276 * except.c (sjlj_emit_function_enter): Use
25277 profile_probability::unlikely.
25279 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25281 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25282 and tocrel_offset be pointer args rather than implicitly using
25284 (legitimate_constant_pool_address_p, rs6000_emit_move,
25285 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25286 tocrel_offset and use in toc_relative_expr_p call.
25287 (print_operand, print_operand_address): Use static tocrel_base_oac
25288 and tocrel_offset_oac.
25289 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25292 2017-06-29 Maya Rashish <coypu@sdf.org>
25294 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25296 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25298 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25299 objects, take into account only the alignment of 'op0' and 'mode1' if
25302 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25304 * ccmp.c (ccmp_tree_comparison_p): New function.
25305 (ccmp_candidate_p): Update to use above function.
25306 (get_compare_parts): New function.
25307 (expand_ccmp_next): Update to use new functions.
25308 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25310 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25311 take mode as argument.
25312 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25313 * expr.c (expand_expr_real_1): Pass mode as argument.
25315 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25317 * combine.c (combine_instructions): Print insns to dump_file, together
25320 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25322 * asan.c (asan_emit_stack_protection): Update.
25323 (create_cond_insert_point): Update.
25324 * auto-profile.c (afdo_propagate_circuit): Update.
25325 * basic-block.h (struct edge_def): Turn probability to
25326 profile_probability.
25327 (EDGE_FREQUENCY): Update.
25328 * bb-reorder.c (find_traces_1_round): Update.
25329 (better_edge_p): Update.
25330 (sanitize_hot_paths): Update.
25331 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25332 (make_single_succ_edge): Update.
25333 (check_bb_profile): Update.
25334 (dump_edge_info): Update.
25335 (update_bb_profile_for_threading): Update.
25336 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25337 probabilitycount to 0.
25338 * cfgbuild.c (compute_outgoing_frequencies): Update.
25339 * cfgcleanup.c (try_forward_edges): Update.
25340 (outgoing_edges_match): Update.
25341 (try_crossjump_to_edge): Update.
25342 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25343 (expand_gimple_tailcall): Update.
25344 (construct_init_block): Use make_single_succ_edge.
25345 (construct_exit_block): Use make_single_succ_edge.
25346 * cfghooks.c (verify_flow_info): Update.
25347 (redirect_edge_succ_nodup): Update.
25348 (split_edge): Update.
25349 (account_profile_record): Update.
25350 * cfgloopanal.c (single_likely_exit): Update.
25351 * cfgloopmanip.c (scale_loop_profile): Update.
25352 (set_zero_probability): Remove.
25353 (duplicate_loop_to_header_edge): Update.
25354 * cfgloopmanip.h (loop_version): Update prototype.
25355 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25356 (force_nonfallthru_and_redirect): Update.
25357 (update_br_prob_note): Update.
25358 (rtl_verify_edges): Update.
25359 (purge_dead_edges): Update.
25360 (rtl_lv_add_condition_to_bb): Update.
25361 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25362 * cgraphunit.c (init_lowered_empty_function): Update.
25363 (cgraph_node::expand_thunk): Update.
25364 * cilk-common.c: Include profile-count.h
25365 * dojump.c (inv): Remove.
25366 (jumpifnot): Update.
25367 (jumpifnot_1): Update.
25368 (do_jump_1): Update.
25370 (do_jump_by_parts_greater_rtx): Update.
25371 (do_compare_rtx_and_jump): Update.
25372 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25373 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25374 * dwarf2cfi.c: Include profile-count.h
25375 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25376 (sjlj_emit_dispatch_table): Likewise.
25377 * explow.c: Include profile-count.h
25378 * expmed.c (emit_store_flag_force): Update.
25379 (do_cmp_and_jump): Update.
25380 * expr.c (compare_by_pieces_d::generate): Update.
25381 (compare_by_pieces_d::finish_mode): Update.
25382 (emit_block_move_via_loop): Update.
25383 (store_expr_with_bounds): Update.
25384 (store_constructor): Update.
25385 (expand_expr_real_2): Update.
25386 (expand_expr_real_1): Update.
25387 * expr.h (try_casesi, try_tablejump): Update prototypes.
25388 * gimple-pretty-print.c (dump_probability): Update.
25389 (dump_profile): New.
25390 (dump_gimple_label): Update.
25391 (dump_gimple_bb_header): Update.
25392 * graph.c (draw_cfg_node_succ_edges): Update.
25393 * hsa-gen.c (convert_switch_statements): Update.
25394 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25395 (find_if_case_1): Update.
25396 (find_if_case_2): Update.
25397 * internal-fn.c (expand_arith_overflow_result_store): Update.
25398 (expand_addsub_overflow): Update.
25399 (expand_neg_overflow): Update.
25400 (expand_mul_overflow): Update.
25401 (expand_vector_ubsan_overflow): Update.
25402 * ipa-cp.c (good_cloning_opportunity_p): Update.
25403 * ipa-split.c (split_function): Use make_single_succ_edge.
25404 * ipa-utils.c (ipa_merge_profiles): Update.
25405 * loop-doloop.c (add_test): Update.
25406 (doloop_modify): Update.
25407 * loop-unroll.c (compare_and_jump_seq): Update.
25408 (unroll_loop_runtime_iterations): Update.
25409 * lra-constraints.c (lra_inheritance): Update.
25410 * lto-streamer-in.c (input_cfg): Update.
25411 * lto-streamer-out.c (output_cfg): Update.
25412 * mcf.c (adjust_cfg_counts): Update.
25413 * modulo-sched.c (sms_schedule): Update.
25414 * omp-expand.c (expand_omp_for_init_counts): Update.
25415 (extract_omp_for_update_vars): Update.
25416 (expand_omp_ordered_sink): Update.
25417 (expand_omp_for_ordered_loops): Update.
25418 (expand_omp_for_generic): Update.
25419 (expand_omp_for_static_nochunk): Update.
25420 (expand_omp_for_static_chunk): Update.
25421 (expand_cilk_for): Update.
25422 (expand_omp_simd): Update.
25423 (expand_omp_taskloop_for_outer): Update.
25424 (expand_omp_taskloop_for_inner): Update.
25425 * omp-simd-clone.c (simd_clone_adjust): Update.
25426 * optabs.c (expand_doubleword_shift): Update.
25427 (expand_abs): Update.
25428 (emit_cmp_and_jump_insn_1): Update.
25429 (expand_compare_and_swap_loop): Update.
25430 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25431 * predict.c (predictable_edge_p): Update.
25432 (edge_probability_reliable_p): Update.
25433 (set_even_probabilities): Update.
25434 (combine_predictions_for_insn): Update.
25435 (combine_predictions_for_bb): Update.
25436 (propagate_freq): Update.
25437 (estimate_bb_frequencies): Update.
25438 (force_edge_cold): Update.
25439 * profile-count.c (profile_count::dump): Add missing space into dump.
25440 (profile_count::debug): Add newline.
25441 (profile_count::differs_from_p): Explicitly convert to unsigned.
25442 (profile_count::stream_in): Update.
25443 (profile_probability::dump): New member function.
25444 (profile_probability::debug): New member function.
25445 (profile_probability::differs_from_p): New member function.
25446 (profile_probability::differs_lot_from_p): New member function.
25447 (profile_probability::stream_in): New member function.
25448 (profile_probability::stream_out): New member function.
25449 * profile-count.h (profile_count_quality): Rename to ...
25450 (profile_quality): ... this one.
25451 (profile_probability): New.
25452 (profile_count): Update.
25453 * profile.c (compute_branch_probabilities): Update.
25454 * recog.c (peep2_attempt): Update.
25455 * sched-ebb.c (schedule_ebbs): Update.
25456 * sched-rgn.c (find_single_block_region): Update.
25457 (compute_dom_prob_ps): Update.
25458 (schedule_region): Update.
25459 * sel-sched-ir.c (compute_succs_info): Update.
25460 * stmt.c (struct case_node): Update.
25461 (do_jump_if_equal): Update.
25462 (get_outgoing_edge_probs): Update.
25463 (conditional_probability): Update.
25464 (emit_case_dispatch_table): Update.
25465 (expand_case): Update.
25466 (expand_sjlj_dispatch_table): Update.
25467 (emit_case_nodes): Update.
25468 * targhooks.c: Update.
25469 * tracer.c (better_p): Update.
25470 (find_best_successor): Update.
25471 * trans-mem.c (expand_transaction): Update.
25472 * tree-call-cdce.c: Update.
25473 * tree-cfg.c (gimple_split_edge): Upate.
25474 (move_sese_region_to_fn): Upate.
25475 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25476 * tree-eh.c (lower_resx): Upate.
25477 (cleanup_empty_eh_move_lp): Upate.
25478 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25479 * tree-inline.c (copy_edges_for_bb): Update.
25480 (copy_cfg_body): Update.
25481 * tree-parloops.c (gen_parallel_loop): Update.
25482 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25483 (gimple_gen_time_profiler): Update.
25484 * tree-ssa-dce.c (remove_dead_stmt): Update.
25485 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25486 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25487 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25488 (unloop_loops): Update.
25489 (try_peel_loop): Update.
25490 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25491 * tree-ssa-loop-split.c (connect_loops): Update.
25492 (split_loop): Update.
25493 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25494 (hoist_guard): Update.
25495 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25496 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25497 (value_replacement): Update.
25498 * tree-ssa-reassoc.c (branch_fixup): Update.
25499 * tree-ssa-tail-merge.c (replace_block_by): Update.
25500 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25501 (create_edge_and_update_destination_phis): Update.
25502 (compute_path_counts): Update.
25503 (recompute_probabilities): Update.
25504 (update_joiner_offpath_counts): Update.
25505 (freqs_to_counts_path): Update.
25506 (duplicate_thread_path): Update.
25507 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25508 (struct switch_conv_info): Update.
25509 (gen_inbound_check): Update.
25510 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25511 (vect_do_peeling): Update.
25512 (vect_loop_versioning): Update.
25513 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25514 (optimize_mask_stores): Update.
25515 * ubsan.c (ubsan_expand_null_ifn): Update.
25516 * value-prof.c (gimple_divmod_fixed_value): Update.
25517 (gimple_divmod_fixed_value_transform): Update.
25518 (gimple_mod_pow2): Update.
25519 (gimple_mod_pow2_value_transform): Update.
25520 (gimple_mod_subtract): Update.
25521 (gimple_mod_subtract_transform): Update.
25522 (gimple_ic): Update.
25523 (gimple_stringop_fixed_value): Update.
25524 (gimple_stringops_transform): Update.
25525 * value-prof.h: Update.
25527 2017-06-29 Carl Love <cel@us.ibm.com>
25529 * config/rs6000/rs6000-c.c: Add support for built-in functions
25530 vector signed int vec_signed (vector float);
25531 vector signed long long vec_signed (vector double);
25532 vector signed int vec_signed2 (vector double, vector double);
25533 vector signed int vec_signede (vector double);
25534 vector signed int vec_signedo (vector double);
25535 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25536 instruction generator.
25537 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25538 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25539 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25541 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25542 vunsignede_v2df): Add define_expands.
25543 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25544 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25545 VEC_UNSIGNEDO): Add definitions.
25546 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25547 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25548 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25549 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25550 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25551 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25552 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25553 vec_unsignede, vec_unsignedo): Add builtin defines.
25554 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25556 * doc/extend.texi: Update the built-in documentation file for the
25557 new built-in functions.
25559 2017-06-29 Richard Biener <rguenther@suse.de>
25561 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25562 reduction chains to LOOP_VINFO_REDUCTIONS.
25563 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25564 SLP reductions after processing reduction chains.
25566 2017-06-29 Nathan Sidwell <nathan@acm.org>
25568 * builtins.c (fold_builtin_FUNCTION): Use
25569 lang_hooks.decl_printable_name.
25571 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25573 PR middle-end/81194
25574 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25575 with only one label.
25576 * stmt.c (expand_case): Assert NCASES is greater than one.
25578 2017-06-29 Richard Biener <rguenther@suse.de>
25580 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25582 (group_case_labels): Likewise.
25583 (find_taken_edge): Push sanity checking on val to workers...
25584 (find_taken_edge_cond_expr): ... here
25585 (find_taken_edge_switch_expr): ... and here, handle cases
25586 with just a default label.
25587 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25588 (group_case_labels): Likewise.
25589 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25590 group_case_labels does anything cleanup the CFG again.
25592 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25594 PR tree-optimization/81196
25595 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25596 exit condition comparing two IVs.
25598 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25600 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25601 profile to the dummy entry at the end of the list of architectures.
25602 * config/arm/arm-cpu-cdata.h: Regenerated.
25604 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25605 Michael Collison <michael.collison@arm.com>
25608 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25610 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25611 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25612 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25613 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25614 of shift when the shift amount is masked with constant equal to
25615 the size of the mode.
25616 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25619 2017-06-29 Martin Liska <mliska@suse.cz>
25621 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25623 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25626 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25629 (edge_growth_cache_entry::edge_growth_cache_entry): New
25631 (reset_edge_growth_cache): Update to use constructor.
25633 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25635 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25636 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25637 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25639 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25641 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25642 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25644 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25646 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25647 (*-linux-uclibc*): Add t-uclibc tmake_file.
25648 * config/t-musl: New.
25649 * config/t-uclibc: New.
25651 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25653 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25655 (gen_comm_data): Emit architectural setting of arch_prof.
25656 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25658 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25659 (armv8-m.base, armv8-m.main): Likewise.
25660 * arm-protos.h (arm_build_target): Add profile field.
25661 (arch_option): Likewise.
25662 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25664 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25665 arm_active_target.profile.
25667 2017-06-28 Richard Biener <rguenther@suse.de>
25669 PR middle-end/81227
25670 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25671 TYPE_OVERFLOW_WRAPS.
25672 * match.pd (negate_expr_p): Likewise.
25673 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25674 fold_build2, not fold_binary.
25676 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25678 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25679 Convert memory address to Pmode.
25680 (aarch64_print_operand): Assert MEM operands are always Pmode.
25682 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25685 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25686 Remove redundant if.
25687 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25688 * config/arm/aarch-common-protos.h
25689 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25690 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25692 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25695 * multiple_target.c (create_dispatcher_calls): Set the default
25696 clone to be static, not public.
25698 2017-06-28 Richard Biener <rguenther@suse.de>
25700 * tree-vect-loop.c (vectorizable_reduction): Move special
25701 cond reduction IV var creation ...
25702 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25703 parameter. Use STMT_VINFO_VECTYPE.
25704 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25707 2017-06-28 Martin Liska <mliska@suse.cz>
25710 * ipa-visibility.c (non_local_p): Handle visibility.
25712 2017-06-28 Martin Liska <mliska@suse.cz>
25715 * common.opt: Add IntegerRange to various options.
25716 * opt-functions.awk (integer_range_info): New function.
25717 * optc-gen.awk: Add integer_range_info to cl_options struct.
25718 * opts-common.c (decode_cmdline_option): Handle
25719 CL_ERR_INT_RANGE_ARG.
25720 (cmdline_handle_error): Likewise.
25721 * opts.c (print_filtered_help): Show valid interval in
25722 when --help is provided.
25723 * opts.h (struct cl_option): Add range_min and range_max fields.
25724 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25726 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25728 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25729 (x * C EQ/NE y * C): New transformation.
25731 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25733 * genmultilib (combination_space): Accept '+' in option names.
25735 2017-06-28 Martin Liska <mliska@suse.cz>
25738 * asan.c (instrument_derefs): Bail out inner references
25739 that are hard register variables.
25741 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25744 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25745 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25747 2017-06-28 Richard Biener <rguenther@suse.de>
25749 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25750 (vect_get_slp_defs): Adjust.
25751 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25753 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25755 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25756 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25757 (vectorizable_reduction): Adjust.
25758 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25760 (vect_get_slp_defs): Likewise.
25761 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25762 (vectorizable_bswap): Adjust.
25763 (vectorizable_call): Likewise.
25764 (vectorizable_conversion): Likewise.
25765 (vectorizable_assignment): Likewise.
25766 (vectorizable_shift): Likewise.
25767 (vectorizable_operation): Likewise.
25768 (vectorizable_store): Likewise.
25769 (vectorizable_condition): Likewise.
25770 (vectorizable_comparison): Likewise.
25772 2017-06-28 Michael Collison <michael.collison@arm.com>
25775 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25777 (arm_gen_movmemqi): Removed unused variable 'i'.
25778 Convert 'for' loop into 'while' loop.
25779 (arm_expand_prologue): Remove last unnecessary set of insn.
25780 (thumb_pop): Remove unused variable 'pushed_words'.
25781 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25783 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25785 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25786 * config/s390/s390-protos.h: Add prototype of
25787 s390_rel_address_ok_p.
25788 * config/s390/s390.c (s390_got_symbol): New function.
25789 (s390_rel_address_ok_p): New function.
25790 (legitimize_pic_address): Use s390_rel_address_ok_p.
25791 (s390_load_got): Use s390_got_symbol.
25792 (s390_option_override): Issue error if
25793 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25794 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25796 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25798 2017-06-27 Andrew Pinski <apinski@cavium.com>
25800 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25801 (X * copysign (1.0, X)): New pattern.
25802 (X * copysign (1.0, -X)): New pattern.
25803 (copysign (-1.0, CST)): New pattern.
25805 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25807 * genmultilib (combination_space): Remove variable.
25808 Validate reuse rules against regular expression for any sequence
25809 of multilib options in any order.
25811 2017-06-27 Michael Collison <michael.collison@arm.com>
25813 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25814 call aarch64_split_simd_combine.
25815 * (aarch64_combine_internal<mode>): Delete pattern.
25816 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25817 Allow register and subreg operands.
25819 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25821 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25822 specific need, just fallback on defaults.
25823 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25825 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25826 Olivier Hainque <hainque@adacore.com>
25828 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25830 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25831 targets. Pick a default if no particular attempt applied.
25832 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25835 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25837 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25838 (x86_64-wrs-vxworks7): Likewise.
25840 2017-06-27 Marek Polacek <polacek@redhat.com>
25843 * ubsan.c (instrument_null): Check get_base_address's result for null.
25845 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25847 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25849 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25851 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25852 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25853 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25854 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25855 New function types.
25856 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25857 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25858 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
25859 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
25860 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
25861 BUILT_IN_FEUPDATEENV): New builtins.
25862 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
25863 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
25864 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
25865 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
25867 (builtin_structptr_types): Adjust size.
25868 * tree.c (builtin_structptr_types): Add four entries.
25870 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25871 Olivier Hainque <hainque@adacore.com>
25873 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
25874 (TLS_SYM): New local macro, forcing reference to __tls__ on
25875 link command lines for VxWorks 7 RTPs, triggering initialization
25877 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
25878 OS features TLS support, true for RTPs on VxWorks 7.
25879 * config/vxworks.c (vxworks_override_options): Setup emutls
25882 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25884 * predict.c (test_prediction_value_range): Use -1U instead of -1
25885 to avoid narrowing conversion warning.
25886 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
25887 to avoid narrowing conversion warning.
25888 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
25890 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
25892 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25894 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
25895 64bit configurations.
25896 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
25897 (SIZE_TYPE): Likewise.
25898 * config/vxworks.c (vxworks_emutls_var_fields): Use
25899 long_unsigned_type_node instead of unsigned_type_node as the offset
25900 field type, which is "pointer" mode in emutls.c.
25902 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25905 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
25907 PR middle-end/81207
25908 * gimple-fold.c (replace_call_with_call_and_fold): Handle
25909 gimple_vuse copying separately from gimple_vdef copying.
25911 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25913 * value-prof.c (free_hist): Remove call to memset and the enclosing if
25916 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25917 Olivier Hainque <hainque@adacore.com>
25919 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
25920 for all vxworks7 targets.
25921 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
25922 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
25923 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
25924 variations for VX6/VX7 and 32/64bits later on in ...
25925 (VXWORKS_LIB_SPEC): Leverage new macros.
25926 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
25927 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
25929 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25931 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
25932 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
25934 2017-06-26 Carl Love <cel@us.ibm.com>
25936 * config/rs6000/rs6000-c.c: Add support for built-in functions
25937 vector bool char vec_reve (vector bool char);
25938 vector signed char vec_reve (vector signed char);
25939 vector unsigned char vec_reve (vector unsigned char);
25940 vector bool int vec_reve (vector bool int);
25941 vector signed int vec_reve (vector signed int);
25942 vector unsigned int vec_reve (vector unsigned int);
25943 vector bool long long vec_reve (vector bool long long);
25944 vector signed long long vec_reve (vector signed long long);
25945 vector unsigned long long vec_reve (vector unsigned long long);
25946 vector bool short vec_reve (vector bool short);
25947 vector signed short vec_reve (vector signed short);
25948 vector double vec_reve (vector double);
25949 vector float vec_reve (vector float);
25950 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
25951 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
25952 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
25953 (altivec_vreve): New pattern.
25954 * config/rs6000/altivec.h (vec_reve): New define.
25955 * doc/extend.texi (vec_rev): Update the built-in documentation file
25956 for the new built-in functions.
25958 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25960 PR tree-optimization/71815
25961 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
25963 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
25965 (slsr_process_phi): Likewise.
25966 (replace_uncond_cands_and_profitable_phis): Don't replace a
25967 multiply candidate with a stride of 1 (copy or cast).
25968 (phi_incr_cost): Call uses_consumed_by_stmt rather than
25970 (lowest_cost_path): Likewise.
25971 (total_savings): Likewise.
25973 2017-06-26 Richard Biener <rguenther@suse.de>
25976 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
25977 Use def_builtin_pure for all gather builtins.
25979 2017-06-26 Richard Biener <rguenther@suse.de>
25981 PR tree-optimization/81203
25982 * tree-tailcall.c (find_tail_calls): Do not move stmts into
25983 non-dominating BBs.
25985 2017-06-26 Marek Polacek <polacek@redhat.com>
25988 * doc/invoke.texi: Document -Wmultistatement-macros.
25990 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
25992 * doc/sourcebuild.texi (ARM-specific attributes): Document new
25993 arm_neon_ok_no_float_abi effective target.
25995 2017-06-26 Richard Biener <rguenther@suse.de>
25997 PR tree-optimization/80928
25998 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
25999 (copy_bbs): Set BB_DUPLICATED flag early.
26000 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26002 (execute_on_shrinking_pred): Likewise.
26003 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26004 BB_DUPLICATED blocks.
26005 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26006 iterate over all PHIs considering removal of *gsi.
26008 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26010 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26013 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26015 * config/rs6000/rs6000-string.c: (expand_block_clear,
26016 do_load_for_compare, select_block_compare_mode,
26017 compute_current_alignment, expand_block_compare,
26018 expand_strncmp_align_check, expand_strn_compare,
26019 expand_block_move, rs6000_output_load_multiple)
26020 Move functions related to string/block move/compare
26021 to a separate file.
26022 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26023 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26024 for this function which is now used in two files.
26025 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26026 * config.gcc: Add rs6000-string.o to extra_objs for
26027 targets powerpc*-*-* and rs6000*-*-*.
26029 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26032 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26033 32-bit, since indexed is not valid for DImode.
26034 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26035 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26036 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26037 (define_peephole2 for Altivec d-form store): Likewise.
26040 * multiple_target.c (create_dispatcher_calls): Only create the
26041 dispatcher call if the function is the default clone of a
26042 versioned function.
26044 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26046 PR middle-end/80902
26047 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26048 a call, force the call to not be a tail call.
26050 2017-06-23 Jeff Law <law@redhat.com>
26052 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26055 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26057 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26058 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26059 handling for early expansion of vector shifts (sl,sr,sra,rl).
26060 (builtin_function_type): Add vector shift right instructions
26061 to the unsigned argument list.
26063 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26065 rtl-optimizatoin/79286
26066 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26067 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26068 trap. PIC register plus a const unspec without offset can never trap.
26070 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26072 * tree.h (builtin_structptr_type): New type.
26073 (builtin_structptr_types): Declare new array.
26074 * tree.c (builtin_structptr_types): New array.
26075 (free_lang_data, build_common_tree_nodes): Use it.
26077 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26080 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26083 2017-06-22 Matt Turner <mattst88@gmail.com>
26085 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26086 Lake models to skylake case. Assume skylake for unknown
26087 models with clflushopt.
26089 2017-06-22 Jeff Law <law@redhat.com>
26091 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26092 frame sizes that do not satisfy aarch64_uimm12_shift.
26094 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26096 * profile-count.h (apply_probability,
26097 apply_scale, probability_in): Fix checks for zero.
26099 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26101 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26102 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26104 2016-06-22 Richard Biener <rguenther@suse.de>
26106 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26107 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26108 REDUC_MAX_EXPR support.
26109 (vectorizable_reduction): Likewise.
26110 (vect_create_epilog_for_reduction): Likewise.
26112 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26114 * match.pd (A / (1 << B) -> A >> B): New.
26115 * generic-match-head.c: Include optabs-tree.h.
26116 * gimple-match-head.c: Likewise.
26117 * optabs-tree.h (target_supports_op_p): New.
26118 * optabs-tree.c (target_supports_op_p): New.
26120 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26122 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26123 $gcc_cv_ld --help output.
26124 (gcc_cv_ld_demangle): Likewise.
26125 (gcc_cv_ld_eh_frame_hdr): Likewise.
26126 (gcc_cv_ld_pie): Likewise.
26127 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26128 (gcc_cv_ld_buildid): Likewise.
26129 (gcc_cv_ld_sysroot): Likewise.
26130 (ld_bndplt_support): Likewise.
26131 (ld_pushpopstate_support): Likewise.
26132 * configure: Regenerate.
26133 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26135 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26138 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26139 operands indexes to avoid gap between operands and match_dups.
26141 2017-06-21 Andrew Pinski <apinski@cavium.com>
26143 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26144 Increment Arith_shift and Arith_shift_reg by 1.
26145 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26147 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26148 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26149 (aarch64_strip_extend): Add new argument and test for it.
26150 (aarch64_cheap_mult_shift_p): New function.
26151 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26152 add a cost if it is true.
26153 Update calls to aarch64_strip_extend.
26154 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26156 2017-06-21 Andrew Pinski <apinski@cavium.com>
26158 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26160 (thunderxt88): Likewise.
26161 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26162 (thunderx_prefetch_tune): New variable.
26163 (thunderx2t99_prefetch_tune): Update for the correct values.
26164 (thunderxt88_tunings): New variable.
26165 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26166 generic_prefetch_tune.
26167 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26169 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26171 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26172 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26173 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26174 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26175 (aarch64_atomic_cas<mode>, GPI): Likewise.
26177 2017-06-21 Martin Liska <mliska@suse.cz>
26179 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26180 statements on cold and hot labels.
26181 * predict.c (tree_estimate_probability_bb): Remove the
26182 prediction from this place.
26184 2017-06-21 Martin Liska <mliska@suse.cz>
26186 PR tree-optimization/79489
26187 * gimplify.c (maybe_add_early_return_predict_stmt): New
26189 (gimplify_return_expr): Call the function.
26190 * predict.c (tree_estimate_probability_bb): Remove handling
26192 * predict.def: Update comment about early return predictor.
26193 * gimple-predict.h (is_gimple_predict): New function.
26194 * predict.def: Change default value of early return to 66.
26195 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26197 * passes.def: Put pass_strip_predict_hints to the beginning of
26200 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26202 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26203 FUNCTION_DECL declarations.
26204 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26206 (dwaf2out_decl): Likewise.
26207 * godump.c (go_early_global_decl): Skip call to the real debug hook
26208 for FUNCTION_DECL declarations.
26209 * passes.c (rest_of_decl_compilation): Skip call to the
26210 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26211 -fdump-go-spec is passed.
26213 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26215 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26216 Reorder for compactness.
26217 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26218 (def_builtin_pure, def_builtin_pure2): New functions.
26219 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26221 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26223 * match.pd (nop_convert): New predicate.
26224 ((A +- CST1) +- CST2): Allow some NOP conversions.
26226 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26229 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26230 with ctors/dtors if GOVD_SHARED is set.
26232 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26234 * config/aarch64/aarch64.md (movti_aarch64):
26235 Emit mov rather than orr.
26236 (movtf_aarch64): Likewise.
26237 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26238 Emit mov rather than orr.
26240 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26242 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26243 Swap alternatives, make integer dup more expensive.
26245 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26247 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26248 Return true for non-tls symbols.
26250 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26252 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26253 (cortex-a75): Likewise.
26254 (cortex-a75.cortex-a55): Likewise.
26255 * config/aarch64/aarch64-tune.md: Regenerate.
26256 * doc/invoke.texi (-mtune): Document new values for -mtune.
26258 2017-06-21 Tom de Vries <tom@codesourcery.com>
26260 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26261 stack_size feature.
26262 (Effective-Target Keywords, Other attributes): Suggest using
26263 dg-add-options stack_size feature to get stack limit in stack_size
26264 effective target documentation.
26266 2017-06-21 Julian Brown <julian@codesourcery.com>
26267 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26269 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26270 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26271 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26273 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26274 attribute type list for neon_multiply.
26275 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26276 attribute type list for neon_multiply.
26277 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26278 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26279 attribute type list for neon_multiply.
26280 * config/arm/types.md (crypto_pmull): Add.
26281 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26282 attribute type list.
26284 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26286 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26289 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26291 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26292 to make sure not to dereference a NULL cost_classes_ptr pointer.
26294 2017-06-20 Carl Love <cel@us.ibm.com>
26296 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26297 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26298 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26299 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26300 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26301 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26302 VMULOSW): New enum "unspec" values.
26303 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26304 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26305 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26306 altivec_vmulosw): New patterns.
26307 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26308 VMULOSW): Add definitions.
26310 2017-06-20 Julia Koval <julia.koval@intel.com>
26312 * config/i386/i386.c: Fix rounding expand for new pattern.
26313 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26315 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26317 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26318 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26320 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26322 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26325 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26327 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26328 architecture, then by implementer ID.
26329 * config/aarch64/aarch64-tune.md: Regenerate.
26331 2017-06-20 Richard Biener <rguenther@suse.de>
26333 PR middle-end/81097
26334 * fold-const.c (split_tree): Fold to type before negating.
26336 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26338 * diagnostic-show-locus.c
26339 (selftest::test_fixit_deletion_affecting_newline): New function.
26340 (selftest::diagnostic_show_locus_c_tests): Call it.
26342 2017-06-20 Andreas Schwab <schwab@suse.de>
26345 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26348 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26350 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26351 __ARM_FEATURE_COPROC according to support.
26353 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26355 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26356 Rewritten to avoid overflow for > 32-bit pointers.
26359 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26360 by removing enum keyword.
26361 (ubsan_type_descriptor): Likewise. Formatting fix.
26364 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26365 splitter): Require TARGET_SSE2 in the condition.
26367 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26370 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26371 for doing vector set of SFmode on ISA 3.0.
26372 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26373 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26375 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26376 SFmode value into a V4SF variable that was extracted from another
26377 V4SF variable without converting the element to double precision
26378 and back to single precision vector format.
26379 (vsx_insert_extract_v4sf_p9_2): Likewise.
26381 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26383 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26384 in UWHI to avoid undefined overflow.
26387 * ubsan.h (enum ubsan_encode_value_phase): New.
26388 (ubsan_encode_value): Change second argument to
26389 enum ubsan_encode_value_phase with default value of
26390 UBSAN_ENCODE_VALUE_GENERIC.
26391 * ubsan.c (ubsan_encode_value): Change second argument to
26392 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26393 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26394 create_tmp_var_raw instead of create_tmp_var and use a
26396 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26397 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26398 ubsan_encode_value callers.
26401 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26402 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26403 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26405 2017-06-19 Richard Biener <rguenther@suse.de>
26407 PR middle-end/81118
26408 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26409 estimates if we changed anything.
26411 2017-06-19 Richard Biener <rguenther@suse.de>
26413 PR tree-optimization/80887
26414 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26415 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26416 simplified lookups, then reset mprts_hook.
26417 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26419 (try_to_simplify): Likewise.
26421 2017-06-19 Martin Liska <mliska@suse.cz>
26424 * gimplify.c (gimplify_switch_expr):
26425 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26427 2017-06-19 Martin Liska <mliska@suse.cz>
26429 * doc/install.texi: Document that PGO runs in 4 stages.
26431 2017-06-19 Martin Liska <mliska@suse.cz>
26434 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26435 to dispatcher function name.
26436 * multiple_target.c (replace_function_decl): New function.
26437 (create_dispatcher_calls): Redirect both edges and references.
26439 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26441 * profile-count.c (profile_count::dump): Dump quality.
26442 (profile_count::differs_from_p): Update for unsigned val.
26443 * profile-count.h (profile_count_quality): New enum.
26444 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26446 2017-06-19 Richard Biener <rguenther@suse.de>
26448 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26449 struct function as arg.
26450 (estimate_numbers_of_iterations): Export overload with loop arg.
26451 (free_numbers_of_iterations_estimates_loop): Use an overload of
26452 free_numbers_of_iterations_estimates instead.
26453 * tree-cfg.c (remove_bb): Adjust.
26454 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26455 * tree-parloops.c (gen_parallel_loop): Likewise.
26456 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26458 (tree_unroll_loops_completely): Likewise.
26459 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26460 Use an overload instead and export.
26461 (estimated_loop_iterations): Adjust.
26462 (max_loop_iterations): Likewise.
26463 (likely_max_loop_iterations): Likewise.
26464 (estimate_numbers_of_iterations): Take struct function as arg
26466 (loop_exits_before_overflow): Adjust.
26467 (free_numbers_of_iterations_estimates_loop): Use an overload.
26468 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26469 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26471 2017-06-19 Richard Biener <rguenther@suse.de>
26474 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26475 RANGE_EXPR conservatively.
26477 2017-06-16 Carl Love <cel@us.ibm.com>
26479 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26480 definitions for vec_float, vec_float2, vec_floato,
26481 vec_floate built-ins.
26482 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26483 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26485 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26486 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26487 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26488 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26489 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26490 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26491 vec_floato): Add builtin defines.
26492 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26493 Update the built-in documentation file for the new built-in
26496 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26498 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26499 (mthumb): Mark as the negative of -marm.
26501 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26503 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26505 (ARM Options, -mtune): Document that this accepts the same
26506 extension options as -mcpu.
26507 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26509 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26511 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26512 permitted extensions.
26514 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26516 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26517 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26518 (armv8-m.main): Add option +nodsp.
26519 * config/arm/arm-cpu-cdata.h: Regenerated.
26521 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26523 * config/arm/t-fuchsia: New file.
26524 * config.gcc (arm*-*-fuchsia*): Use it.
26526 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26528 * config/arm/t-symbian: Rewrite for new option infrastructure.
26530 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26532 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26533 (MULTILIB_REQUIRED): Likewise.
26535 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26537 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26538 (MULTILIB_RESUE): Likewise.
26539 (MULTILIB_MATCHES): Likewise.
26540 (MULTLIB_REQUIRED): Likewise.
26542 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26544 * config/arm/t-rtems: Rewrite for new option framework.
26546 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26548 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26549 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26550 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26551 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26552 * config/arm/t-multilib: ... here.
26553 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26554 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26555 armv7-a and armv8*-a when A-profile libraries have not been built.
26556 * config/arm/t-rmprofile: Rewrite.
26558 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26560 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26561 with a backslash. Remove the backslash after substituting unescaped
26563 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26565 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26567 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26568 the driver to pass through the default setting of -mfloat-abi.
26569 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26571 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26572 (all_feat_combs): New rule.
26573 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26575 * config/arm/t-aprofile: Rewrite.
26577 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26579 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26580 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26581 fpu is not specified by the user/command-line.
26582 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26583 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26584 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26585 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26586 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26587 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26589 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26591 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26592 * config/arm/t-arm-elf: Rewritten.
26594 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26596 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26597 have some floating-point instructions.
26598 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26599 (TARGET_MAYBE_HARD_FLOAT): New macro.
26600 * config/arm/arm-builtins.c (arm_init_builtins): Use
26601 TARGET_MAYBE_HARD_FLOAT.
26602 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26604 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26606 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26607 (configargs.h): Include it.
26608 (arm_print_hint_for_fpu_option): New function.
26609 (arm_parse_fpu_option): New function.
26610 (candidate_extension): New class.
26611 (arm_canon_for_multilib): New function.
26612 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26613 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26614 (ARCH_CANONICAL_SPECS): New macro.
26615 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26617 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26619 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26620 are set after handling multilib fragments. Set target_cpu_default2
26623 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26625 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26627 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26628 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26630 (check_arch): Likewise.
26631 * config/arm/arm.c (arm_configure_build_target): Handle
26632 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26633 options in the default.
26635 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26637 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26638 when an option is an alias of another.
26639 * config/arm/parsecpu.awk (optalias): New parser token.
26640 (gen_comm_data): Mark non-alias options as such. Emit entries
26641 for extension aliases.
26642 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26643 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26644 (armv6kz, armv6zk, armv6t2): Likewise.
26645 (armv7): Make vfpv3-d16 an alias.
26646 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26648 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26649 Sort in canonical order.
26650 (armv8-a): Sort in canonical order.
26651 (armv8.1-a, armv8.2-a): Likewise.
26652 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26654 (cortex-a9): Sort in canonical order.
26655 * config/arm/arm.c (selftests.h): Include it.
26656 (arm_test_cpu_arch_data): New function.
26657 (arm_run_self_tests): New function.
26658 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26659 (targetm): Move declaration to the end of the file.
26660 * arm-cpu-cdata.h: Regenerated.
26662 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26664 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26665 call to target_mode_check describing the type of option passed.
26666 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26667 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26668 arm_parse_cpu_option_name to match parameters against list of
26670 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26671 arm_arch_core_flags data structure.
26672 * config/arm/arm-cpu_cdata.h: Regenerated.
26674 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26676 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26678 (arm_print_hint_for_cpu_option): Likewise.
26679 (arm_print_hint_for_arch_option): Likewise.
26680 (arm_parse_cpu_option_name): Likewise.
26681 (arm_parse_arch_option_name): Likewise.
26682 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26683 of entries in the all_fpus list.
26684 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26685 (arm_parse_cpu_option_name): Declare.
26686 (arm_parse_arch_option_name): Declare.
26687 (arm_parse_option_features): Declare.
26688 (arm_intialize_isa): Declare.
26689 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26691 (gen_comm_data): ... here. Make definitions non-static.
26692 * config/arm/arm-cpu-data.h: Regenerated.
26693 * config/arm/arm-cpu-cdata.h: Regenerated.
26695 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26697 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26698 (cpu_arch_extension): New structure.
26699 (cpu_arch_option, arch_option, cpu_option): New structures.
26700 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26701 architecture types.
26702 (gen_data): Generate new format data tables.
26703 * config/arm/arm.c (cpu_tune): New structure.
26704 (cpu_option, processors): Delete.
26705 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26706 (arm_print_hint_for_cpu_option): ... this and ...
26707 (arm_print_hint_for_arch_option): ... this.
26708 (arm_parse_arch_cpu_name): Delete. Replace with ...
26709 (arm_parse_cpu_option_name): ... this and ...
26710 (arm_parse_arch_option_name): ... this.
26711 (arm_unrecognized_feature): Change type of target parameter to
26713 (arm_parse_arch_cpu_features): Delete. Replace with ...
26714 (arm_parse_option_features): ... this.
26715 (arm_configure_build_target): Rework to use new configuration data
26717 (arm_print_tune_info): Rework for new configuration data tables.
26718 * config/arm/arm-cpu-data.h: Regenerated.
26719 * config/arm/arm-cpu.h: Regenerated.
26721 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26723 * Makefile.in (OBJS): Move sbitmap.o from here ...
26724 (OBJS-libcommon): ... to here.
26726 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26728 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26729 (ISA_ALL_CRYPTO): New macro.
26730 (ISA_ALL_SIMD): New macro
26731 (ISA_ALL_FP): New macro.
26732 * config/arm/arm.c (fpu_bitlist): Update initializer.
26733 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26735 (arm9e): Add fpu. Add option for nofp
26736 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26737 (arm926ej-s, arm1026ej-s): Likewise.
26738 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26739 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26740 neon-fp16, neon-vfpv4, nofp and nosimd.
26741 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26742 (cortex-a8): Add fpu. Add option for nofp.
26743 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26744 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26745 (cortex-r4f): Add fpu.
26746 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26747 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26749 (cortex-r8): Likewise.
26750 (cortex-m4): Add fpu. Add option for nofp.
26751 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26752 (cortex-a17.cortex-a7): Likewise.
26753 (cortex-a32): Add fpu. Add options for crypto and nofp.
26754 (cortex-a35, cortex-a53): Likewise.
26755 (cortex-a57): Add fpu. Add option for crypto.
26756 (cortex-a72, cortex-a73): Likewise.
26757 (exynos-m1): Likewise.
26758 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26759 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26760 (cortex-m33): Add fpu. Add option for nofp.
26761 * config/arm/arm-cpu-cdata.h: Regenerated
26762 * config/arm/arm-cpu-data.h: Regenerated.
26764 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26766 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26767 (armv5te, armv5tej): Likewise.
26768 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26769 (armv7): Add options fp and vfpv3-d16.
26770 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26771 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26773 (armv7ve): Likewise.
26774 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26775 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26776 (armv8-a): Add nocrypto option.
26777 (armv8.1-a, armv8.2-a): Likewise.
26778 (armv8-m.main): add options fp, fp.dp and nofp.
26780 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26782 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26784 (armv8-a+crc): Delete.
26785 (armv8.1-a): Add options simd, crypto and nofp.
26786 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26787 (armv8.2-a+fp16): Delete.
26788 (armv8-m.main): Add option dsp.
26789 (armv8-m.main+dsp): Delete.
26790 (cortex-a8): Add fpu. Add nofp option.
26791 (cortex-a9): Add fpu. Add nofp and nosimd options.
26792 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26793 link to main cpu and architecture data structures.
26794 (gen_comm_data): Only put isa attributes from the main architecture
26796 (option): New statement for architecture and CPU entries.
26797 * arm.c (struct cpu_option): New structure.
26798 (struct processors): Add entry for options.
26799 (arm_unrecognized_feature): New function.
26800 (arm_parse_arch_cpu_name): Ignore any characters after the first
26802 (arm_parse_arch_cpu_feature): New function.
26803 (arm_configure_build_target): Separate out any CPU and architecture
26804 features and parse separately. Don't error out if -mfpu=auto is
26805 used with only an architecture string.
26806 (arm_print_asm_arch_directives): New function.
26807 (arm_file_start): Call it.
26808 * config/arm/arm-cpu-cdata.h: Regenerated.
26809 * config/arm/arm-cpu-data.h: Likewise.
26810 * config/arm/arm-tables.opt: Likewise.
26812 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26814 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26815 assembler when it is not -mfpu=auto.
26817 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26819 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26820 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26821 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26822 (ASM_CPU_SPEC): Rewrite.
26823 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26824 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26825 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26826 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26827 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26828 copied string is NUL-terminated. Also strip any characters prefixed
26830 (arm_rewrite_selected_arch): New function.
26831 (arm_rewrite_march): New function.
26833 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26835 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26836 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26837 (march, mcpu, mtune): Convert to string-based options.
26838 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26839 (arm_parse_arch_cpu_name): New function.
26840 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26841 identify selected architecture or CPU.
26842 (arm_option_save): New function.
26843 (TARGET_OPTION_SAVE): Redefine.
26844 (arm_option_restore): Restore string options.
26845 (arm_option_print): Print string options.
26847 2017-06-16 Martin Sebor <msebor@redhat.com>
26849 PR tree-optimization/80933
26850 PR tree-optimization/80934
26851 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26852 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26853 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26854 (gimple_fold_builtin): Call them.
26856 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26858 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
26859 as unlikely; update profile.
26861 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26863 * predict.c (force_edge_cold): Handle declaring edges impossible
26866 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26868 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
26870 (try_unroll_loop_completely): Fix reporting.
26872 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26874 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
26876 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
26879 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
26880 if given a non-constant argument for an intrinsic which requires a
26883 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26885 * profile.c (compare_freqs): New function.
26886 (branch_prob): Sort edge list.
26887 (find_spanning_tree): Assume that the list is priority sorted.
26889 2017-06-16 Richard Biener <rguenther@suse.de>
26891 PR tree-optimization/81090
26892 * passes.def (pass_record_bounds): Remove.
26893 * tree-pass.h (make_pass_record_bounds): Likewise.
26894 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
26895 make_pass_record_bounds): Likewise.
26896 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
26897 not free niter estimates at the beginning but at the end.
26898 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
26900 2017-06-16 Richard Biener <rguenther@suse.de>
26902 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
26903 initializer to workaround ICE in host GCC 4.8.
26905 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26907 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
26909 (clone_inlined_nodes): Update.
26911 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26913 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
26914 prefetch settings, and enable prefetching by default at -O3.
26916 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26918 * config/aarch64/aarch64.c (aarch64_override_options_internal):
26919 Set flag_prefetch_loop_arrays according to tuning data.
26921 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26923 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
26924 New tune structure.
26925 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
26926 [Unrelated to main purpose of the patch] Place the pointer field last
26927 to enable type checking errors when tune structure are wrongly merged.
26928 * config/aarch64/aarch64.c (generic_prefetch_tune,)
26929 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
26930 (thunderx2t99_prefetch_tune): New tune constants.
26931 (tune_params *_tunings): Update all tunings (no functional change).
26932 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
26933 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
26934 from tunings structures.
26936 2017-06-16 Jakub Jelinek <jakub@redhat.com>
26939 * ubsan.c (instrument_null): Add T argument, use it instead
26940 of computing it based on IS_LHS.
26941 (instrument_object_size): Likewise.
26942 (pass_ubsan::execute): Adjust instrument_null and
26943 instrument_object_size callers to pass gimple_get_lhs or
26944 gimple_assign_rhs1 result to it. Use instrument_null instead of
26945 calling get_base_address and instrument_mem_ref. Handle
26946 aggregate call arguments for object-size sanitization.
26948 2017-06-16 Yury Gribov <tetra2005@gmail.com>
26950 PR tree-optimization/81089
26951 * tree-vrp.c (is_masked_range_test): Validate operands of
26954 2017-06-15 Martin Sebor <msebor@redhat.com>
26957 * dumpfile.c (dump_register): Avoid calling memset to initialize
26958 a class with a default ctor.
26959 * gcc.c (struct compiler): Remove const qualification.
26960 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
26961 * hash-table.h: Ditto.
26962 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
26964 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
26965 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
26967 * params.h (struct param_info): Make struct members non-const.
26968 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
26969 with default initialization.
26970 * vec.h (vec_copy_construct, vec_default_construct): New helper
26972 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
26973 with vec_copy_construct.
26974 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
26975 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
26976 * doc/invoke.texi (-Wclass-memaccess): Document.
26978 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26980 * emit-rtl.h (is_leaf): Update comment about local
26981 register allocator.
26983 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26986 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
26987 for a variable to have a section before checking if the section has a
26989 Set section to.persistent if persistent attribute is set.
26990 Warn if .persistent attribute is used on an automatic variable.
26992 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
26994 PR rtl-optimization/80474
26995 * reorg.c (update_block): Do not ignore instructions in a delay slot.
26997 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
26999 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27002 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27004 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27005 (casesi): Emit bounds checking as RTL.
27006 (casesi_internal_mips16_<mode>): Remove bounds checking.
27008 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27010 * config/xtensa/xtensa.c (xtensa_option_override): Append
27011 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27012 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27013 xtensa_doloop_hooks): Define unconditionally.
27014 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27016 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27017 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27018 for it in xtensa_option_override.
27019 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27020 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27022 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27024 * doc/cppopts.texi: Document '-' special value to -MF.
27026 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27028 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27029 (cortex_a53_fconst): Likewise.
27030 (cortex_a53_fpmul): Likewise.
27031 (cortex_a53_f_load_64): Likewise.
27032 (cortex_a53_f_load_many): Likewise.
27033 (cortex_a53_advsimd_alu): Likewise.
27034 (cortex_a53_advsimd_alu_q): Likewise.
27035 (cortex_a53_advsimd_mul): Likewise.
27036 (cortex_a53_advsimd_mul_q): Likewise.
27037 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27038 Add missing fmul, r2f_cvt and fconst cases.
27040 2017-06-14 Richard Biener <rguenther@suse.de>
27042 PR middle-end/81088
27043 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27045 (fold_binary_loc): When associating do not treat pre-existing
27046 TREE_OVERFLOW on literal constants as a reason to allow
27047 TREE_OVERFLOW on associated literal constants.
27049 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27051 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27052 (MASK_FEATURES): New macro.
27053 * config/sparc/sparc.c (sparc_option_override): Remove the special
27054 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27056 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27058 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27059 a division of 0 if non-call exceptions are enabled.
27061 2017-06-14 Andrew Pinski <apinski@cavium.com>
27062 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27065 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27066 Improve vector initialization code gen for only variable case.
27068 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27070 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27072 2017-06-14 Richard Biener <rguenther@suse.de>
27074 PR tree-optimization/81083
27075 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27078 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27080 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27081 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27082 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27083 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27084 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27085 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27087 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27089 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27090 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27092 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27094 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27096 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27098 * config/rs6000/t-rtems: Don't handle SPE.
27100 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27102 * config/rs6000/t-linux: Don't handle SPE.
27104 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27106 * config/rs6000/eabispe.h: Delete file.
27108 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27110 * config/rs6000/t-spe: Delete file.
27112 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27114 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27115 (rs6000_legitimate_offset_address_p): Return false for anything in
27116 V2SImode or V2SFmode.
27118 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27120 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27121 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27122 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27124 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27125 (rs6000_legitimate_offset_address_p): Ditto.
27126 (rs6000_emit_move): Ditto.
27127 (rs6000_init_builtins): Remove V4HI_type_node.
27129 2017-06-13 Martin Liska <mliska@suse.cz>
27132 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27133 (gate_asan): Likewise.
27134 * asan.h (asan_no_sanitize_address_p): Remove the function.
27135 (sanitize_flags_p): New function.
27136 * builtins.def: Fix coding style.
27137 * common.opt: Use renamed enum value.
27138 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27139 * doc/extend.texi: Document no_sanitize attribute.
27140 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27141 to SANITIZE_UNDEFINED_NONDEFAULT.
27142 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27143 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27144 Use sanitize_flags_p.
27145 * gimplify.c (gimplify_function_tree): Likewise.
27146 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27147 * opts.c (parse_no_sanitize_attribute): New function.
27148 (common_handle_option): Use renamed enum value.
27149 * opts.h (parse_no_sanitize_attribute): Declare.
27150 * tree.c (sanitize_flags_p): New function.
27151 * tree.h: Declared here.
27152 * tsan.c: Use sanitize_flags_p.
27153 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27154 (instrument_mem_ref): Likewise.
27155 (instrument_bool_enum_load): Likewise.
27156 (do_ubsan_in_current_function): Remove the function.
27157 (pass_ubsan::execute): Use sanitize_flags_p.
27158 * ubsan.h: Remove do_ubsan_in_current_function
27159 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27160 (dump_function_to_file): Use it here.
27162 2017-06-13 Martin Jambor <mjambor@suse.cz>
27164 PR tree-optimization/80803
27165 PR tree-optimization/81063
27166 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27167 (propagate_subaccesses_across_link): Enqueue subtree whenever
27168 necessary instead of relying on the caller.
27170 2017-06-13 Martin Jambor <mjambor@suse.cz>
27172 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27173 that have a first_link.
27174 (sort_and_splice_var_accesses): Do not check first_link before
27176 (subtree_mark_written_and_enqueue): Likewise.
27177 (propagate_all_subaccesses): Likewise and do not stop at first
27178 parent with a first_link.
27180 2017-06-13 Martin Jambor <mjambor@suse.cz>
27182 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27185 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27187 * match.pd: New pattern.
27189 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27191 * tree-vrp.c (is_masked_range_test): New function.
27192 (register_edge_assert_for): Determine ranges for
27195 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27197 PR tree-optimization/67328
27198 * fold-const.c (maskable_range_p): New function.
27199 (build_range_check): Generate bittests if possible.
27201 2017-06-13 Martin Liska <mliska@suse.cz>
27203 * gimple-pretty-print.c (dump_probability): Add new argument.
27204 (dump_edge_probability): Dump both probability and count.
27205 (dump_gimple_label): Likewise.
27206 (dump_gimple_bb_header): Likewise.
27208 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27211 * config/avr/avr-devices.c: Fix indentation.
27212 * config/avr/gen-avr-mmcu-specs.c: Dito.
27214 2017-06-13 Richard Biener <rguenther@suse.de>
27216 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27217 instead get vector type from stmt_info.
27218 (vectorizable_reduction): Adjust. Remove dead code.
27220 2017-06-13 Richard Biener <rguenther@suse.de>
27222 PR middle-end/81065
27223 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27224 case of C * (x * C2 + C3).
27225 (fold_addr_of_array_ref_difference): Properly fold index difference.
27227 2017-06-12 David S. Miller <davem@davemloft.net>
27230 * config/sparc/sparc.md (return expander): Emit frame blockage if
27231 function uses alloca.
27233 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27235 * combine.c (make_field_assignment): Check len rather than the mode
27236 precision when calling force_to_mode.
27238 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27240 Support multilibs and devices that see flash in RAM address range.
27243 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27244 (avr_mcu_t) <flash_pm_offset>: New field.
27245 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27246 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27247 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27248 (AVR_TINY_PM_OFFSET): Remove macro.
27249 * config/avr/avr.opt (-mshort-calls): New option.
27250 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27251 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27252 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27253 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27254 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27255 instead of avr_arch->have_jmp_call.
27256 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27257 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27258 avr_arch->flash_pm_offset to define.
27259 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27260 new field flash_pm_offset. Add entry for avrxmega3.
27261 (avr_texinfo): Add entry for avrxmega3.
27262 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27263 attiny212, attiny214,
27264 attiny412, attiny414, attiny416, attiny417,
27265 attiny814, attiny816, attiny817,
27266 attiny1614, attiny1616, attiny1617,
27267 attiny3214, attiny3216, attiny3217.
27268 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27269 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27270 (avr_print_operand_address) [AVR_TINY]: Same.
27271 (avr_asm_init_sections) <readonly_data_section>: Only patch
27272 callback if avr_arch->flash_pm_offset = 0.
27273 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27274 for rodata if avr_arch->flash_pm_offset != 0.
27275 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27276 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27277 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27278 (m_options): Append opt_rcall.
27279 (m_dirnames): Append dir_rcall.
27280 * config/avr/t-multilib: Regenerate.
27282 * configure.ac [target=avr]: Check whether avrxmega3 default
27283 linker description file works as needed.
27284 * configure: Regenerate.
27285 * doc/avr-mmcu.texi: Regenerate.
27286 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27287 <__AVR_ARCH__>: Document avrxmega3 and 103.
27288 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27289 <__AVR_SHORT_CALLS__>: Document it.
27290 <__AVR_PM_BASE_ADDRESS__>: Document it.
27291 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27292 (AVR Variable Attributes) <progmem>: Document this is
27293 not needed for avrxmega3.
27294 (AVR Named Address Spaces) <__flash>: Dito.
27296 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27298 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27300 2017-06-12 Doug Rupp <rupp@adacore.com>
27302 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27303 Append vxworks-stdint.h to the tm_file list.
27304 * config/vxworks-stdint.h: New file.
27306 2017-06-12 Martin Liska <mliska@suse.cz>
27308 PR tree-optimization/81041
27309 * tree-profile.c (gimple_gen_ic_func_profiler):
27310 Create an extra BB in profile-generate
27311 (gimple_gen_time_profiler): Likewise.
27313 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27315 PR tree-optimization/81003
27316 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27317 (update_range_test): Use it instead of force_gimple_operand_gsi.
27319 2017-06-12 Richard Biener <rguenther@suse.de>
27321 PR tree-optimization/81053
27322 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27323 with backedge value not defined in loop. Simplify def stmt
27326 2017-06-11 Tom de Vries <tom@codesourcery.com>
27329 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27331 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27332 nvptx_cannot_force_const_mem.
27334 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27336 * opts.c (finish_options): Move test for flag_split_stack after
27337 it has been initialized.
27339 2017-06-11 Jason Merrill <jason@redhat.com>
27341 * tree.h (id_equal): New.
27342 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27343 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27344 instead of strcmp of IDENTIFIER_POINTER.
27346 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27348 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27349 (mark_all_inlined_calls_cdtor): Fix formating.
27350 (inline_transform): Rescale profile before inlining.
27352 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27354 * cgraph.h (cgraph_edge::clone): Update prototype.
27355 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27356 (cgraph_node::create_clone): Update.
27357 (cgraph_node::create_version_clone): Update.
27358 * tree-inline.c (copy_bb): Update.
27359 (expand_call_inline): Update.
27361 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27363 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27364 factored out from ...
27365 (rs6000_emit_prologue): ... here.
27367 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27369 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27370 factored out from ...
27371 (rs6000_emit_prologue): ... here.
27373 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27375 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27377 (handle_missing_profiles): Fix computation of tp_first_run.
27378 (counts_to_freqs): Do not touch freqs when count is 0.
27380 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27382 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27385 2017-06-10 Tom de Vries <tom@codesourcery.com>
27387 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27388 attributes): Document signal effective target.
27390 2017-06-10 Tom de Vries <tom@codesourcery.com>
27392 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27393 Document effective target stack_size.
27395 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27397 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27398 to the edit_context if they can be auto-applied.
27400 2017-06-9 Ian Lance Taylor <iant@golang.org>
27402 * opts.c (finish_options): If -fsplit-stack, disable implicit
27403 -forder-blocks-and-partition.
27404 * doc/invoke.texi (Optimize Options): Document that when using
27405 -fsplit-stack -forder-blocks-and-partition is not implicitly
27408 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27410 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27411 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27412 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27413 * builtins.def (abort, trap, unreachable): Declare cold.
27414 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27415 * tree-core.h (ECF_COLD): New.
27416 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27417 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27419 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27421 * predict.c (unlikely_executed_stmt_p): Cleanup.
27423 2017-06-09 Richard Biener <rguenther@suse.de>
27425 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27426 model if the ref is always written to.
27428 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27430 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27432 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27434 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27437 2017-06-09 Tom de Vries <tom@codesourcery.com>
27440 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27441 "target cannot support label values" when encountering LABEL_REF.
27443 2017-06-09 Martin Liska <mliska@suse.cz>
27445 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27446 (gimple_gen_ic_func_profiler): Emit direct comparison
27447 of __gcov_indirect_call_callee with NULL.
27448 (gimple_gen_time_profiler): Change probability from
27449 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27451 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27453 * profile.c (edge_gcov_counts): Turn to pointer.
27454 (compute_branch_probabilities, compute_branch_probabilities): Update.
27455 (branch_prob): Do not clear edge_gcov_count.
27456 * profile.h (edge_gcov_counts): Turn to pointer.
27457 (edge_gcov_count): Update.
27459 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27461 * gimple.h (gimple_check_failed): Mark cold.
27463 2017-06-09 Richard Biener <rguenther@suse.de>
27465 PR tree-optimization/66623
27466 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27467 refactor check_reduction into two parts, properly computing
27468 whether we have to check reduction validity for outer loop
27471 2017-06-09 Richard Biener <rguenther@suse.de>
27473 PR tree-optimization/79483
27474 * graphite-scop-detection.c (order): New global.
27475 (get_order): Compute bb to order mapping that satisfies code
27476 generation constraints.
27477 (cmp_pbbs): New helper.
27478 (build_scops): Start domwalk at entry block, sort generated
27481 2017-06-09 Richard Biener <rguenther@suse.de>
27483 PR middle-end/81007
27484 * ipa-polymorphic-call.c
27485 (ipa_polymorphic_call_context::restrict_to_inner_class):
27486 Skip FIELD_DECLs with error_mark_node type.
27487 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27490 2017-06-09 Martin Liska <mliska@suse.cz>
27492 * predict.c (struct branch_predictor): New struct.
27493 (test_prediction_value_range): New test.
27494 (predict_c_tests): New function.
27495 * selftest-run-tests.c (selftest::run_tests): Run the function.
27496 * selftest.h: Declare new tests.
27498 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27501 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27502 gen_add3_insn did not fail.
27503 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27504 r0, construct that number in a temporary reg and add that reg to r0.
27505 If asked to put the result in r0 as well, fail.
27507 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27509 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27510 for early expansion of vec_eqv.
27512 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27514 PR middle-end/81005
27515 * ubsan.c (instrument_null): Avoid pointless code temporary.
27516 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27518 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27522 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27525 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27528 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27530 * predict.c (unlikely_executed_edge_p): Move ahead.
27531 (probably_never_executed_edge_p): Use it.
27533 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27535 PR middle-end/79988
27536 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27537 gimple_call_builtin_p call.
27539 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27541 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27542 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27543 rtl_check_failed_type2, rtl_check_failed_code1,
27544 rtl_check_failed_code2, rtl_check_failed_code_mode,
27545 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27546 rtvec_check_failed_bounds, rtl_check_failed_flag,
27547 _fatal_insn_not_found, _fatal_insn): Likewise.
27548 * tree.h (tree_contains_struct_check_failed,
27549 tree_check_failed, tree_not_check_failed,
27550 tree_class_check_failed, tree_range_check_failed,
27551 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27552 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27553 tree_operand_check_failed, omp_clause_check_failed,
27554 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27557 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27559 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27560 flag_branch_probabilities.
27561 * ipa-inline.c (edge_badness): Likewise.
27562 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27563 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27564 * predict.c (maybe_hot_frequency_p): Likewise.
27565 (probably_never_executed): Likewise.
27566 * sched-ebb.c (schedule_ebbs): Likewise.
27567 * sched-rgn.c (find_single_block_region): Likewise.
27568 * tracer.c (tail_duplicate): Likewise.
27570 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27572 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27573 longer requires x_flag_profile_use.
27575 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27577 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27578 instead of flag_reorder_blocks_and_partition.
27579 * dbxout.c (dbxout_function_end): Likewise.
27580 * dwarf2out.c (gen_subprogram_die): Likewise.
27581 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27582 * hw-doloop.c (reorg_loops): Likewise.
27583 * varasm.c (assemble_start_function,
27584 assemble_end_function): Likewise.
27585 (decide_function_section): Do not check for
27586 flag_reorder_blocks_and_partition.
27588 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27590 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27592 (chkp_get_hard_register_fake_addr_expr): Ditto.
27593 (chkp_build_addr_expr): Add check for hard reg case.
27594 (chkp_parse_array_and_component_ref): Ditto.
27595 (chkp_find_bounds_1): Ditto.
27596 (chkp_process_stmt): Don't generate bounds store for
27599 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27601 * predict.c (maybe_hot_bb_p): Do not check profile status.
27602 (maybe_hot_edge_p): Likewise.
27603 (probably_never_executed): Check for zero counts even if profile
27605 (unlikely_executed_edge_p): New function.
27606 (unlikely_executed_stmt_p): New function.
27607 (unlikely_executed_bb_p): New function.
27608 (set_even_probabilities): Use unlikely predicates.
27609 (combine_predictions_for_bb): Likewise.
27610 (predict_paths_for_bb): Likewise.
27611 (predict_paths_leading_to_edge): Likewise.
27612 (determine_unlikely_bbs): New function.
27613 (estimate_bb_frequencies): Use it.
27614 (compute_function_frequency): Use zero counts even if profile is
27616 * profile-count.h: Fix typo.
27618 2017-08-08 Julia Koval <julia.koval@intel.com>
27620 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27621 _mm512_mask_cvtsepi16_storeu_epi8,
27622 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27623 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27624 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27625 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27626 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27627 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27628 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27629 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27630 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27631 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27632 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27633 __builtin_ia32_pmovuswb256mem_mask,
27634 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27635 __builtin_ia32_pmovwb512mem_mask): New builtins.
27637 2017-08-08 Julia Koval <julia.koval@intel.com>
27639 PR target/73350,80862
27640 * config/i386/subst.md (round): Fix round pattern.
27641 * config/i386/i386.c (ix86_erase_embedded_rounding):
27642 Fix erasing rounding for the fixed pattern.
27644 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27646 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27648 2017-06-08 Martin Liska <mliska@suse.cz>
27650 PR gcov-profile/80911
27651 * gcov.c (block_info::block_info): New constructor.
27653 2017-06-07 Carl Love <cel@us.ibm.com>
27655 * config/rs6000/rs6000-c: The return type of the following
27656 built-in functions was implemented as int not long long. Fix sign
27657 of return value for the unsigned version of vec_mulo and vec_mule.
27658 vector unsigned long long vec_bperm (vector unsigned long long,
27659 vector unsigned char)
27660 vector signed long long vec_mule (vector signed int,
27662 vector unsigned long long vec_mule (vector unsigned int,
27663 vector unsigned int)
27664 vector signed long long vec_mulo (vector signed int,
27666 vector unsigned long long vec_mulo (vector unsigned int,
27667 vector unsigned int)
27668 * doc/extend.texi: Fix the documentation for the built-in
27671 2017-06-07 Carl Love <cel@us.ibm.com>
27674 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27677 2017-06-07 Carl Love <cel@us.ibm.com>
27679 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27680 support, Generate doublehv for signed int/float for BE case only.
27682 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27684 * doc/invoke.texi (mcx16): Rewrite.
27686 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27688 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27689 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27690 *mov<mode>_softfloat, and an anonymous splitter): Use
27691 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27693 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27695 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27697 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27698 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27699 (rs6000_debug_reg_global): Adjust.
27700 (rs6000_init_hard_regno_mode_ok): Adjust.
27701 (rs6000_dbx_register_number): Adjust.
27702 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27703 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27704 Remove SPE_ACC and SPEFSCR.
27705 (REG_ALLOC_ORDER): Ditto.
27706 (FRAME_POINTER_REGNUM): Change to 111.
27707 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27708 (REG_CLASS_NAMES): Ditto.
27709 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27710 (REGISTER_NAMES): Ditto.
27711 (ADDITIONAL_REG_NAMES): Ditto.
27712 (rs6000_reg_names): Ditto.
27713 * config/rs6000/rs6000.md: Renumber some register number
27716 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27718 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27720 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27721 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27723 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27724 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27725 Delete the SPE high registers.
27726 (REG_ALLOC_ORDER): Ditto.
27727 (enum reg_class): Remove SPE_HIGH_REGS.
27728 (REG_CLASS_NAMES): Ditto.
27729 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27730 (REGISTER_NAMES): Ditto.
27731 (rs6000_reg_names): Ditto.
27732 * doc/tm.texi.in: Remove SPE as example.
27733 * doc/tm.texi: Regenerate.
27735 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27737 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27738 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27739 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27740 * config/rs6000/rs6000.md (type): Remove "brinc".
27742 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27744 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27745 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27746 * config/rs6000/linuxspe.h: Delete file.
27747 * config/rs6000/rs6000.md: Don't include spe.md.
27748 * config/rs6000/spe.h: Delete file.
27749 * config/rs6000/spe.md: Delete file.
27750 * config/rs6000/t-rs6000: Remove spe.md.
27752 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27754 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27755 (reg_or_none500mem_operand): Delete.
27756 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27757 instead of reg_or_none500mem_operand.
27759 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27761 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27762 handling of SPE flags.
27763 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27765 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27767 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27769 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27770 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27771 paired_divv2sf3): Similar.
27772 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27773 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27774 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27776 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27777 Rename the paired_* instruction patterns.
27778 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27780 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27781 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27782 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27783 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27784 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27785 PAIRED_VECTOR_MODE.
27786 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27787 (spe_func_has_64bit_regs_p): Delete.
27788 (spe_expand_predicate_builtin): Delete.
27789 (spe_expand_evsel_builtin): Delete.
27790 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27791 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27792 (invalid_e500_subreg): Delete.
27793 (rs6000_legitimize_address): Always force_reg op2 as well, for
27794 paired single memory accesses.
27795 (rs6000_member_type_forces_blk): Delete.
27796 (rs6000_spe_function_arg): Delete.
27797 (rs6000_expand_unop_builtin): Delete SPE handling.
27798 (rs6000_expand_binop_builtin): Ditto.
27799 (spe_expand_stv_builtin): Delete.
27800 (bdesc_2arg_spe): Delete.
27801 (spe_expand_builtin): Delete.
27802 (spe_expand_predicate_builtin): Delete.
27803 (spe_expand_evsel_builtin): Delete.
27804 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27805 (spe_init_builtins): Delete.
27806 (spe_func_has_64bit_regs_p): Delete.
27807 (savres_routine_name): Delete "info" parameter. Adjust callers.
27808 (rs6000_emit_stack_reset): Ditto.
27809 (rs6000_dwarf_register_span): Delete.
27810 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27811 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27812 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27814 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27816 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27817 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27818 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27819 mulv2sf3, divv2sf3): Delete expanders.
27821 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27823 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27825 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27827 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27828 * config/rs6000/rs6000.c: Ditto.
27830 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27832 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27833 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27834 comparison_operator.
27836 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27838 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27839 * config/rs6000/rs6000.opt: Ditto.
27840 * config/rs6000/t-rtems: Ditto.
27842 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27844 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27845 TARGET_E500_SINGLE by 0, simplify.
27846 * config/rs6000/rs6000.c: Ditto.
27847 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27848 (spe_build_register_parallel): Delete.
27849 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27850 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27851 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27852 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27853 (E500_CONVERT): Delete.
27854 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27856 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27858 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
27859 * config/rs6000/dfp.md: Ditto.
27860 (negdd2, *negdd2_fpr): Merge.
27861 (absdd2, *absdd2_fpr): Merge.
27862 (negtd2, *negtd2_fpr): Merge.
27863 (abstd2, *abstd2_fpr): Merge.
27864 * config/rs6000/e500.h: Delete file.
27865 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
27866 TARGET_FPRS by 1 and simplify.
27867 * config/rs6000/rs6000-c.c: Ditto.
27868 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
27869 TARGET_DF_SPE by 0.
27870 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
27872 * config/rs6000/rs6000.md: Ditto.
27873 (floatdidf2, *floatdidf2_fpr): Merge.
27874 (move_from_CR_gt_bit): Delete.
27875 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
27876 (E500_CR_IOR_COMPARE): Delete.
27877 (All patterns that require !TARGET_FPRS): Delete.
27878 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
27880 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27882 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
27884 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27886 * graphds.c (add_edge): Intitialize edge's attached data.
27887 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
27888 pointer parameter. Call pointed function on each edge during
27889 graph traversing. Skip traversing the edge when the function
27891 (graphds_dfs, graphds_scc): Ditto.
27892 (for_each_edge): New parameter. Pass the new parameter to callback
27894 * graphds.h (skip_edge_callback): New function pointer type.
27895 (graphds_dfs, graphds_scc): New function pointer parameter.
27896 (graphds_edge_callback, for_each_edge): New parameter.
27898 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27900 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
27901 out code checking if runtime alias check is possible to below ...
27902 Call the new function.
27903 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
27904 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
27906 2017-06-07 Marek Polacek <polacek@redhat.com>
27909 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
27910 TYPE_OVERFLOW_WRAPS check.
27912 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27914 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
27915 if versioning is required.
27916 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
27917 peeling with the check for versioning.
27919 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27921 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
27922 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
27923 Set true to new parameter if new ssa variable is defined.
27924 (vect_gen_vector_loop_niters): Refactor. Set range information
27925 for the new vector loop bound variable.
27926 (vect_do_peeling): Ditto.
27928 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27930 * tree-affine.c (ssa.h): Include header file.
27931 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
27932 has wrapping overflow behavior.
27934 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27936 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
27938 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27940 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
27941 (tree_to_aff_combination): ... here.
27943 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27945 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
27946 reg_pressure model function.
27947 (ivopts_global_cost_for_size): Delete.
27948 (determine_set_costs, iv_ca_recount_cost): Call new model function
27949 ivopts_estimate_reg_pressure.
27951 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27953 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
27954 expensive than udiv. Remove floating point cases from mod.
27956 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27958 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
27959 Increase idiv cost.
27961 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27963 * config/aarch64/aarch64.md
27964 (copysignsf3): Fix mask generation.
27966 2017-06-07 Jakub Jelinek <jakub@redhat.com>
27968 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
27970 (class dump_manager): Add register_dumps method.
27971 * dumpfile.c: Include langhooks.h.
27972 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
27973 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
27974 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
27975 (dump_manager::dump_register): Start with 512 entries instead of 32.
27976 (dump_manager::register_dumps): New method.
27977 * toplev.c (general_init): Instead of invoking register_dumps
27978 langhook, invoke register_dumps method on the dump manager.
27979 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
27982 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
27984 * doc/md.texi: Clarify the restrictions on a define_insn condition.
27985 Say that # requires an associated define_split to exist, and that
27986 the define_split must be suitable for use after register allocation.
27988 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27990 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
27991 (compute_outgoing_frequencies): Also initialize zero counts.
27992 (find_many_sub_basic_blocks): Do not produce uninitialized profile
27993 around loops; preserve more of profile when nothing changes.
27995 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
27997 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
27999 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28000 * config/arm/arm-cpu-cdata.h: Regenerate.
28001 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28002 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28003 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28004 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28005 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28007 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28009 * config/arm/t-rmprofile: Likewise.
28010 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28012 2017-06-06 David S. Miller <davem@davemloft.net>
28015 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28016 blockage if function uses alloca.
28018 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28020 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28021 New "uid" fields to hold pretty-print IDs of group and ref.
28022 Memory references are now identified as <group_id>:<ref_id>
28023 instead of using [random] addresses.
28024 (dump_mem_details): Simplify, no functional change.
28025 (dump_mem_ref): Simplify and make output more concise.
28026 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28027 (find_or_create_group): Initialize group uid.
28028 (record_ref): Initialize ref uid. Improve debug output.
28029 (prune_group_by_reuse, should_issue_prefetch_p,)
28030 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28031 (mark_nontemporal_store, determine_loop_nest_reuse):
28032 Improve debug output.
28034 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28036 * dbgcnt.def (prefetch): New debug counter.
28037 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28038 (schedule_prefetches): Stop issueing prefetches if debug counter
28041 2017-06-06 Tom de Vries <tom@codesourcery.com>
28043 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28044 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28046 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28048 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28049 Use aarch64_reg_or_zero predicate for operand 4.
28050 (aarch64_compare_and_swap<mode> define_insn_and_split):
28051 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28052 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28054 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28056 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28057 (arm_compute_save_core_reg_mask): This.
28058 (thumb1_compute_save_reg_mask): Rename into ...
28059 (thumb1_compute_save_core_reg_mask): This.
28060 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28061 (arm_compute_frame_layout): Likewise.
28063 2017-06-06 Richard Biener <rguenther@suse.de>
28065 PR tree-optimization/80974
28066 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28067 keep or clear leaders SSA info.
28069 2017-06-06 Tom de Vries <tom@codesourcery.com>
28071 * config/nvptx/nvptx.c (split_mode_p): New function.
28072 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28074 2017-06-06 Tom de Vries <tom@codesourcery.com>
28076 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28078 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28081 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28084 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28086 * shrink-wrap.c (handle_simple_exit): Update profile.
28087 (try_shrink_wrapping): Upate profile.
28089 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28091 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28092 (tree_guess_outgoing_edge_probabilities): New.
28093 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28094 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28096 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28098 * ipa-split.c (split_function): Initialize return bb profile.
28100 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28102 * profile.c (compute_branch_probabilities): Also initialize
28103 EXIT_BLOCK profile.
28105 2017-06-06 Richard Biener <rguenther@suse.de>
28107 PR tree-optimization/80928
28108 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28109 (vect_analyze_loop_operations): Properly guard analysis for
28111 (vect_transform_loop): Likewise.
28112 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28113 (vect_model_induction_cost): Do not cost for pure SLP.
28114 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28115 of induction in inner loop vectorization.
28116 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28117 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28118 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28120 (vect_analyze_slp_cost_1): Cost induction.
28121 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28122 (vect_get_slp_vect_defs): Likewise.
28123 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28124 (vect_transform_stmt): Handle SLP reductions.
28125 * tree-vectorizer.h (vectorizable_induction): Adjust.
28127 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28129 * config/rs6000/rs6000.c (make_resolver_func): Update
28130 init_lowered_empty_function call.
28132 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28134 * doc/invoke.texi: Document the -fprofile-abs-path option.
28135 * common.opt (fprofile-abs-path): New option.
28136 * gcov-io.h (gcov_write_filename): Declare.
28137 * gcov-io.c (gcov_write_filename): New function.
28138 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28139 * profile.c (output_location): Likewise.
28141 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28143 * shring-wrap.c: Revert accidental commit.
28145 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28147 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28149 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28151 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28153 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28154 profile in callgraph edge.
28155 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28156 (apply_scale): Likewise.
28157 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28158 Also scale profile when inlining function with zero profile.
28159 (initialize_cfun): Update exit block profile even when it is zero.
28160 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28161 when profile is read.
28163 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28165 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28166 (CLONE_*): New constants to define the processors we can generate
28167 code for with the target_clone attribute.
28168 (rs6000_clone_map): New array to identify which clone processors
28169 the current program is running on.
28170 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28171 target_clone attribute.
28172 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28173 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28174 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28175 (cpu_expand_builtin): Add support for target_clone attribute.
28176 (rs6000_valid_attribute_p): Allow "default" attribute.
28177 (get_decl_name): New debug function to simplify printing the
28178 current function name in debugging statements.
28179 (rs6000_clone_priority): New functions to support the target_clone
28180 attribute, and be able to generate code to switch between ISA 2.05
28181 through ISA 3.0 (power6 through power9).
28182 (rs6000_compare_version_priority): Likewise.
28183 (rs6000_get_function_versions_dispatcher): Likewise.
28184 (make_resolver_func): Likewise.
28185 (add_condition_to_bb): Likewise.
28186 (dispatch_function_versions): Likewise.
28187 (rs6000_generate_version_dispatcher_body): Likewise.
28188 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28189 (fusion_gpr_load_p): Fix a spacing issue.
28190 * doc/extend.texi (Common Function Attributes): Document that the
28191 PowerPC supports the target_clone attribute.
28193 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28195 * config/arm/arm.h: explain F symbol found in description of ARM
28196 register allocation in its legend.
28198 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28200 * config/mips/frame-header-opt.c: Include profile-count.h.
28201 * config/riscv/riscv.c: Include profile-count.h
28203 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28205 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28207 (sm_set_flag_if_changed): Add bbs field.
28208 (execute_sm_if_changed_flag_set): Pass BBS.
28209 (execute_sm): Update.
28211 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28213 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28216 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28218 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28219 (peephole2): New peephole2 to emit the above.
28220 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28222 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28224 * config/aarch64/aarch64.c (define_peephole2 above
28225 *sub_<shift>_<mode>): New peephole.
28227 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28229 * config/i386/i386.c (make_resolver_func): Update.
28230 * Makefile.in: Add profile-count.h and profile-count.o
28231 * auto-profile.c (afdo_indirect_call): Update to new API.
28232 (afdo_set_bb_count): Update.
28233 (afdo_propagate_edge): Update.
28234 (afdo_propagate_circuit): Update.
28235 (afdo_calculate_branch_prob): Update.
28236 (afdo_annotate_cfg): Update.
28237 * basic-block.h: Include profile-count.h
28238 (struct edge_def): Turn count to profile_count.
28239 (struct basic_block_def): Likewie.
28240 (REG_BR_PROB_BASE): Move to profile-count.h
28241 (RDIV): Move to profile-count.h
28242 * bb-reorder.c (max_entry_count): Turn to profile_count.
28243 (find_traces): Update.
28244 (rotate_loop):Update.
28245 (connect_traces):Update.
28246 (sanitize_hot_paths):Update.
28247 * bt-load.c (migrate_btr_defs): Update.
28248 * cfg.c (RDIV): Remove.
28249 (init_flow): Use alloc_block.
28250 (alloc_block): Uninitialize count.
28251 (unchecked_make_edge): Uninitialize count.
28252 (check_bb_profile): Update.
28253 (dump_edge_info): Update.
28254 (dump_bb_info): Update.
28255 (update_bb_profile_for_threading): Update.
28256 (scale_bbs_frequencies_int): Update.
28257 (scale_bbs_frequencies_gcov_type): Update.
28258 (scale_bbs_frequencies_profile_count): New.
28259 * cfg.h (update_bb_profile_for_threading): Update.
28260 (scale_bbs_frequencies_profile_count): Declare.
28261 * cfgbuild.c (compute_outgoing_frequencies): Update.
28262 (find_many_sub_basic_blocks): Update.
28263 * cfgcleanup.c (try_forward_edges): Update.
28264 (try_crossjump_to_edge): Update.
28265 * cfgexpand.c (expand_gimple_tailcall): Update.
28266 (construct_exit_block): Update.
28267 * cfghooks.c (verify_flow_info): Update.
28268 (dump_bb_for_graph): Update.
28269 (split_edge): Update.
28270 (make_forwarder_block): Update.
28271 (duplicate_block): Update.
28272 (account_profile_record): Update.
28273 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28274 (get_estimated_loop_iterations): Update.
28275 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28276 (single_likely_exit): Update.
28277 * cfgloopmanip.c (scale_loop_profile): Update.
28279 (set_zero_probability): Update.
28280 (lv_adjust_loop_entry_edge): Update.
28281 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28282 (purge_dead_edges): Update.
28283 (rtl_account_profile_record): Update.
28284 * cgraph.c (cgraph_node::create): Uninitialize count.
28285 (symbol_table::create_edge): Uninitialize count.
28286 (cgraph_update_edges_for_call_stmt_node): Update.
28287 (cgraph_edge::dump_edge_flags): Update.
28288 (cgraph_node::dump): Update.
28289 (cgraph_edge::maybe_hot_p): Update.
28290 * cgraph.h: Include profile-count.h
28291 (create_clone), create_edge, create_indirect_edge): Update.
28292 (cgraph_node): Turn count to profile_count.
28293 (cgraph_edge0: Likewise.
28294 (make_speculative, clone): Update.
28295 (create_edge): Update.
28296 (init_lowered_empty_function): Update.
28297 * cgraphclones.c (cgraph_edge::clone): Update.
28298 (duplicate_thunk_for_node): Update.
28299 (cgraph_node::create_clone): Update.
28300 * cgraphunit.c (cgraph_node::analyze): Update.
28301 (cgraph_node::expand_thunk): Update.
28302 * final.c (dump_basic_block_info): Update.
28303 * gimple-streamer-in.c (input_bb): Update.
28304 * gimple-streamer-out.c (output_bb): Update.
28305 * graphite.c (print_global_statistics): Update.
28306 (print_graphite_scop_statistics): Update.
28307 * hsa-brig.c: Include basic-block.h.
28308 * hsa-dump.c: Include basic-block.h.
28309 * hsa-gen.c (T sum_slice): Update.
28310 (convert_switch_statements):Update.
28311 * hsa-regalloc.c: Include basic-block.h.
28312 * ipa-chkp.c (chkp_produce_thunks): Update.
28313 * ipa-cp.c (struct caller_statistics): Update.
28314 (init_caller_stats): Update.
28315 (gather_caller_stats): Update.
28316 (ipcp_cloning_candidate_p): Update.
28317 (good_cloning_opportunity_p): Update.
28318 (get_info_about_necessary_edges): Update.
28319 (dump_profile_updates): Update.
28320 (update_profiling_info): Update.
28321 (update_specialized_profile): Update.
28322 (perhaps_add_new_callers): Update.
28323 (decide_about_value): Update.
28324 (ipa_cp_c_finalize): Update.
28325 * ipa-devirt.c (struct odr_type_warn_count): Update.
28326 (struct decl_warn_count): Update.
28327 (struct final_warning_record): Update.
28328 (possible_polymorphic_call_targets): Update.
28329 (ipa_devirt): Update.
28330 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28331 * ipa-icf.c (sem_function::merge): Update.
28332 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28333 * ipa-inline.c (compute_uninlined_call_time): Update.
28334 (compute_inlined_call_time): Update.
28335 (want_inline_small_function_p): Update.
28336 (want_inline_self_recursive_call_p): Update.
28337 (edge_badness): Update.
28338 (lookup_recursive_calls): Update.
28339 (recursive_inlining): Update.
28340 (inline_small_functions): Update.
28341 (dump_overall_stats): Update.
28342 (dump_inline_stats): Update.
28343 * ipa-profile.c (ipa_profile_generate_summary): Update.
28344 (ipa_propagate_frequency): Update.
28345 (ipa_profile): Update.
28346 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28347 * ipa-utils.c (ipa_merge_profiles): Update.
28348 * loop-doloop.c (doloop_modify): Update.
28349 * loop-unroll.c (report_unroll): Update.
28350 (unroll_loop_runtime_iterations): Update.
28351 * lto-cgraph.c (lto_output_edge): Update.
28352 (lto_output_node): Update.
28353 (input_node): Update.
28354 (input_edge): Update.
28355 (merge_profile_summaries): Update.
28356 * lto-streamer-in.c (input_cfg): Update.
28357 * lto-streamer-out.c (output_cfg): Update.
28358 * mcf.c (create_fixup_graph): Update.
28359 (adjust_cfg_counts): Update.
28360 (sum_edge_counts): Update.
28361 * modulo-sched.c (sms_schedule): Update.
28362 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28363 * predict.c (maybe_hot_count_p): Update.
28364 (probably_never_executed): Update.
28365 (dump_prediction): Update.
28366 (combine_predictions_for_bb): Update.
28367 (propagate_freq): Update.
28368 (handle_missing_profiles): Update.
28369 (counts_to_freqs): Update.
28370 (rebuild_frequencies): Update.
28371 (force_edge_cold): Update.
28372 * predict.h: Include profile-count.h
28373 (maybe_hot_count_p, counts_to_freqs): UPdate.
28374 * print-rtl-function.c: Do not include cfg.h
28375 * print-rtl.c: Include basic-block.h
28376 * profile-count.c: New file.
28377 * profile-count.h: New file.
28378 * profile.c (is_edge_inconsistent): Update.
28379 (correct_negative_edge_counts): Update.
28380 (is_inconsistent): Update.
28381 (set_bb_counts): Update.
28382 (read_profile_edge_counts): Update.
28383 (compute_frequency_overlap): Update.
28384 (compute_branch_probabilities): Update; Initialize and deinitialize
28386 (branch_prob): Update.
28387 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28388 (edge_gcov_count): New.
28389 (bb_gcov_count): New.
28390 * shrink-wrap.c (try_shrink_wrapping): Update.
28391 * tracer.c (better_p): Update.
28392 * trans-mem.c (expand_transaction): Update.
28393 (ipa_tm_insert_irr_call): Update.
28394 (ipa_tm_insert_gettmclone_call): Update.
28395 * tree-call-cdce.c: Update.
28396 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28397 (gimple_duplicate_sese_tail): Update.
28398 (gimple_account_profile_record): Update.
28399 (execute_fixup_cfg): Update.
28400 * tree-inline.c (copy_bb): Update.
28401 (copy_edges_for_bb): Update.
28402 (initialize_cfun): Update.
28403 (freqs_to_counts): Update.
28404 (copy_cfg_body): Update.
28405 (expand_call_inline): Update.
28406 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28407 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28408 (try_unroll_loop_completely): Update.
28409 (try_peel_loop): Update.
28410 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28411 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28412 * tree-ssa-loop-split.c (connect_loops): Update.
28413 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28414 * tree-ssa-reassoc.c (branch_fixup): Update.
28415 * tree-ssa-tail-merge.c (replace_block_by): Update.
28416 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28417 (compute_path_counts): Update.
28418 (update_profile): Update.
28419 (recompute_probabilities): Update.
28420 (update_joiner_offpath_counts): Update.
28421 (estimated_freqs_path): Update.
28422 (freqs_to_counts_path): Update.
28423 (clear_counts_path): Update.
28424 (ssa_fix_duplicate_block_edges): Update.
28425 (duplicate_thread_path): Update.
28426 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28427 (struct switch_conv_info): Update.
28428 * tree-tailcall.c (decrease_profile): Update.
28429 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28430 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28431 * value-prof.c (check_counter): Update.
28432 (gimple_divmod_fixed_value): Update.
28433 (gimple_mod_pow2): Update.
28434 (gimple_mod_subtract): Update.
28435 (gimple_ic_transform): Update.
28436 (gimple_stringop_fixed_value): Update.
28437 * value-prof.h (gimple_ic): Update.
28439 2017-06-02 Carl Love <cel@us.ibm.com>
28441 * config/rs6000/rs6000-c: Add support for built-in functions
28442 vector double vec_doublee (vector signed int);
28443 vector double vec_doublee (vector unsigned int);
28444 vector double vec_doublee (vector float);
28445 vector double vec_doubleh (vector signed int);
28446 vector double vec_doubleh (vector unsigned int);
28447 vector double vec_doubleh (vector float);
28448 vector double vec_doublel (vector signed int);
28449 vector double vec_doublel (vector unsigned int);
28450 vector double vec_doublel (vector float);
28451 vector double vec_doubleo (vector signed int);
28452 vector double vec_doubleo (vector unsigned int);
28453 vector double vec_doubleo (vector float);.
28454 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28455 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28457 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28458 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28459 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28461 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28462 vec_doublel, vec_doubleh.
28463 * doc/extend.texi: Update the built-in documentation file for the
28464 new built-in functions.
28466 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28469 * ipa-inline-analysis.c (free_growth_caches): Set
28470 edge_removal_hook_holder to NULL after removing it.
28472 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28474 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28475 comparision with zero.
28477 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28478 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28479 for early expansion of vec_min and vec_max builtins.
28480 (builtin_function_type): Add min/max unsigned variants to those
28481 identified as having unsigned arguments.
28483 2017-06-02 Olivier Hainque <hainque@adacore.com>
28485 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28487 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28489 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28490 Use VALL_F16 iterator rather than VALL.
28492 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28494 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28495 Emit CBNZ inside loop when doing a strong exchange and comparing
28496 against zero. Generate the CC flags after the loop.
28498 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28500 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28501 (dl_section_ref): New.
28502 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28503 On AIX, append an expression to subtract the size of the
28504 section length to dl_section_ref.
28506 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28508 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28509 for early expansion of vector absolute builtins.
28511 2017-06-02 Richard Biener <rguenther@suse.de>
28513 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28514 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28516 2017-06-02 Richard Biener <rguenther@suse.de>
28518 PR tree-optimization/80948
28519 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28520 stmt order as well.
28522 2017-06-02 Richard Biener <rguenther@suse.de>
28524 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28526 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28527 for inductions as relevant.
28529 2017-06-02 Richard Biener <rguenther@suse.de>
28531 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28532 (vectorizable_induction): ... this. Remove dead code.
28534 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28536 * builtins. (expand_builtin_alloca): Remove second parameter and
28537 infer its value from the first parameter instead.
28538 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28540 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28542 PR rtl-optimization/80903
28543 * loop-doloop.c (add_test): Unshare sequence.
28545 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28547 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28549 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28551 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28553 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28554 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28555 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28556 (xlogue_layout::get_stub_rtx): Make static.
28557 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28558 (xlogue_layout::compute_stub_managed_regs): Rename to...
28559 (xlogue_layout::count_stub_managed_regs): ...this.
28560 (xlogue_layout::is_stub_managed_reg): New function.
28561 (xlogue_layout::m_stub_names): Rename to...
28562 (xlogue_layout::s_stub_names): ...this, make static.
28563 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28564 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28565 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28566 xlogue_layout::s_stub_names): Instantiate statics.
28567 (stub_managed_regs): Remove.
28568 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28569 (disable_call_ms2sysv_xlogues): Rename to...
28570 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28571 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28573 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28574 change after reload_completed.
28575 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28577 (ix86_expand_prologue): Likewise.
28578 (ix86_expand_epilogue): Likewise.
28579 (ix86_expand_split_stack_prologue): Likewise.
28580 (ix86_compute_frame_layout): Remove frame parameter ...
28581 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28582 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28584 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28585 (ix86_frame): Move from here ...
28586 * config/i386/i386.h (ix86_frame): ... to here.
28587 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28588 complete ix86_frame data structure instead. Remove some_ld_name.
28590 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28592 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28593 symbols that hold a DECL_VALUE_EXPR.
28595 2017-06-01 Martin Jambor <mjambor@suse.cz>
28597 PR tree-optimization/80898
28598 * tree-sra.c (process_subtree_disqualification): Removed.
28599 (disqualify_candidate): Do not acll
28600 process_subtree_disqualification.
28601 (subtree_mark_written_and_enqueue): New function.
28602 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28603 RHS has been disqualified and re-queue LHS if necessary. Apart
28604 from that, ignore disqualified RHS.
28606 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28608 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28609 address fetch for z10 or later.
28611 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28613 * config/arc/arc.md (tst_movb): Add guard when splitting.
28615 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28617 * config/arc/arc.c (arc_can_eliminate): Test against
28618 arc_frame_pointer_needed.
28620 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28622 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28623 to prevent store reordering.
28624 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28625 (type): Add block type.
28626 (stack_tie): Define special instruction to be used in
28629 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28631 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28632 constraint. It is not valid for the pattern.
28633 (noncommutative_binary_comparison): Likewise.
28635 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28637 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28640 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28642 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28643 be used by the reg-alloc.
28645 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28647 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28648 reg-alloc when having mul64 or mul32x16 instructions.
28649 (mulsidi3): Likewise.
28650 (umulsidi3): Likewise.
28651 (mulsi32x16): New pattern.
28652 (mulsi64): Likewise.
28653 (mulsidi64): Likewise.
28654 (umulsidi64): Likewise.
28655 (MUL32x16_REG): Define.
28656 (mul64_600): Use MUL32x16_REG.
28657 (mac64_600): Likewise.
28658 (umul64_600): Likewise.
28659 (umac64_600): Likewise.
28661 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28663 * config/arc/arc.md (mulsi3_700): Make it commutative.
28665 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28667 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28669 (*sign_extendsidi2_insn): Likewise for movstosw.
28671 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28673 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28674 the type of the input discriminant value. Convert the
28675 discriminant value of signedness vary.
28677 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28679 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28680 Add to -Wall section.
28682 2017-06-01 Richard Biener <rguenther@suse.de>
28684 PR middle-end/66313
28685 * fold-const.c (fold_plusminus_mult_expr): If the factored
28686 factor may be zero use a wrapping type for the inner operation.
28687 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28688 and handle moved defs.
28689 (process_assignment): Properly guard the unary op case. Return a
28690 tri-state indicating that moving the stmt before the call may allow
28691 to continue. Pass through to_move.
28692 (find_tail_calls): Handle moving unrelated defs before
28695 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28698 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28699 splitter result in the canonical way.
28701 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28703 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28704 also for 32bit target. Update insn attributes.
28705 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28707 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28709 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28710 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28711 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28712 (_mm_maskz_min_ss): New intrinsics.
28714 2017-05-31 Martin Liska <mliska@suse.cz>
28716 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28717 Change comment style to one we normally use.
28718 (vectorizable_reduction): Likewise.
28719 (vectorizable_induction): Likewise.
28720 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28721 (vectorizable_call): Likewise.
28722 (vectorizable_simd_clone_call): Likewise.
28723 (vectorizable_conversion): Likewise.
28724 (vectorizable_assignment): Likewise.
28725 (vectorizable_shift): Likewise.
28726 (vectorizable_operation): Likewise.
28727 (vectorizable_store): Likewise.
28728 (vectorizable_load): Likewise.
28729 * tree-vectorizer.h: Likewise.
28731 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28733 * passes.c (emergency_dump_function): New.
28734 * tree-pass.h (emergency_dump_function): Declare.
28735 * plugin.c (plugins_internal_error_function): Remove.
28736 * plugin.h (plugins_internal_error_function): Remove declaration.
28737 * toplev.c (internal_error_function): New static function. Use it...
28738 (general_init): ...here.
28740 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28742 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28743 (arc_rtx_costs): Add costs for new patterns.
28744 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28745 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28747 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28749 * tree-ssa-strlen.c (get_next_strinfo): New function.
28750 (get_stridx_plus_constant): Use it.
28751 (zero_length_string): Likewise.
28752 (adjust_related_strinfos): Likewise.
28753 (adjust_last_stmt): Likewise.
28755 2017-05-31 Richard Biener <rguenther@suse.de>
28758 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28759 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28761 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28763 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28764 loop_vinfo argument and use of dependence distance vectors.
28765 Check instead whether the two references differ only in their
28766 initial value and assume that they have the same alignment if the
28767 difference is a multiple of the vector alignment.
28768 (vect_analyze_data_refs_alignment): Update call accordingly.
28770 2017-05-31 Martin Liska <mliska@suse.cz>
28773 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28775 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28777 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28778 (create_intersect_range_checks): Move from ...
28779 * tree-data-ref.c (create_intersect_range_checks_index)
28780 (create_intersect_range_checks): ... to here.
28781 (create_runtime_alias_checks): New function factored from ...
28782 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28783 here. Call above function.
28784 * tree-data-ref.h (create_runtime_alias_checks): New function.
28786 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28788 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28789 segment length for dr_b and compute it in wide_int.
28791 2017-05-31 Richard Biener <rguenther@suse.de>
28793 PR tree-optimization/80906
28794 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28795 and pass through iv_map.
28796 (copy_bb_and_scalar_dependences): Adjust.
28797 (translate_pending_phi_nodes): Likewise.
28798 (copy_loop_close_phi_args): Handle code-generating IVs instead
28801 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28803 * diagnostic-color.c (color_dict): Add "type-diff".
28804 (parse_gcc_colors): Update comment.
28805 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28806 -fdiagnostics-show-template-tree and -fno-elide-type.
28807 (GCC_COLORS): Add type-diff to example.
28809 (-fdiagnostics-show-template-tree): New.
28810 (-fno-elide-type): New.
28811 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28812 the pp_format_decoder callback. Call any m_format_postprocessor's
28814 (pretty_printer::pretty_printer): Initialize
28815 m_format_postprocessor.
28816 (pretty_printer::~pretty_printer): Delete any
28817 m_format_postprocessor.
28818 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28819 (class format_postprocessor): New class.
28820 (struct pretty_printer::format_decoder): Document the new parameters.
28821 (struct pretty_printer::m_format_postprocessor): New field.
28822 * tree-diagnostic.c (default_tree_printer): Update for new
28823 bool and const char ** params.
28824 * tree-diagnostic.h (default_tree_printer): Likewise.
28826 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28828 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28829 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28830 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28831 rs6000_gen_cell_microcode code.
28832 (rs6000_final_prescan_insn): Delete.
28833 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28834 "warn-cell-microcode" entries.
28835 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28836 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28837 throughout. Change cc_reg_not_micro_cr0_operand to
28838 cc_reg_not_cr0_operand throughout.
28839 (*extendhi<mode>2_noload): Delete.
28840 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28841 (mwarn-cell-microcode): Delete.
28842 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28843 -mgen-cell-microcode and -mwarn-cell-microcode.
28845 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28848 * config/i386/constraints.md (Yd): New constraint.
28850 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28851 and (?Yd, r) alternatives. Update insn attributes.
28852 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28853 and (?*Yd, r) alternatives. Update insn attributes.
28854 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28856 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28858 * gimplify.c (gimplify_modify_expr): Don't create a
28859 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
28862 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
28864 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
28866 2017-05-30 Richard Biener <rguenther@suse.de>
28868 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
28869 and reduc_def fields.
28870 (STMT_VINFO_REDUC_TYPE): New define.
28871 (STMT_VINFO_REDUC_DEF): Likewise.
28872 (vect_force_simple_reduction): Adjust prototype.
28873 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
28874 (vect_is_simple_reduction): Remove check_reduction argument.
28875 (vect_force_simple_reduction): Adjust and set
28876 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28877 (vectorizable_reduction): Do not re-do reduction analysis
28878 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28879 * tree-parloops.c (gather_scalar_reductions): Adjust.
28881 2017-05-30 Richard Biener <rguenther@suse.de>
28883 PR middle-end/80901
28884 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
28887 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
28889 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
28890 Introduce unknown_misalignment parameter and remove vf.
28891 (vect_peeling_hash_get_lowest_cost):
28892 Pass unknown_misalignment parameter.
28893 (vect_enhance_data_refs_alignment):
28894 Fix unsupportable data ref treatment.
28896 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28898 * tree-vect-data-refs.c (vect_get_data_access_cost):
28899 Workaround for SLP handling.
28900 (vect_enhance_data_refs_alignment):
28901 Compute costs for doing no peeling at all, compare to the best
28902 peeling costs so far and avoid peeling if cheaper.
28904 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28906 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
28907 Return peeling info and set costs to zero for unlimited cost
28909 (vect_enhance_data_refs_alignment): Also inspect all datarefs
28910 with unknown misalignment. Compute and costs for unknown
28911 misalignment, compare them to the costs for known misalignment
28912 and choose the cheapest for peeling.
28914 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28916 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
28917 (vect_get_peeling_costs_all_drs): Create function.
28918 (vect_peeling_hash_get_lowest_cost):
28919 Use vect_get_peeling_costs_all_drs.
28920 (vect_peeling_supportable): Create function.
28921 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
28923 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28925 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
28926 DR_HAS_NEGATIVE_STEP.
28927 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
28928 (vect_enhance_data_refs_alignment): Use.
28929 (vect_duplicate_ssa_name_ptr_info): Use.
28930 * tree-vectorizer.h (dr_misalignment): Use.
28931 (known_alignment_for_access_p): Use.
28933 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28936 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
28938 (has_section_name): New function.
28940 2017-05-30 Martin Liska <mliska@suse.cz>
28943 * auto-profile.c (get_function_decl_from_block): Fix
28946 2017-05-30 Richard Biener <rguenther@suse.de>
28948 PR middle-end/80876
28949 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
28951 2017-05-30 Martin Liska <mliska@suse.cz>
28953 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
28954 * dumpfile.h (struct dump_file_info): Remove ctors.
28956 2017-05-30 Martin Liska <mliska@suse.cz>
28958 * predict.def: Fix GNU coding style.
28960 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28962 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28963 Mark 'to' argument with ATTRIBUTE_UNUSED.
28965 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28967 * config/xtensa/xtensa.c (xtensa_emit_call): Use
28968 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
28969 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
28972 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
28974 * doc/install.texi (Options specification): Restore entry of
28975 --enable-sjlj-exceptions.
28977 2017-05-27 Michael Eager <eager@eagercon.com>
28980 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
28982 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
28984 * config/microblaze/microblaze.h
28985 (FIXED_REGISTERS): Update in macro.
28986 (CALL_USED_REGISTERS): Update in macro.
28988 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
28990 * doc/install.texi: Add links to macOS binary distributions.
28992 2017-05-27 Jakub Jelinek <jakub@redhat.com>
28996 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28998 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29000 2017-05-26 Martin Liska <mliska@suse.cz>
29002 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29004 2017-05-26 Martin Liska <mliska@suse.cz>
29006 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29007 always leading ';; '.
29008 (dump_bb_info): Likewise.
29009 (brief_dump_cfg): Likewise.
29010 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29011 * dumpfile.c: Remove usage of TDF_VERBOSE.
29012 * dumpfile.h (enum dump_kind): Likewise.
29013 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29014 * print-tree.c (debug_verbose): Remove.
29015 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29016 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29017 * tree-diagnostic.c (default_tree_printer): Replace
29018 TDF_DIAGNOSTIC with TDF_SLIM.
29020 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29022 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29023 in parameter loop, rather than loop_vinfo.
29024 (create_intersect_range_checks): Ditto.
29025 (vect_create_cond_for_alias_checks): Update call to above functions.
29027 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29029 PR tree-optimization/80815
29030 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29031 for merging runtime alias checks. Handle negative DR_STEPs.
29033 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29035 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29037 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29038 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29039 out code pruning runtime alias checks.
29040 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29041 factored out from above.
29042 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29044 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29046 (prune_runtime_alias_test_list): New decalaration.
29048 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29050 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29051 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29052 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29053 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29054 (operator==, comp_dr_with_seg_len_pair): Ditto.
29055 (vect_prune_runtime_alias_test_list): Ditto.
29057 2017-05-26 Martin Liska <mliska@suse.cz>
29060 * params.def: Bound partial-inlining-entry-probability param.
29062 2017-05-26 Marek Polacek <polacek@redhat.com>
29065 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29068 2017-05-26 Richard Biener <rguenther@suse.de>
29070 PR tree-optimization/80842
29071 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29074 2017-05-26 Richard Biener <rguenther@suse.de>
29076 PR tree-optimization/80844
29077 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29079 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29081 * doc/md.texi (Machine Constraints): Update x86 family
29082 machine constraints section to match 'config/i386/constraints.md'.
29084 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29086 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29088 2017-05-25 Nathan Sidwell <nathan@acm.org>
29090 * doc/invoke.texi (--enable-languages): Update documentation.
29092 2017-05-25 Martin Liska <mliska@suse.cz>
29094 * dumpfile.c: Add TDF_FOLDING.
29095 * dumpfile.h (enum dump_kind): Likewise.
29096 * genmatch.c (dt_simplify::gen_1): Use it.
29098 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29100 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29102 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29104 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29105 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29107 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29109 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29110 * match.pd (X == C): Rewrite it here.
29111 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29112 with_certain_nonzero_bits2): New predicates.
29113 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29115 2017-05-24 Nathan Sidwell <nathan@acm.org>
29117 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29120 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29123 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29125 * config/powerpcspe: New port. Files are copied from the rs6000
29126 port, with "rs6000" in filenames replaced by "powerpcspe".
29128 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29130 PR rtl-optimization/80754
29131 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29133 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29135 * config/sparc/sparc.md (length): Return the correct value for -mflat
29136 sibcalls to match output_sibcall.
29138 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29142 * config/rs6000/rs6000.c (struct machine_function): Add new field
29144 (rs6000_get_separate_components): Init that field, use it.
29145 (rs6000_components_for_bb): Use the field.
29147 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29149 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29151 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29153 PR middle-end/80823
29154 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29156 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29159 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29160 address against address_operand predicate.
29161 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29163 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29165 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29166 parameters passed indirectly.
29168 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29170 * config/i386/i386.md (*movdi_internal): Remove SSE4
29171 alternative 18 (?r, *v). Update insn attributes.
29172 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29173 Update insn attributes.
29174 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29175 Update insn attributes.
29176 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29177 alternative 1 (r, v). Remove isa attribute.
29178 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29179 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29180 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29182 2017-05-23 Tom de Vries <tom@codesourcery.com>
29184 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29187 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29189 * cgraphunit.c (symbol_table::process_new_functions): Update.
29190 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29191 (inline_generate_summary): Rename to ...
29192 (ipa_fn_summary_generate): ... this one.
29193 (inline_read_summary): Rename to ...
29194 (ipa_fn_summary_read): ... this one.
29195 (inline_write_summary): Rename to ...
29196 (ipa_fn_summary_write): ... this one.
29197 (inline_free_summary): Rename to ...
29198 (ipa_free_fn_summary): ... this one.
29199 (pass_data_local_fn_summary, pass_local_fn_summary,
29200 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29201 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29202 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29203 make_pass_ipa_fn_summary): New.
29204 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29205 inline_write_summary, inline_free_summary): Remove.
29206 (ipa_free_fn_summary) : New.
29207 * ipa-inline.c (ipa_inline): Update.
29208 (pass_ipa_inline): Do not generate summaries.
29209 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29211 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29212 and add pass_ipa_fn_summary.
29213 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29215 (make_pass_inline_parameters): Remove.
29217 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29219 * omp-low.c (struct omp_context): Remove "default_kind" member.
29222 * omp-offload.c (execute_oacc_device_lower): Remove the
29223 parallelism dimensions function attributes for unparallelized
29224 OpenACC kernels constructs.
29226 2017-05-23 Martin Liska <mliska@suse.cz>
29228 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29230 (cgraph_edge::make_speculative): Likewise.
29231 (cgraph_edge::resolve_speculation): Likewise.
29232 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29233 (cgraph_node::dump): Likewise.
29234 * cgraph.h: Likewise.
29235 * cgraphunit.c (analyze_functions): Likewise.
29236 (symbol_table::compile): Likewise.
29237 * ipa-cp.c (print_all_lattices): Likewise.
29238 (determine_versionability): Likewise.
29239 (initialize_node_lattices): Likewise.
29240 (ipcp_verify_propagated_values): Likewise.
29241 (estimate_local_effects): Likewise.
29242 (update_profiling_info): Likewise.
29243 (create_specialized_node): Likewise.
29244 (perhaps_add_new_callers): Likewise.
29245 (decide_about_value): Likewise.
29246 (decide_whether_version_node): Likewise.
29247 (identify_dead_nodes): Likewise.
29248 (ipcp_store_bits_results): Likewise.
29249 * ipa-devirt.c (dump_targets): Likewise.
29250 (ipa_devirt): Likewise.
29251 * ipa-icf.c (sem_item::dump): Likewise.
29252 (sem_function::equals): Likewise.
29253 (sem_variable::equals): Likewise.
29254 (sem_item_optimizer::read_section): Likewise.
29255 (sem_item_optimizer::execute): Likewise.
29256 (congruence_class::dump): Likewise.
29257 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29258 (dump_inline_summary): Likewise.
29259 (estimate_node_size_and_time): Likewise.
29260 (inline_analyze_function): Likewise.
29261 * ipa-inline-transform.c (inline_call): Likewise.
29262 * ipa-inline.c (report_inline_failed_reason): Likewise.
29263 (want_early_inline_function_p): Likewise.
29264 (edge_badness): Likewise.
29265 (update_edge_key): Likewise.
29266 (inline_small_functions): Likewise.
29267 * ipa-profile.c (ipa_profile): Likewise.
29268 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29269 (ipa_make_edge_direct_to_target): Likewise.
29270 (remove_described_reference): Likewise.
29271 (ipa_impossible_devirt_target): Likewise.
29272 (propagate_controlled_uses): Likewise.
29273 (ipa_print_node_params): Likewise.
29274 (ipcp_transform_function): Likewise.
29275 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29276 (propagate_pure_const): Likewise.
29277 * ipa-reference.c (generate_summary): Likewise.
29278 (read_write_all_from_decl): Likewise.
29279 (propagate): Likewise.
29280 (ipa_reference_read_optimization_summary): Likewise.
29281 * ipa-utils.c (ipa_merge_profiles): Likewise.
29282 * ipa.c (walk_polymorphic_call_targets): Likewise.
29283 (symbol_table::remove_unreachable_nodes): Likewise.
29284 (ipa_single_use): Likewise.
29285 * passes.c (execute_todo): Likewise.
29286 * predict.c (drop_profile): Likewise.
29287 * symtab.c (symtab_node::get_dump_name): New function.
29288 (symtab_node::dump_name): Likewise.
29289 (symtab_node::dump_asm_name): Likewise.
29290 (symtab_node::dump_references): Likewise.
29291 (symtab_node::dump_referring): Likewise.
29292 (symtab_node::dump_base): Likewise.
29293 (symtab_node::debug_symtab): Likewise.
29294 * tree-sra.c (convert_callers_for_node): Likewise.
29295 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29296 * value-prof.c (init_node_map): Likewise.
29298 2017-05-23 Martin Liska <mliska@suse.cz>
29300 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29301 and symtab_node::debug_symtab to symbol_table::debug.
29302 * cgraphunit.c (analyze_functions): Use the renamed function.
29303 (symbol_table::compile): Likewise.
29304 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29305 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29306 * passes.c (execute_todo): Likewise.
29307 * symtab.c (symbol_table::dump): New function.
29308 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29310 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29312 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29313 that nonconst implies exec.
29315 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29317 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29318 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29319 (inline_edge_summary_vec): Turn into ...
29320 (ipa_call_summaries): ... this one.
29321 (redirect_to_unreachable, edge_set_predicate,
29322 evaluate_properties_for_edge, inline_summary_alloc,
29323 reset_ipa_call_summary, reset_inline_summary,
29324 inline_summary_t::duplicate): Update.
29325 (inline_edge_duplication_hook): Turn to ...
29326 (ipa_call_summary_t::duplicate): ... this one.
29327 (inline_edge_removal_hook): Turn to ...
29328 (ipa_call_summary_t::remove): ... this one.
29329 (dump_inline_edge_summary): Turn to ...
29330 (dump_ipa_call_summary): ... this one.
29331 (estimate_function_body_sizes): Update.
29332 (inline_update_callee_summaries): Update.
29333 (remap_edge_change_prob): Update.
29334 (remap_edge_summaries): Update.
29335 (inline_merge_summary): Update.
29336 (do_estimate_edge_time): Update.
29337 (inline_generate_summary): Update.
29338 (inline_read_section): Update.
29339 (inline_read_summary): Update.
29340 (inline_free_summary): Update.
29341 * ipa-inline.c (can_inline_edge_p): Update.
29342 (compute_inlined_call_time): Update.
29343 (want_inline_small_function_p): Update.
29344 (edge_badness): Update.
29345 (early_inliner): Update.
29346 * ipa-inline.h (inline_edge_summary): Turn to ...
29347 (ipa_call_summary): ... this one.
29348 (ipa_call_summary_t): New class.
29349 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29350 (ipa_call_summaries): New.
29351 (inline_edge_summary): Remove.
29352 (estimate_edge_growth): Update.
29353 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29354 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29355 * ipa-split.c (execute_split_functions): Update.
29356 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29358 2017-05-23 Tom de Vries <tom@codesourcery.com>
29360 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29361 attributes): Document rdrand effective target.
29363 2017-05-23 Tom de Vries <tom@codesourcery.com>
29365 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29366 attributes): Sort alphabetically.
29368 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29370 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29372 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29375 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29376 V2DF/V2DI splat into two separate patterns, one that handles
29377 registers, and the other that only handles memory. Drop support
29378 for splatting from a GPR on ISA 2.07 and then splitting the
29379 splat into direct move and splat.
29380 (vsx_splat_<mode>_reg): Likewise.
29381 (vsx_splat_<mode>_mem): Likewise.
29383 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29385 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29387 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29389 PR middle-end/80809
29390 * omp-low.c (finish_taskreg_remap): New function.
29391 (finish_taskreg_scan): If unit size of ctx->record_type
29392 is non-constant, unshare the size expression and replace
29393 decls in it with possible outer var refs.
29395 PR middle-end/80809
29396 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29397 GOVD_SHARED rather than GOVD_PRIVATE with it.
29398 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29399 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29401 PR middle-end/80853
29402 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29403 as last argument to build_outer_var_ref for pointer bases of array
29404 section reductions.
29406 2017-05-19 Martin Sebor <msebor@redhat.com>
29408 * print-tree.c (print_node): Print DECL_READ_P flag.
29410 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29412 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29413 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29414 * cgraph.c: Likewise.
29415 * cgraphunit.c: Likewise.
29416 * gengtype.c: Likewise.
29417 * ipa-cp.c: Likewise.
29418 * ipa-devirt.c: Likewise.
29419 * ipa-icf.c: Likewise.
29420 * ipa-predicate.c: Likewise.
29421 * ipa-profile.c: Likewise.
29422 * ipa-prop.c: Likewise.
29423 * ipa-split.c: Likewise.
29425 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29426 edge_predicate_pool, dump_inline_hints,
29427 inline_summary::account_size_time, redirect_to_unreachable,
29428 edge_set_predicate, set_hint_predicate,
29429 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29430 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29431 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29432 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29433 ipa_call_summary_t::remove, initialize_growth_caches,
29434 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29435 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29436 mark_modified, unmodified_parm_1, unmodified_parm,
29437 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29438 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29439 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29440 will_be_nonconstant_predicate, record_modified_bb_info,
29441 get_minimal_bb, record_modified, param_change_prob,
29442 phi_result_unknown_predicate, predicate_for_phi_result,
29443 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29444 estimate_function_body_sizes, compute_inline_parameters,
29445 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29446 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29447 inline_update_callee_summaries, remap_edge_change_prob,
29448 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29449 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29450 inline_analyze_function, inline_summary_t::insert,
29451 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29452 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29453 inline_free_summary): Move to ipa-fnsummary.h
29454 (predicate_t): Remove.
29455 * ipa-fnsummary.c: New file.
29456 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29457 (enum inline_hints_vals, inline_hints, agg_position_info,
29458 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29459 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29460 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29461 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29462 inline_read_summary, inline_write_summary, inline_free_summary,
29463 inline_analyze_function, initialize_inline_failed,
29464 inline_merge_summary, inline_update_overall_summary,
29465 compute_inline_parameters): Move to ipa-fnsummary.h
29466 * ipa-fnsummary.h: New file.
29467 * ipa-inline-transform.h: Include ipa-inline.h.
29468 * ipa-inline.c: LIkewise.
29470 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29472 * ipa-inline.c (edge_badness): Use inlined_time instead of
29473 inline_summaries->get.
29475 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29477 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29479 2017-05-22 Nathan Sidwell <nathan@acm.org>
29481 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29482 (fdump-lang): Document 'raw' option.
29483 * dumpfile.h (TDI_tu): Delete.
29484 * dumpfile.c (dump_files): Remove translation-unit.
29485 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29487 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29489 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29490 command option from $(AWK) call.
29491 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29492 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29493 [FORMAT]: Remove handling of variable.
29494 * config/avr/t-multilib: Regenerate.
29496 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29498 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29500 (dump_inline_summary): Do not print self_time.
29501 (estimate_function_body_sizes): Do not set self_time.
29502 (compute_inline_parameters): Likewise.
29503 (inline_read_section, inline_write_summary): Do not stream self_time.
29504 * ipa-inline.h (inline_summary): Drop self_time.
29506 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29508 * ipa-inline-analysis.c (account_size_time): Rename to ...
29509 (inline_summary::account_size_time): ... this one.
29510 (reset_ipa_call_summary): Turn to ...
29511 (ipa_call_summary::reset): ... this one.
29512 (reset_inline_summary): Turn to ...
29513 (inline_summary::reset): ... this one.
29514 (inline_summary_t::remove): Update.
29515 (inline_summary_t::duplicate): Update.
29516 (ipa_call_summary_t::remove): Update.
29517 (dump_inline_summary): Update.
29518 (estimate_function_body_sizes): Update.
29519 (compute_inline_parameters): Update.
29520 (estimate_node_size_and_time): Update.
29521 (inline_merge_summary): Update.
29522 (inline_update_overall_summary): Update.
29523 (inline_read_section): Update.
29524 (inline_write_summary): Update.
29525 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29526 add account_size_time and reset member functions.
29527 (ipa_call_summary): Add reset function.
29528 * ipa-predicate.h (predicate::operator &): Constify.
29530 2017-05-22 Richard Biener <rguenther@suse.de>
29532 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29534 2017-05-19 Jason Merrill <jason@redhat.com>
29536 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29538 2017-05-19 Marek Polacek <polacek@redhat.com>
29541 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29542 TYPE_OVERFLOW_WRAPS checks.
29544 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29546 * tree-core.h (enum omp_clause_default_kind): Add
29547 "OMP_CLAUSE_DEFAULT_PRESENT".
29548 * tree-pretty-print.c (dump_omp_clause): Handle it.
29549 * gimplify.c (enum gimplify_omp_var_data): Add
29550 "GOVD_MAP_FORCE_PRESENT".
29551 (gimplify_adjust_omp_clauses_1): Map it to
29552 "GOMP_MAP_FORCE_PRESENT".
29553 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29555 * gimplify.c (oacc_default_clause): Clarify.
29557 2017-05-19 Nathan Sidwell <nathan@acm.org>
29559 LANG_HOOK_REGISTER_DUMPS
29560 * toplev.c (general_init): Call register dump lang hook.
29561 * doc/invoke.texi: Document -fdump-lang option family.
29562 * dumpfile.c (dump_files): Remove class dump here.
29563 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29564 * dumpfile.h (tree_dump_index): Remove TDI_class.
29565 * langhooks-def.h (lhd_register_dumps): Declare.
29566 (LANG_HOOKS_REGISTER_DUMPS): Define.
29567 (LANG_HOOKS_INITIALIZER): Add it.
29568 * langhooks.c (lhd_register_dumps): Define.
29569 * langhooks.h (struct lang_hooks): Add register_dumps.
29571 2017-05-19 Nathan Sidwell <nathan@acm.org>
29573 * context.h (context::set_passes): New.
29574 * context.c (context::context): Do not create pass manager.
29575 * toplev.c (general_init): Create pass manager here.
29577 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29579 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29580 use this splitter if two add or or instructions would also work for
29581 the constant we want to generate.
29583 2017-05-19 Richard Biener <rguenther@suse.de>
29586 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29587 predicate evaluation.
29589 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29591 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29593 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29596 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29598 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29599 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29600 (inline_edge_summary_vec): Turn into ...
29601 (ipa_call_summaries): ... this one.
29602 (redirect_to_unreachable, edge_set_predicate,
29603 evaluate_properties_for_edge, inline_summary_alloc,
29604 reset_ipa_call_summary, reset_inline_summary,
29605 inline_summary_t::duplicate): Update.
29606 (inline_edge_duplication_hook): Turn to ...
29607 (ipa_call_summary_t::duplicate): ... this one.
29608 (inline_edge_removal_hook): Turn to ...
29609 (ipa_call_summary_t::remove): ... this one.
29610 (dump_inline_edge_summary): Turn to ...
29611 (dump_ipa_call_summary): ... this one.
29612 (estimate_function_body_sizes): Update.
29613 (inline_update_callee_summaries): Update.
29614 (remap_edge_change_prob): Update.
29615 (remap_edge_summaries): Update.
29616 (inline_merge_summary): Update.
29617 (do_estimate_edge_time): Update.
29618 (inline_generate_summary): Update.
29619 (inline_read_section): Update.
29620 (inline_read_summary): Update.
29621 (inline_free_summary): Update.
29622 * ipa-inline.c (can_inline_edge_p): Update.
29623 (compute_inlined_call_time): Update.
29624 (want_inline_small_function_p): Update.
29625 (edge_badness): Update.
29626 (early_inliner): Update.
29627 * ipa-inline.h (inline_edge_summary): Turn to ...
29628 (ipa_call_summary): ... this one.
29629 (ipa_call_summary_t): New class.
29630 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29631 (ipa_call_summaries): New.
29632 (inline_edge_summary): Remove.
29633 (estimate_edge_growth): Update.
29634 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29635 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29636 * ipa-split.c (execute_split_functions): Update.
29637 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29639 2017-05-19 Richard Biener <rguenther@suse.de>
29641 PR middle-end/80764
29642 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29644 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29646 * config/rs6000/rs6000.c (struct machine_function): Add field
29647 fpr_is_wrapped_separately.
29648 (rs6000_get_separate_components): Use 64 components. Handle the
29649 new FPR components.
29650 (rs6000_components_for_bb): Handle the FPR components.
29651 (rs6000_emit_prologue_components): Handle the FPR components.
29652 (rs6000_emit_epilogue_components): Handle the FPR components.
29653 (rs6000_set_handled_components): Handle the FPR components.
29654 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29655 that are already separately shrink-wrapped.
29656 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29657 that are already separately shrink-wrapped.
29659 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29662 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29665 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29666 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29667 cases where the register allocator uses a move and an offsettable
29668 memory operation to/from a FPR register on ISA 2.06/2.07.
29669 (define_peephole2 for Altivec d-form store): Likewise.
29671 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29674 * config/i386/mmx.md (*mov<mode>_internal): Enable
29675 alternatives 11, 12, 13 and 14 also for 32bit targets.
29676 Remove alternatives 15, 16, 17 and 18.
29677 * config/i386/sse.md (vec_concatv2di): Change
29678 alternative (!x, *y) to (x, ?!*Yn).
29680 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29682 * dumpfile.h (enum dump_kind): Remove stray comma.
29684 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29686 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29687 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29688 predicate::num_conditions
29689 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29690 (CHANGED): turn into predicate::changed.
29691 (agg_position_info): Move to ipa-predicate.h
29692 (add_condition, predicate::add_clause, predicate::operator &=,
29693 predicate::or_with, predicate::evaluate, predicate::probability,
29694 dump_condition, dump_clause, predicate::dump,
29695 predicate::remap_after_duplication, predicate::remap_after_inlining,
29696 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29697 (evaluate_conditions_for_known_args): Update.
29698 (set_cond_stmt_execution_predicate): Update.
29699 * ipa-inline.h: Include ipa-predicate.h
29700 (condition, inline_param_summary, conditions, agg_position_info,
29701 predicate): Move to ipa-predicate.h
29702 * ipa-predicate.c: New file.
29703 * ipa-predicate.h: New file.
29705 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29707 * final.c (leaf_function_p): Check we are not in a sequence.
29709 2017-05-18 Martin Liska <mliska@suse.cz>
29711 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29712 * dumpfile.c (dump_register): Use new enum dump_kind.
29713 (get_dump_file_name): Likewise.
29714 (dump_enable_all): Likewise.
29715 (dump_switch_p_1): Likewise.
29716 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29717 * dumpfile.h (enum dump_kind): New enum type.
29718 (struct dump_file_info): Create constructor and
29719 format fields and comments.
29720 * passes.c (pass_manager::register_one_dump_file):
29722 * statistics.c (statistics_early_init): Likewise.
29723 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29724 TDF_TREE with TDF_SLIM.
29725 (gather_memory_references_ref): Likewise.
29727 2017-05-18 Martin Liska <mliska@suse.cz>
29729 * vec.h (struct vnull): Use it.
29731 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29733 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29734 (true_predicate, false_predicate, true_predicate_p,
29735 false_predicate_p): Remove.
29736 (single_cond_predicate, not_inlined_predicate): Turn to member function
29738 (add_condition): Update.
29739 (add_clause): Turn to...
29740 (predicate::add_clause): ... this one; update; allow passing NULL
29742 (and_predicates): Turn to ...
29743 (predicate::operator &=): ... this one.
29744 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29745 (or_predicates): Turn to ...
29746 (predicate::or_with): ... this one.
29747 (evaluate_predicate): Turn to ...
29748 (predicate::evaluate): ... this one.
29749 (predicate_probability): Turn to ...
29750 (predicate::probability): ... this one.
29751 (dump_condition): Update.
29752 (dump_predicate): Turn to ...
29753 (predicate::dump): ... this one.
29754 (account_size_time): Update.
29755 (edge_set_predicate): Update.
29756 (set_hint_predicate): UPdate.
29757 (evaluate_conditions_for_known_args): Update.
29758 (evaluate_properties_for_edge): Update.
29759 (remap_predicate_after_duplication): Turn to...
29760 (predicate::remap_after_duplication): ... this one.
29761 (remap_hint_predicate_after_duplication): Update.
29762 (inline_summary_t::duplicate): UPdate.
29763 (dump_inline_edge_summary): Update.
29764 (dump_inline_summary): Update.
29765 (set_cond_stmt_execution_predicate): Update.
29766 (set_switch_stmt_execution_predicate): Update.
29767 (compute_bb_predicates): Update.
29768 (will_be_nonconstant_expr_predicate): Update.
29769 (will_be_nonconstant_predicate): Update.
29770 (phi_result_unknown_predicate): Update.
29771 (predicate_for_phi_result): Update.
29772 (array_index_predicate): Update.
29773 (estimate_function_body_sizes): Update.
29774 (estimate_node_size_and_time): Update.
29775 (estimate_ipcp_clone_size_and_time): Update.
29776 (remap_predicate): Rename to ...
29777 (predicate::remap_after_inlining): ... this one.
29778 (remap_hint_predicate): Update.
29779 (inline_merge_summary): Update.
29780 (inline_update_overall_summary): Update.
29781 (estimate_size_after_inlining): Update.
29782 (read_predicate): Rename to ...
29783 (predicate::stream_in): ... this one.
29784 (read_inline_edge_summary): Update.
29785 (write_predicate): Rename to ...
29786 (predicate::stream_out): ... this one.
29787 (write_inline_edge_summary): Update.
29788 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29789 (clause_t): Turn to uint32_t
29790 (predicate): Turn to class; implement constructor and operators
29792 (size_time_entry): Update.
29793 (inline_summary): Update.
29794 (inline_edge_summary): Update.
29796 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29798 * fold-const.c (fold_binary_loc): Move transformation...
29799 * match.pd (C - X CMP X): ... here.
29801 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29803 * config/sparc/sparc.c (sparc_option_override): Set function
29804 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29805 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29807 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29809 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29811 2017-05-18 Marek Polacek <polacek@redhat.com>
29814 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29815 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29816 gimple_assign_load_p.
29818 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29820 PR middle-end/80692
29821 * real.c (do_compare): Give decimal_do_compare preference over
29822 comparing just the signs.
29824 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29826 * doc/md.texi (Canonicalization of Instructions): Describe the
29827 canonical form of instructions that inherently set a condition
29830 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29832 PR middle-end/80775
29833 * tree-cfg.c: Move deletion of unreachable case statements to after
29834 the merging of consecutive case labels.
29836 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29838 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29839 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29840 restoring of callee-saved registers.
29842 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29844 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29845 * config/visium/visium.c (single_set_and_flags): Likewise.
29846 * config/visium/visium.md (Substitutions): Likewise.
29848 2017-05-17 Martin Liska <mliska@suse.cz>
29850 * cfg.c: Introduce dump_flags_t type and
29851 use it instead of int type.
29853 * cfghooks.c: Likewise.
29854 * cfghooks.h (struct cfg_hooks): Likewise.
29855 * cfgrtl.c: Likewise.
29856 * cfgrtl.h: Likewise.
29857 * cgraph.c (cgraph_node::get_body): Likewise.
29858 * coretypes.h: Likewise.
29859 * domwalk.c: Likewise.
29860 * domwalk.h: Likewise.
29861 * dumpfile.c (struct dump_option_value_info): Likewise.
29862 (dump_enable_all): Likewise.
29863 (dump_switch_p_1): Likewise.
29864 (opt_info_switch_p): Likewise.
29865 * dumpfile.h (enum tree_dump_index): Likewise.
29866 (struct dump_file_info): Likewise.
29867 * genemit.c: Likewise.
29868 * generic-match-head.c: Likewise.
29869 * gengtype.c (open_base_files): Likewise.
29870 * gimple-pretty-print.c: Likewise.
29871 * gimple-pretty-print.h: Likewise.
29872 * graph.c (print_graph_cfg): Likewise.
29873 * graphite-scop-detection.c (dot_all_sese): Likewise.
29874 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
29875 * loop-unroll.c (report_unroll): Likewise.
29876 * passes.c (pass_manager::register_one_dump_file): Likewise.
29877 * print-tree.c: Likewise.
29878 * statistics.c: Likewise.
29879 * tree-cfg.c: Likewise.
29880 * tree-cfg.h: Likewise.
29881 * tree-dfa.c: Likewise.
29882 * tree-dfa.h: Likewise.
29883 * tree-dump.c (dump_function): Likewise.
29884 * tree-dump.h (struct dump_info): Likewise.
29885 * tree-pretty-print.c: Likewise.
29886 * tree-pretty-print.h: Likewise.
29887 * tree-ssa-live.c: Likewise.
29888 * tree-ssa-live.h: Likewise.
29889 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
29890 * tree-vect-loop.c: Likewise.
29891 * tree-vect-slp.c: Likewise.
29893 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
29894 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29896 PR tree-optimization/80457
29897 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
29898 of all arguments to a statement as scalar_to_vec operations.
29899 (vectorizable_call): Adjust call to vect_model_simple_cost for
29901 (vectorizable_conversion): Likewise.
29902 (vectorizable_assignment): Likewise.
29903 (vectorizable_shift): Likewise.
29904 (vectorizable_operation): Likewise.
29905 (vectorizable_comparison): Likewise.
29906 (vect_is_simple_cond): Record the def types for operands.
29907 (vectorizable_condition): Likewise, call vect_model_simple_cost.
29908 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
29909 for statement argument count.
29911 2017-05-16 Carl Love <cel@us.ibm.com>
29913 * config/rs6000/rs6000-c: Add support for built-in functions
29914 vector unsigned long long vec_bperm (vector unsigned long long,
29915 vector unsigned char)
29916 vector signed long long vec_mule (vector signed int,
29918 vector unsigned long long vec_mule (vector unsigned int,
29919 vector unsigned int)
29920 vector signed long long vec_mulo (vector signed int,
29922 vector unsigned long long vec_mulo (vector unsigned int,
29923 vector unsigned int)
29924 vector signed char vec_sldw (vector signed char,
29925 vector signed char,
29927 vector unsigned char vec_sldw (vector unsigned char,
29928 vector unsigned char,
29930 vector signed short vec_sldw (vector signed short,
29931 vector signed short,
29933 vector unsigned short vec_sldw (vector unsigned short,
29934 vector unsigned short,
29936 vector signed int vec_sldw (vector signed int,
29939 vector unsigned int vec_sldw (vector unsigned int,
29940 vector unsigned int,
29942 vector signed long long vec_sldw (vector signed long long,
29943 vector signed long long,
29945 vector unsigned long long vec_sldw (vector unsigned long long,
29946 vector unsigned long long,
29948 * config/rs6000/rs6000-c: Add support for built-in functions
29949 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
29950 * config/rs6000/altivec.h: Add defintion for vec_sldw.
29951 * doc/extend.texi: Update the built-in documentation for the
29952 new built-in functions.
29954 2017-05-16 Marek Polacek <polacek@redhat.com>
29958 * tree.c (save_expr): Don't fold the expression.
29960 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
29962 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
29963 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
29964 and (?*y,m). Update insn attributes.
29966 2017-05-16 Martin Liska <mliska@suse.cz>
29968 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
29969 flags argument of print_gimple_stmt, print_gimple_expr,
29970 print_generic_stmt and print_generic_expr.
29971 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
29972 * coretypes.h: Likewise.
29973 * except.c (dump_eh_tree): Likewise.
29974 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
29975 * gimple-pretty-print.h: Likewise.
29976 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
29977 (backprop::push_to_worklist): Likewise.
29978 (backprop::pop_from_worklist): Likewise.
29979 (backprop::process_use): Likewise.
29980 (backprop::intersect_uses): Likewise.
29981 (note_replacement): Likewise.
29982 * gimple-ssa-store-merging.c
29983 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
29984 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
29985 (pass_store_merging::execute): Likewise.
29986 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
29987 (ssa_base_cand_dump_callback): Likewise.
29988 (dump_incr_vec): Likewise.
29989 (replace_refs): Likewise.
29990 (replace_mult_candidate): Likewise.
29991 (create_add_on_incoming_edge): Likewise.
29992 (create_phi_basis): Likewise.
29993 (insert_initializers): Likewise.
29994 (all_phi_incrs_profitable): Likewise.
29995 (introduce_cast_before_cand): Likewise.
29996 (replace_one_candidate): Likewise.
29997 * gimplify.c (gimplify_expr): Likewise.
29998 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
29999 (set_rename): Likewise.
30000 (rename_uses): Likewise.
30001 (copy_loop_phi_nodes): Likewise.
30002 (add_close_phis_to_merge_points): Likewise.
30003 (copy_loop_close_phi_args): Likewise.
30004 (copy_cond_phi_args): Likewise.
30005 (graphite_copy_stmts_from_block): Likewise.
30006 (translate_pending_phi_nodes): Likewise.
30007 * graphite-poly.c (print_pdr): Likewise.
30008 (dump_gbb_cases): Likewise.
30009 (dump_gbb_conditions): Likewise.
30010 (print_scop_params): Likewise.
30011 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30012 (build_cross_bb_scalars_use): Likewise.
30013 (gather_bbs::before_dom_children): Likewise.
30014 * hsa-dump.c (dump_hsa_immed): Likewise.
30015 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30016 (get_replacement_map): Likewise.
30017 * ipa-inline-analysis.c (dump_condition): Likewise.
30018 (estimate_function_body_sizes): Likewise.
30019 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30020 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30021 * ipa-prop.c (ipa_dump_param): Likewise.
30022 (ipa_print_node_jump_functions_for_edge): Likewise.
30023 (ipa_modify_call_arguments): Likewise.
30024 (ipa_modify_expr): Likewise.
30025 (ipa_dump_param_adjustments): Likewise.
30026 (ipa_dump_agg_replacement_values): Likewise.
30027 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30028 * ipa-pure-const.c (check_stmt): Likewise.
30029 (pass_nothrow::execute): Likewise.
30030 * ipa-split.c (execute_split_functions): Likewise.
30031 * omp-offload.c (dump_oacc_loop_part): Likewise.
30032 (dump_oacc_loop): Likewise.
30033 * trans-mem.c (tm_log_emit): Likewise.
30034 (tm_memopt_accumulate_memops): Likewise.
30035 (dump_tm_memopt_set): Likewise.
30036 (dump_tm_memopt_transform): Likewise.
30037 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30038 (print_loop): Likewise.
30039 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30040 (chkp_gather_checks_info): Likewise.
30041 (chkp_get_check_result): Likewise.
30042 (chkp_remove_check_if_pass): Likewise.
30043 (chkp_use_outer_bounds_if_possible): Likewise.
30044 (chkp_reduce_bounds_lifetime): Likewise.
30045 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30046 (chkp_mark_completed_bounds): Likewise.
30047 (chkp_register_incomplete_bounds): Likewise.
30048 (chkp_mark_invalid_bounds): Likewise.
30049 (chkp_maybe_copy_and_register_bounds): Likewise.
30050 (chkp_build_returned_bound): Likewise.
30051 (chkp_get_bound_for_parm): Likewise.
30052 (chkp_build_bndldx): Likewise.
30053 (chkp_get_bounds_by_definition): Likewise.
30054 (chkp_generate_extern_var_bounds): Likewise.
30055 (chkp_get_bounds_for_decl_addr): Likewise.
30056 * tree-chrec.c (chrec_apply): Likewise.
30057 * tree-data-ref.c (dump_data_reference): Likewise.
30058 (dump_subscript): Likewise.
30059 (dump_data_dependence_relation): Likewise.
30060 (analyze_overlapping_iterations): Likewise.
30061 * tree-inline.c (expand_call_inline): Likewise.
30062 (tree_function_versioning): Likewise.
30063 * tree-into-ssa.c (dump_defs_stack): Likewise.
30064 (dump_currdefs): Likewise.
30065 (dump_names_replaced_by): Likewise.
30066 (dump_update_ssa): Likewise.
30067 (update_ssa): Likewise.
30068 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30069 * tree-parloops.c (build_new_reduction): Likewise.
30070 (try_create_reduction_list): Likewise.
30071 (ref_conflicts_with_region): Likewise.
30072 (oacc_entry_exit_ok_1): Likewise.
30073 (oacc_entry_exit_single_gang): Likewise.
30074 * tree-pretty-print.h: Likewise.
30075 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30076 (get_scalar_evolution): Likewise.
30077 (add_to_evolution): Likewise.
30078 (get_loop_exit_condition): Likewise.
30079 (analyze_evolution_in_loop): Likewise.
30080 (analyze_initial_condition): Likewise.
30081 (analyze_scalar_evolution): Likewise.
30082 (instantiate_scev): Likewise.
30083 (number_of_latch_executions): Likewise.
30084 (gather_chrec_stats): Likewise.
30085 (final_value_replacement_loop): Likewise.
30086 (scev_const_prop): Likewise.
30087 * tree-sra.c (dump_access): Likewise.
30088 (disqualify_candidate): Likewise.
30089 (create_access): Likewise.
30090 (reject): Likewise.
30091 (maybe_add_sra_candidate): Likewise.
30092 (create_access_replacement): Likewise.
30093 (analyze_access_subtree): Likewise.
30094 (analyze_all_variable_accesses): Likewise.
30095 (sra_modify_assign): Likewise.
30096 (initialize_constant_pool_replacements): Likewise.
30097 (find_param_candidates): Likewise.
30098 (decide_one_param_reduction): Likewise.
30099 (replace_removed_params_ssa_names): Likewise.
30100 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30101 * tree-ssa-copy.c (dump_copy_of): Likewise.
30102 (copy_prop_visit_cond_stmt): Likewise.
30103 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30104 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30105 (record_equivalences_from_stmt): Likewise.
30106 * tree-ssa-dse.c (compute_trims): Likewise.
30107 (delete_dead_call): Likewise.
30108 (delete_dead_assignment): Likewise.
30109 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30110 (forward_propagate_into_cond): Likewise.
30111 (pass_forwprop::execute): Likewise.
30112 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30113 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30115 (move_computations_worker): Likewise.
30116 (execute_sm): Likewise.
30117 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30118 (remove_exits_and_undefined_stmts): Likewise.
30119 (remove_redundant_iv_tests): Likewise.
30120 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30121 (adjust_iv_update_pos): Likewise.
30122 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30123 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30124 (value_replacement): Likewise.
30125 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30126 * tree-ssa-pre.c (print_pre_expr): Likewise.
30127 (get_representative_for): Likewise.
30128 (create_expression_by_pieces): Likewise.
30129 (insert_into_preds_of_block): Likewise.
30130 (eliminate_insert): Likewise.
30131 (eliminate_dom_walker::before_dom_children): Likewise.
30132 (eliminate): Likewise.
30133 (remove_dead_inserted_code): Likewise.
30134 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30135 * tree-ssa-reassoc.c (get_rank): Likewise.
30136 (eliminate_duplicate_pair): Likewise.
30137 (eliminate_plus_minus_pair): Likewise.
30138 (eliminate_not_pairs): Likewise.
30139 (undistribute_ops_list): Likewise.
30140 (eliminate_redundant_comparison): Likewise.
30141 (update_range_test): Likewise.
30142 (optimize_range_tests_var_bound): Likewise.
30143 (optimize_vec_cond_expr): Likewise.
30144 (rewrite_expr_tree): Likewise.
30145 (rewrite_expr_tree_parallel): Likewise.
30146 (linearize_expr): Likewise.
30147 (break_up_subtract): Likewise.
30148 (linearize_expr_tree): Likewise.
30149 (attempt_builtin_powi): Likewise.
30150 (attempt_builtin_copysign): Likewise.
30151 (transform_stmt_to_copy): Likewise.
30152 (transform_stmt_to_multiply): Likewise.
30153 (dump_ops_vector): Likewise.
30154 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30155 (print_scc): Likewise.
30156 (set_ssa_val_to): Likewise.
30157 (visit_reference_op_store): Likewise.
30158 (visit_use): Likewise.
30159 (sccvn_dom_walker::before_dom_children): Likewise.
30160 (run_scc_vn): Likewise.
30161 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30163 (expr_hash_elt::print): Likewise.
30164 (const_and_copies::pop_to_marker): Likewise.
30165 (const_and_copies::record_const_or_copy_raw): Likewise.
30166 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30167 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30168 (dump_predicates): Likewise.
30169 (find_uninit_use): Likewise.
30170 (warn_uninitialized_phi): Likewise.
30171 (pass_late_warn_uninitialized::execute): Likewise.
30172 * tree-ssa.c (verify_vssa): Likewise.
30173 (verify_ssa): Likewise.
30174 (maybe_optimize_var): Likewise.
30175 * tree-vrp.c (dump_value_range): Likewise.
30176 (dump_all_value_ranges): Likewise.
30177 (dump_asserts_for): Likewise.
30178 (register_edge_assert_for_2): Likewise.
30179 (vrp_visit_cond_stmt): Likewise.
30180 (vrp_visit_switch_stmt): Likewise.
30181 (vrp_visit_stmt): Likewise.
30182 (vrp_visit_phi_node): Likewise.
30183 (simplify_cond_using_ranges_1): Likewise.
30184 (fold_predicate_in): Likewise.
30185 (evrp_dom_walker::before_dom_children): Likewise.
30186 (evrp_dom_walker::push_value_range): Likewise.
30187 (evrp_dom_walker::pop_value_range): Likewise.
30188 (execute_early_vrp): Likewise.
30190 2017-05-16 Richard Biener <rguenther@suse.de>
30192 * dwarf2out.c (loc_list_from_tree_1): Do not create
30193 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30195 2017-05-16 Richard Biener <rguenther@suse.de>
30197 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30199 (note_variable_value_in_expr): If we resolved the decl ref
30200 do not push to the stack.
30202 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30204 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30205 operations in fast-math mode.
30206 (vaddq_f16): Likewise.
30207 (vmul_f16): Likewise.
30208 (vmulq_f16): Likewise.
30209 (vsub_f16): Likewise.
30210 (vsubq_f16): Likewise.
30211 * config/arm/neon.md (add<mode>3): New.
30213 (fma:<VH:mode>3): New. Also remove outdated comment.
30216 2017-05-16 Martin Liska <mliska@suse.cz>
30220 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30221 (odr_types_equivalent_p): Likewise.
30223 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30225 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30227 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30230 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30231 non-interunit SSE move alternatives with '?'.
30232 (zero-extendsidi peephole2): New peephole to skip intermediate
30233 general register in SSE zero-extend sequence.
30235 2017-05-15 Jeff Law <law@redhat.com>
30237 * reorg.c (relax_delay_slots): Create a new variable to hold
30238 the temporary target rather than clobbering TARGET_LABEL.
30240 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30241 missing argument to extract_bit_field call.
30242 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30244 2017-05-15 Martin Liska <mliska@suse.cz>
30247 * gcc.c (process_command): Do not allow empty argument of -o option.
30249 2017-05-15 Renlin Li <renlin.li@arm.com>
30251 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30252 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30253 * config/aarch64/constraints.md (Usf): Add long call check.
30254 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30255 (call_value): Likewise.
30256 (sibcall): Likewise.
30257 (sibcall_value): Likewise.
30259 (call_value_insn): New.
30260 (sibcall_insn): Update rtx pattern.
30261 (sibcall_value_insn): Likewise.
30262 (call_internal): Remove.
30263 (call_value_internal): Likewise.
30264 (sibcall_internal): Likewise.
30265 (sibcall_value_internal): Likewise.
30266 (call_reg): Likewise.
30267 (call_symbol): Likewise.
30268 (call_value_reg): Likewise.
30269 (call_value_symbol): Likewise.
30271 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30274 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30276 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30278 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30279 compatible with CCGOCmode and with CCZmode.
30281 2017-05-14 Martin Sebor <msebor@redhat.com>
30283 PR middle-end/77671
30284 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30285 (gimple_fold_builtin_snprintf): Same.
30286 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30287 (gimple_fold_builtin_snprintf): Same.
30288 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30289 of character types.
30290 (is_call_safe): New function.
30291 (try_substitute_return_value): Call it.
30292 (try_simplify_call): New function.
30293 (pass_sprintf_length::handle_gimple_call): Call it.
30295 2017-05-14 Martin Sebor <msebor@redhat.com>
30297 PR middle-end/80669
30298 * builtins.c (expand_builtin_stpncpy): Simplify.
30300 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30302 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30303 * config/i386/i386.h
30304 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30305 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30306 (struct machine_function): Add new members call_ms2sysv,
30307 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30308 (struct machine_frame_state): New fields sp_realigned and
30309 sp_realigned_offset.
30310 * config/i386/i386.c
30311 (enum xlogue_stub): New enum.
30312 (enum xlogue_stub_sets): New enum.
30313 (class xlogue_layout): New class.
30314 (struct ix86_frame): New fields stack_realign_allocate_offset,
30315 stack_realign_offset and outlined_save_offset. Modify comments to
30316 detail stack layout when using out-of-line stubs.
30317 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30318 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30319 -mcall-ms2sysv-xlogues.
30320 (stub_managed_regs): New static variable.
30321 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30322 registers managed by out-of-line stub.
30323 (disable_call_ms2sysv_xlogues): New function.
30324 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30325 m->call_ms2sysv when appropriate and compute frame layout for
30327 (sp_valid_at, fp_valid_at): New inline functions.
30328 (choose_basereg): New function.
30329 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30331 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30332 Use align parameter of choose_baseaddr to generated aligned SSE movs
30334 (pro_epilogue_adjust_stack): Modify to track
30335 machine_frame_state::sp_realigned.
30336 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30337 (ix86_nsaved_sseregs): Likewise.
30338 (ix86_emit_save_regs): Likewise.
30339 (ix86_emit_save_regs_using_mov): Likewise.
30340 (ix86_emit_save_sse_regs_using_mov): Likewise.
30341 (get_scratch_register_on_entry): Likewise.
30342 (gen_frame_set): New function.
30343 (gen_frame_load): Likewise.
30344 (gen_frame_store): Likewise.
30345 (emit_outlined_ms2sysv_save): Likewise.
30346 (emit_outlined_ms2sysv_restore): Likewise.
30347 (ix86_expand_prologue): Modify stack re-alignment code and call
30348 emit_outlined_ms2sysv_save when appropriate.
30349 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30350 parameter rtx_insn *insn, which allows the function to be used to only
30351 generate the notes.
30352 (ix86_expand_epilogue): Modify validity checks of frame and stack
30353 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30354 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30355 * config/i386/predicates.md
30356 (save_multiple): New predicate.
30357 (restore_multiple): Likewise.
30358 * config/i386/sse.md
30359 (save_multiple<mode>): New pattern.
30360 (save_multiple_realign<mode>): Likewise.
30361 (restore_multiple<mode>): Likewise.
30362 (restore_multiple_and_return<mode>): Likewise.
30363 (restore_multiple_leave_return<mode>): Likewise.
30364 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30366 2017-05-14 Julia Koval <julia.koval@intel.com>
30368 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30369 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30370 (__builtin_ia32_xsetbv): New builtins.
30371 * config/i386/i386.c (ix86_expand_special_args_builtin):
30373 (ix86_expand_builtin): Special expand for new intrinsics.
30374 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30375 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30376 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30378 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30380 * cfganal.c (inverted_post_order_compute): Change argument type
30382 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30383 * df-core.c (rest_of_handle_df_initialize): Adjust.
30384 (rest_of_handle_df_finish): Likewise.
30385 (df_analyze_1): Likewise.
30386 (df_analyze): Likewise.
30387 (loop_inverted_post_order_compute): Change argument to be a vec *.
30388 (df_analyze_loop): Adjust.
30389 (df_get_n_blocks): Likewise.
30390 (df_get_postorder): Likewise.
30391 * df.h (struct df_d): Change field to be a vec.
30392 * lcm.c (compute_laterin): Adjust.
30393 (compute_available): Likewise.
30394 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30395 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30396 * tree-ssa-pre.c (compute_antic): Likewise.
30398 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30400 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30401 (depth_first_search::depth_first_search): Change structure init
30402 function to this constructor.
30403 (depth_first_search::add_bb): Rename function to this member.
30404 (depth_first_search::execute): Likewise.
30405 (flow_dfs_compute_reverse_finish): Adjust.
30407 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30409 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30410 (longest_simple_path): Likewise.
30411 * shrink-wrap.c (spread_components): Likewise.
30412 (disqualify_problematic_components): Likewise.
30413 (emit_common_heads_for_components): Likewise.
30414 (emit_common_tails_for_components): Likewise.
30415 (insert_prologue_epilogue_for_components): Likewise.
30417 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30419 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30422 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30424 * df-core.c (df_set_blocks): Start using auto_bitmap.
30425 (df_compact_blocks): Likewise.
30426 * df-problems.c (df_rd_confluence_n): Likewise.
30427 * df-scan.c (df_insn_rescan_all): Likewise.
30428 (df_process_deferred_rescans): Likewise.
30429 (df_update_entry_block_defs): Likewise.
30430 (df_update_exit_block_uses): Likewise.
30431 (df_entry_block_bitmap_verify): Likewise.
30432 (df_exit_block_bitmap_verify): Likewise.
30433 (df_scan_verify): Likewise.
30434 * lra-constraints.c (lra_constraints): Likewise.
30435 (undo_optional_reloads): Likewise.
30436 (lra_undo_inheritance): Likewise.
30437 * lra-remat.c (calculate_gen_cands): Likewise.
30438 (do_remat): Likewise.
30439 * lra-spills.c (assign_spill_hard_regs): Likewise.
30440 (spill_pseudos): Likewise.
30441 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30442 (bitmap_set_subtract_values): Likewise.
30444 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30446 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30447 management with auto_bitmap.
30448 (fix_inter_tick): Likewise.
30449 (fix_recovery_deps): Likewise.
30450 * ira.c (add_store_equivs): Likewise.
30451 (find_moveable_pseudos): Likewise.
30452 (split_live_ranges_for_shrink_wrap): Likewise.
30453 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30454 (rtx_reuse_manager::seen_def_p): Likewise.
30455 (rtx_reuse_manager::set_seen_def): Likewise.
30456 * print-rtl.h (class rtx_reuse_manager): Likewise.
30458 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30460 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30462 (migrate_btr_def): Likewise.
30463 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30464 * df-core.c (loop_post_order_compute): Likewise.
30465 (loop_inverted_post_order_compute): Likewise.
30466 * hsa-common.h: Likewise.
30467 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30468 * init-regs.c (initialize_uninitialized_regs): Likewise.
30469 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30470 (inline_small_functions): Likewise.
30471 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30472 * ira.c (combine_and_move_insns): Likewise.
30473 (build_insn_chain): Likewise.
30474 * loop-invariant.c (find_invariants): Likewise.
30475 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30476 * predict.c (tree_predict_by_opcode): Likewise.
30477 (predict_paths_leading_to): Likewise.
30478 (predict_paths_leading_to_edge): Likewise.
30479 (estimate_loops_at_level): Likewise.
30480 (estimate_loops): Likewise.
30481 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30482 (spread_components): Likewise.
30483 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30484 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30485 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30486 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30487 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30488 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30489 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30490 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30491 (mark_threaded_blocks): Likewise.
30492 (thread_through_all_blocks): Likewise.
30493 * tree-ssa.c (verify_ssa): Likewise.
30494 (execute_update_addresses_taken): Likewise.
30495 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30497 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30499 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30501 (post_order_compute): Likewise.
30502 (inverted_post_order_compute): Likewise.
30503 (pre_and_rev_post_order_compute_fn): Likewise.
30505 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30507 * genrecog.c (int_set::int_set): Explicitly construct our
30508 auto_vec base class.
30509 * vec.h (auto_vec::auto_vec): New constructor.
30511 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30513 * bitmap.h (class auto_bitmap): New constructor taking
30514 bitmap_obstack * argument.
30516 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30518 * bitmap.h (class auto_bitmap): Change type of m_bits to
30519 bitmap_head, and adjust ctor / dtor and member operators.
30521 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30523 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30524 when returned register mode doesn't match original mode.
30526 2017-05-12 Jeff Law <law@redhat.com>
30527 Jakub Jelinek <jakub@redhat.com>
30529 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30530 we look for cc setter after the compare-elim changes.
30531 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30532 within the vector to match what compare-elim now expects.
30533 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30534 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30536 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30537 after the compare-elim changes.
30538 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30539 the vector to match what compare-elim now expects.
30540 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30541 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30542 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30543 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30544 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30546 * config/visium/visium.c (single_set_and_flags): Fix where
30547 we look for cc setter after the compare-elim changes.
30548 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30549 with the vector to match what compare-elim now expects.
30550 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30551 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30552 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30553 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30554 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30555 (neg<mode>2_insn_set_overflow): Likewise.
30557 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30559 PR middle-end/79794
30560 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30561 maybe_expand_insn call, set ops[0].target. If still set after call,
30562 set alt_rtl. Add extra arg to recursive calls.
30563 (extract_bit_field): Add alt_rtl argument. Pass to
30565 * expmed.h (extract_bit_field): Fix prototype.
30566 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30567 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30568 to extract_bit_field_calls.
30569 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30570 Pass alt_rtl to extract_bit_field calls.
30571 * calls.c (store_unaligned_arguments_into_psuedos)
30572 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30573 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30575 * optabs.h (struct expand_operand): Add target bitfield.
30577 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30579 * compare-elim.c (try_eliminate_compare): Canonicalize
30580 operation with embedded compare to
30581 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30582 (set (reg) (operation)].
30584 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30586 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30589 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30590 cost of adding a carry flag for ADC instruction.
30591 [case MINUS]: Ignore the cost of subtracting a carry flag
30592 for SBB instruction.
30594 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30596 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30598 * config/rs6000/bmiintrin.h: New file.
30599 * config/rs6000/bmi2intrin.h: New file.
30600 * config/rs6000/x86intrin.h: New file.
30602 2017-05-12 Jeff Law <law@redhat.com>
30604 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30607 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30609 PR middle-end/80707
30610 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30612 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30614 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30615 early expansion of vector divide builtins.
30616 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30617 builtins identified as having unsigned arguments.
30619 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30621 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30622 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30623 expansion of vector logical operations (and, andc, or, xor,
30626 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30628 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30629 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30631 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30633 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30634 early GIMPLE expansion of vector multiplies.
30636 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30638 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30639 TARGET_HAVE_MOVT conditional.
30640 (movt splitter): Likewise.
30642 2017-05-12 Richard Biener <rguenther@suse.de>
30644 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30645 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30646 Fold all stmts not inplace.
30648 2017-05-12 Richard Biener <rguenther@suse.de>
30650 PR tree-optimization/80713
30651 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30652 inserted_exprs bit for not removed stmts.
30654 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30656 PR middle-end/69921
30657 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30658 parallelized" attribute for parallelized OpenACC kernels.
30659 * omp-offload.c (execute_oacc_device_lower): Use it.
30661 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30662 Set "oacc kernels" attribute.
30663 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30664 parameter. Adjust all users.
30665 (oacc_fn_attrib_kernels_p): Remove function.
30666 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30667 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30668 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30669 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30670 assert "oacc kernels" attribute is set.
30672 2017-05-11 Carl Love <cel@us.ibm.com>
30674 * config/rs6000/rs6000-c: Add support for built-in functions
30675 vector unsigned char vec_popcnt (vector signed char)
30676 vector unsigned char vec_popcnt (vector unsigned char)
30677 vector unsigned short vec_popcnt (vector signed short)
30678 vector unsigned short vec_popcnt (vector unsigned short)
30679 vector unsigned int vec_popcnt (vector signed int)
30680 vector unsigned int vec_popcnt (vector unsigned int)
30681 vector unsigned long long vec_popcnt (vector signed long long)
30682 vector unsigned long long vec_popcnt (vector unsigned long long)
30683 vector signed long long vec_slo (vector signed long long,
30684 vector signed char)
30685 vector signed long long vec_slo (vector signed long long,
30686 vector unsigned char)
30687 vector unsigned long long vec_slo (vector unsigned long long,
30688 vector signed char)
30689 vector unsigned long long vec_slo (vector unsigned long long,
30690 vector unsigned char)
30691 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30692 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30693 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30694 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30695 * doc/extend.texi: Update the built-in documentation file for the
30696 new built-in functions.
30698 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30700 * attribs.h (sorted_attr_string): Move machine independent
30701 functions for target clone support from the i386 port to common
30702 code. Rename ix86_function_versions to common_function_versions.
30703 Rename make_name to make_unique_name.
30704 (common_function_versions): Likewise.
30705 (make_unique_name): Likewise.
30706 (make_dispatcher_decl): Likewise.
30707 (is_function_default_version): Likewise.
30708 * attribs.c (attr_strcmp): Likewise.
30709 (sorted_attr_string): Likewise.
30710 (common_function_versions): Likewise.
30711 (make_unique_name): Likewise.
30712 (make_dispatcher_decl): Likewise.
30713 (is_function_default_version): Likewise.
30714 * config/i386/i386.c (attr_strcmp): Likewise.
30715 (sorted_attr_string): Likewise.
30716 (ix86_function_versions): Likewise.
30717 (make_name): Likewise.
30718 (make_dispatcher_decl): Likewise.
30719 (is_function_default_version): Likewise.
30720 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30722 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30725 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30726 Account for direct move costs for vec_construct of integer
30729 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30732 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30733 (UNSPEC_STX_ATOMIC): Ditto.
30734 (loaddi_via_sse): New insn.
30735 (storedi_via_sse): Ditto.
30736 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30737 Update corresponding peephole2 patterns.
30738 (atomic_storedi_fpu): Ditto.
30740 2017-05-11 Julia Koval <julia.koval@intel.com>
30742 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30743 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30745 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30746 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30747 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30749 2017-05-11 Nathan Sidwell <nathan@acm.org>
30751 * graphite-poly.c: Include dumpfile.h.
30753 * dumpfle.h (dump_function): Declare here ...
30754 * tree-dump.h (dump_function): ... not here.
30755 * dumpfile.c: #include tree-cfg.h.
30756 (dump_function): Move here from ...
30757 * tree-dump.c (dump_function): ... here.
30758 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30759 * graphite-poly.c: Don't include tree-dump.h.
30760 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30761 * print-tree.c: Likewise.
30762 * stor-layout.c: Likewise.
30763 * tree-nested.c: Likewise.
30765 * dumpfile.c (dump_start): Use TDF_FLAGS.
30766 (dump_enable_all): Fix TDF_KIND check thinko.
30768 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30770 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30771 array entries to represent two legal parameterizations of the
30772 overloaded __builtin_cmpb function, as represented by the
30773 P6_OV_BUILTIN_CMPB constant.
30774 (altivec_resolve_overloaded_builtin): Add special case handling
30775 for the __builtin_cmpb function, as represented by the
30776 P6_OV_BUILTIN_CMPB constant.
30777 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30778 (BU_P6_64BIT_2): New macro.
30779 (BU_P6_OVERLOAD_2): New macro
30780 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30781 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30782 (CMPB): Add overload support to represent both 32-bit and 64-bit
30783 compare-bytes function.
30784 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30785 support for TARGET_CMPB.
30786 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30787 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30788 documentation of the __builtin_cmpb overloaded built-in function.
30790 2017-05-11 Richard Biener <rguenther@suse.de>
30792 PR tree-optimization/80705
30793 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30794 bases are not vectorizable.
30796 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30798 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30799 when counting register pressure.
30801 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30803 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30804 (iv_ca_more_deps): Renamed to ...
30805 (iv_ca_compare_deps): ... this.
30806 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30808 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30810 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30812 (determine_group_iv_costs): ... here.
30813 (find_inv_vars_cb): Record inv var if it's not recorded before.
30815 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30817 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30818 (get_shiftadd_cost): Ditto.
30820 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30822 * tree-ssa-address.c: Include header file.
30823 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30825 (add_to_parts): Refactor.
30826 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30827 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30830 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30832 PR tree-optimization/53090
30833 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30835 (extract_cond_operands): Detect condition with IV on both sides
30836 and return COMP_IV_EXPR_2.
30837 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30838 (rewrite_use_compare): Simplify by removing call to function
30839 extract_cond_operands.
30841 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30843 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30844 (extract_cond_operands): Detect condition comparing against non-
30845 invariant bound and return appropriate enum value.
30846 (find_interesting_uses_cond): Update use of extract_cond_operands.
30847 Handle its return value accordingly.
30848 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30850 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30852 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30853 nonlinear iv_use computation in loop invariant sensitive way.
30855 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30857 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30858 (find_iv_candidates): Call relate_compare_use_with_all_cands.
30860 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30862 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
30863 (dump_cand): Support iv_cand.inv_exprs.
30864 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
30866 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
30869 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30871 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
30873 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
30874 as local function. Include necessary header files.
30875 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
30877 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30879 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
30881 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30883 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
30884 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
30885 RSHIFT_EXPR and BIT_NOT_EXPR.
30887 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30889 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
30890 (adjust_setup_cost): New parameter supporting round up adjustment.
30891 (struct address_cost_data): Delete.
30892 (force_expr_to_var_cost): Don't bound cost with spill_cost.
30893 (split_address_cost, ptr_difference_cost): Delete.
30894 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
30895 (struct ainc_cost_data): New struct.
30896 (get_address_cost_ainc): New function.
30897 (get_address_cost, get_computation_cost): Reimplement.
30898 (determine_group_iv_cost_address): Record inv_expr for all uses of
30900 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
30901 (iv_ca_has_deps): Reimplemented to ...
30902 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
30904 (iv_ca_extend): Call iv_ca_more_deps.
30906 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30908 * tree-ssa-address.c (struct mem_address): Move to header file.
30909 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
30910 * tree-ssa-address.h (struct mem_address): Move from C file.
30911 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
30913 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30915 * tree-affine.h (aff_combination_type): New interface.
30916 (aff_combination_zero_p): Remove static.
30917 (aff_combination_const_p): New interface.
30918 (aff_combination_singleton_var_p): New interfaces.
30920 2017-05-11 Richard Biener <rguenther@suse.de>
30922 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30923 Skip unreachable blocks and destinations.
30924 (eliminate): Move stmt removal and fixup ...
30925 (fini_eliminate): ... here. Skip inserted exprs.
30926 (pass_pre::execute): Move fini_pre after fini_eliminate.
30927 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
30928 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
30929 PRE to get rid of dead code that has invalid SSA form and
30930 split critical edges again.
30932 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30934 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30936 2017-05-11 Richard Biener <rguenther@suse.de>
30938 * passes.c (execute_function_todo): Verify loops if they are
30939 said to be up-to-date.
30940 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
30941 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
30943 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
30946 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
30947 handle calling assemble_external ourself.
30950 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
30951 modes with zero size. Enhance comment.
30953 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30955 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
30956 built-ins for vec_xl and vec_xst with short and char pointer
30959 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30961 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
30962 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
30963 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
30964 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
30965 (_mm_maskz_min_round_ss): New intrinsics.
30966 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
30967 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
30968 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
30969 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
30970 (__builtin_ia32_minss_mask_round): New builtins.
30971 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30972 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30973 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
30975 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
30976 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
30978 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
30981 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30983 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
30984 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
30985 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
30986 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
30987 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
30988 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
30989 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
30990 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30991 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30992 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
30993 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
30994 (__builtin_ia32_mulss_mask_round): New builtins.
30995 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30996 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30997 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
30999 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31000 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31002 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31005 2017-05-10 Julia Koval <julia.koval@intel.com>
31007 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31008 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31009 (_mm256_setr_m128i): New intrinsics.
31011 2017-05-10 Julia Koval <julia.koval@intel.com>
31013 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31014 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31015 (_mm_maskz_rcp14_ss): New intrinsics.
31016 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31017 (__builtin_ia32_rcp14ss_mask): New builtins.
31018 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31020 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31022 PR tree-optimization/51513
31023 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31024 (assert_unreachable_fallthru_edge_p): Use it.
31025 (group_case_labels_stmt): Likewise.
31026 * tree-cfg.h: Prototype it.
31027 * stmt.c: Include cfghooks.h and tree-cfg.h.
31028 (emit_case_dispatch_table) <gap_label>: New local variable.
31029 Use it to fill dispatch table gaps.
31030 Test for default_label before updating probabilities.
31031 (expand_case) <default_label>: Remove unneeded initialization.
31032 Test for unreachable default case statement and remove its edge.
31033 Set default_label accordingly.
31034 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31036 2017-05-10 Carl Love <cel@us.ibm.com>
31038 * config/rs6000/rs6000-c: Add support for built-in functions
31039 vector signed char vec_neg (vector signed char)
31040 vector signed short int vec_neg (vector short int)
31041 vector signed int vec_neg (vector signed int)
31042 vector signed long long vec_neg (vector signed long long)
31043 vector float vec_neg (vector float)
31044 vector double vec_neg (vector double)
31045 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31047 * config/rs6000/altivec.h: Add define for vec_neg
31048 * doc/extend.texi: Update the built-in documentation for the
31049 new built-in functions.
31051 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31053 PR tree-optimization/77644
31054 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31056 2017-05-10 Nathan Sidwell <nathan@acm.org>
31058 * dumpfile.h (TDI_lang_all): New.
31059 (TDF_KIND): New. Renumber others
31060 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31062 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31064 (get_dump_file_name): Adjust suffix generation.
31065 (dump_enable_all): Use TDF_KIND.
31066 * doc/invoke.texi (-fdump-lang-all): Document.
31068 * dumpfile.h: Tabify.
31070 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31073 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31074 Move member access before delete.
31076 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31078 * tree-inline.c (expand_call_inline): Split block at stmt
31081 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31084 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31085 are now unused after splitting mov{sf,sd}_hardfloat.
31086 (f32_lr2): Likewise.
31087 (f32_lm): Likewise.
31088 (f32_lm2): Likewise.
31089 (f32_li): Likewise.
31090 (f32_li2): Likewise.
31091 (f32_lv): Likewise.
31092 (f32_sr): Likewise.
31093 (f32_sr2): Likewise.
31094 (f32_sm): Likewise.
31095 (f32_sm2): Likewise.
31096 (f32_si): Likewise.
31097 (f32_si2): Likewise.
31098 (f32_sv): Likewise.
31099 (f32_dm): Likewise.
31100 (f32_vsx): Likewise.
31101 (f32_av): Likewise.
31102 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31103 For movsf, order stores so the VSX stores occur before the GPR
31104 store which encourages the register allocator to use a traditional
31105 FPR instead of a GPR. For movsd, order the stores so that the GPR
31106 store comes before the VSX stores to allow the power6 to work.
31107 This is due to the power6 not having a 32-bit integer store
31108 instruction from a FPR.
31109 (movsf_hardfloat): Likewise.
31110 (movsd_hardfloat): Likewise.
31112 2017-05-09 Martin Sebor <msebor@redhat.com>
31114 PR translation/80280
31115 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31118 PR translation/80280
31119 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31120 data member added in r247778.
31121 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31123 2017-05-09 Nathan Sidwell <nathan@acm.org>
31125 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31127 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31130 2017-05-09 Marek Polacek <polacek@redhat.com>
31132 * doc/invoke.texi: Fix typo.
31134 2017-05-09 Richard Biener <rguenther@suse.de>
31136 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31137 (vrp_val_is_min): Likewise.
31138 (set_value_range_to_value): Likewise.
31139 (set_value_range_to_nonnegative): Likewise.
31140 (gimple_assign_nonzero_p): Likewise.
31141 (gimple_stmt_nonzero_p): Likewise.
31142 (vrp_int_const_binop): Likewise. Remove unreachable case.
31143 (adjust_range_with_scev): Adjust comments.
31144 (compare_range_with_value): Likewise.
31145 (extract_range_from_phi_node): Likewise.
31146 (test_for_singularity): Likewise.
31148 2017-05-09 Richard Biener <rguenther@suse.de>
31150 * tree-vrp.c (get_single_symbol): Add assert that we don't
31151 get overflowed constants as invariant part.
31152 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31153 checks. Use wi::cmp instead of recursing for integer constants.
31154 (compare_values): Just ignore whether we assumed undefined
31155 overflow instead of failing the compare.
31156 (extract_range_for_var_from_comparison_expr): Add comment before the
31157 TREE_NO_WARNING sets.
31158 (test_for_singularity): Likewise.
31159 (extract_range_from_comparison): Do not disable optimization
31160 when we assumed undefined overflow.
31161 (extract_range_basic): Remove init of unused var.
31163 2017-05-09 Richard Biener <rguenther@suse.de>
31165 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31166 (extract_range_from_multiplicative_op_1): Adjust.
31167 (extract_range_from_binary_expr_1): Use int_const_binop.
31169 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31172 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31173 rs6000_store_data_bypass_p in seven define_bypass directives and
31174 in several comments.
31175 * config/rs6000/rs6000-protos.h: Add prototype for
31176 rs6000_store_data_bypass_p function.
31177 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31178 function implements slightly different (rs6000-specific) semantics
31179 than store_data_bypass_p, returning false rather than aborting
31180 with assertion error when arguments do not satisfy the
31181 requirements of store data bypass.
31182 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31183 rs6000_store_data_bypass_p.
31185 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31187 * config/xtensa/xtensa-protos.h
31188 (xtensa_initial_elimination_offset): New declaration.
31189 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31190 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31191 macro definition, add case for FRAME_POINTER_REGNUM when
31192 FRAME_GROWS_DOWNWARD.
31193 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31194 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31195 xtensa_initial_elimination_offset.
31197 2017-05-08 Nathan Sidwell <nathan@acm.org>
31199 * doc/invoke.texi: Alphabetize -fdump options.
31201 2017-05-08 Martin Sebor <msebor@redhat.com>
31203 PR translation/80280
31204 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31206 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31208 * target.def (compute_frame_layout): New optional target hook.
31209 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31210 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31211 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31213 * reload1.c (verify_initial_elim_offsets): Likewise.
31214 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31215 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31216 (arm_get_frame_offsets): Split up into this ...
31217 (arm_compute_frame_layout): ... and this function.
31219 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31221 * config/aarch64/constraints.md (Usa): New constraint.
31222 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31224 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31226 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31227 with_multilib_list after it has been checked.
31229 2017-05-08 Richard Biener <rguenther@suse.de>
31231 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31232 (bitmap_set_subtract_values): Likewise.
31234 2017-05-08 Richard Biener <rguenther@suse.de>
31236 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31237 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31239 (gimple_stmt_nonzero_warnv_p): Rename to ...
31240 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31242 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31243 (extract_range_basic): Adjust, do not disable propagation on
31244 strict overflow sensitive simplification.
31245 (vrp_visit_cond_stmt): Likewise.
31247 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31249 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31250 body size unconditionally.
31252 2017-05-07 Jeff Law <law@redhat.com>
31255 2017-05-06 Jeff Law <law@redhat.com>
31256 PR tree-optimization/78496
31257 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31260 PR tree-optimization/78496
31261 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31262 (simplify_stmt_using_ranges): Call it.
31263 (vrp_dom_walker::before_dom_children): Extract equivalences
31264 from an ASSERT_EXPR with an equality comparison against a
31267 2017-05-06 Jeff Law <law@redhat.com>
31269 PR tree-optimization/78496
31270 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31273 PR tree-optimization/78496
31274 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31275 (simplify_stmt_using_ranges): Call it.
31276 (vrp_dom_walker::before_dom_children): Extract equivalences
31277 from an ASSERT_EXPR with an equality comparison against a
31280 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31282 * lra-constraints.c (lra_copy_reg_equiv): New function.
31283 (split_reg): Use it to copy equivalence information from the
31284 original register to the spill register.
31286 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31288 PR rtl-optimization/75964
31289 * simplify-rtx.c (simplify_const_relational_operation): Remove
31290 invalid handling of comparisons of integer ABS.
31292 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31294 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31295 initialize to zero.
31296 (init_regs): Remove declaration.
31297 (function_arg_advance_32): Initialize error_p as boolean variable.
31299 2017-05-05 Nathan Sidwell <nathan@acm.org>
31301 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31302 lines. Use for (;;).
31304 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31306 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31307 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31308 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31309 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31310 VF=2 that require versioning.
31312 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31314 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31317 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31319 * diagnostic.h (diagnostic_override_option_index): Convert from
31320 macro to inline function.
31322 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31324 * diagnostic.c (last_module_changed_p): New function.
31325 (set_last_module): New function.
31326 (diagnostic_report_current_module): Convert macro usage to
31327 the above functions.
31328 * diagnostic.h (diagnostic_context::last_module): Strengthen
31329 from const line_map * to const line_map_ordinary *.
31330 (diagnostic_last_module_changed): Delete macro.
31331 (diagnostic_set_last_module): Delete macro.
31333 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31335 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31336 with diagnostic_report_diagnostic.
31337 (diagnostic_n_impl_richloc): Likewise.
31338 * diagnostic.h (report_diagnostic): Delete macro.
31339 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31340 with diagnostic_report_diagnostic.
31341 * substring-locations.c (format_warning_va): Likewise.
31343 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31345 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31346 save/restor of format_spec. Move option-printing code to...
31347 (print_option_information): ...this new function, and
31348 reimplement by simply printing to the pretty_printer,
31349 rather than appending to the format string.
31351 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31353 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31354 handling logic into...
31355 (update_effective_level_from_pragmas): ...this new function.
31357 2017-05-04 Andrew Waterman <andrew@sifive.com>
31359 * config/riscv/riscv.opt (mstrict-align): New option.
31360 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31361 (SLOW_UNALIGNED_ACCESS): Define.
31362 (riscv_slow_unaligned_access): Declare.
31363 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31365 (riscv_slow_unaligned_access): New variable.
31366 (rocket_tune_info): Set slow_unaligned_access to true.
31367 (optimize_size_tune_info): Set slow_unaligned_access to false.
31368 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31369 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31370 (riscv_option_override): Set riscv_slow_unaligned_access.
31371 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31373 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31375 * config/riscv/riscv.md: Unify indentation.
31377 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31382 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31384 (extendsi<mode>2): Add support for doing sign extension via
31385 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31386 don't have ISA 3.0 instructions.
31387 (extendsi<mode>2 splitter): Likewise.
31388 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31389 generate the normal insns since SImode can now go in vector
31390 registers. Disallow the special UNSPECs needed for previous
31391 machines to hide SImode being used. Add new insns
31392 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31393 (fix_trunc<mode>si2_stfiwx): Likewise.
31394 (fix_trunc<mode>si2_internal): Likewise.
31395 (fixuns_trunc<mode>si2): Likewise.
31396 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31397 (fctiw<u>z_<mode>_smallint): Likewise.
31398 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31399 of floating point to 32-bit integer from doing a direct move to
31400 the GPR registers to do a store.
31401 (fctiwz_<mode>): Break long line.
31403 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31405 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31406 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31407 (addr_list, addr_offset_valid_p): New.
31408 (split_address_groups): Check offset validity with above function.
31409 (gt-tree-ssa-loop-ivopts.h): Include header file.
31411 2017-05-05 Nathan Sidwell <nathan@acm.org>
31413 * config.gcc (arm*-*-*): Add missing 'fi'.
31415 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31417 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31418 included in -fopt-info does not matter.
31419 * doc/optinfo.texi (-fopt-info): Fix description of default
31420 behavour. Explicitly say order of options included in -fopt-info
31423 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31425 * config.gcc: Allow combinations of aprofile and rmprofile values for
31426 --with-multilib-list.
31427 * config/arm/t-multilib: New file.
31428 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31429 variables. Remove setting of ISA and floating-point ABI in
31430 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31431 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31432 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31433 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31435 * config/arm/t-rmprofile: Likewise except for the matches changes.
31436 * doc/install.texi (--with-multilib-list): Document the combination of
31437 aprofile and rmprofile values and warn about pitfalls in doing that.
31439 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31441 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31442 (movdi_aarch64): Likewise.
31444 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31446 PR tree-optimization/80632
31447 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31449 (build_arrays): Initialize it for virtual phis.
31450 (fix_phi_nodes): Use it for virtual phis.
31452 PR tree-optimization/80558
31453 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31454 [x, y] op z into [x op, y op z] for op & or | if conditions
31457 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31458 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31461 * config/arm/arm.md (use_literal_pool): Remove.
31462 (64-bit immediate split): No longer takes cost into consideration
31463 if arm_disable_literal_pool is enabled.
31464 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31465 used when arm_disable_literal_pool is enabled.
31466 (arm_max_const_double_inline_cost): Remove use of
31467 arm_disable_literal_pool.
31468 (push_minipool_fix): Add assert.
31469 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31470 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31471 (no_literal_pool_sf_immediate): New.
31473 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31475 PR tree-optimization/80613
31476 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31477 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31479 2017-05-05 Richard Biener <rguenther@suse.de>
31481 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31483 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31485 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31486 of this flag from insn conditions due to removal from r247495.
31488 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31490 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31492 (arm_early_store_addr_dep_ptr): Likewise.
31493 * config/arm/aarch-common-protos.h
31494 (arm_early_load_addr_dep_ptr): Add prototype.
31495 (arm_early_store_addr_dep_ptr): Likewise.
31496 * config/arm/cortex-a53.md: Add new bypasses.
31498 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31500 * tree.c (next_type_uid): Change type to unsigned.
31501 (type_hash_canon): Decrement back next_type_uid if
31502 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31503 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31506 2017-05-04 Martin Sebor <msebor@redhat.com>
31508 * builtins.c: Fix a trivial typo in a comment.
31510 PR middle-end/79234
31511 * builtins.c (check_sizes): Adjust to handle reading past the end.
31512 Avoid printing excessive upper bound of ranges. Use %E to print
31513 tree nodes instead of converting them to %wu.
31514 (expand_builtin_memchr): New function.
31515 (compute_dest_size): Rename...
31516 (compute_objsize): ...to this.
31517 (expand_builtin_memcpy): Adjust.
31518 (expand_builtin_mempcpy): Adjust.
31519 (expand_builtin_strcat): Adjust.
31520 (expand_builtin_strcpy): Adjust.
31521 (check_strncat_sizes): Adjust.
31522 (expand_builtin_strncat): Adjust.
31523 (expand_builtin_strncpy): Adjust and simplify.
31524 (expand_builtin_memset): Adjust.
31525 (expand_builtin_bzero): Adjust.
31526 (expand_builtin_memcmp): Adjust.
31527 (expand_builtin): Handle memcmp.
31528 (maybe_emit_chk_warning): Check strncat just once.
31530 2017-05-04 Martin Sebor <msebor@redhat.com>
31532 PR preprocessor/79214
31533 PR middle-end/79222
31534 PR middle-end/79223
31535 * builtins.c (check_sizes): Add inlining context and issue
31536 warnings even when -Wno-system-headers is set.
31537 (check_strncat_sizes): Same.
31538 (expand_builtin_strncat): Same.
31539 (expand_builtin_memmove): New function.
31540 (expand_builtin_stpncpy): Same.
31541 (expand_builtin): Handle memmove and stpncpy.
31543 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31545 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31546 which is not used any more.
31548 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31550 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31552 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31554 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31555 (cortexa53_tunings): Likewise.
31556 (cortexa57_tunings): Likewise.
31557 (cortexa72_tunings): Likewise.
31558 (cortexa73_tunings): Likewise.
31560 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31562 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31563 Set loop alignment to 8.
31565 2017-05-04 Martin Sebor <msebor@redhat.com>
31567 PR translation/80280
31568 * builtins.c (expand_builtin_object_size): Add missing quoting to
31569 %D and like directives.
31570 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31571 (hsa_type_for_tree_type): Same.
31572 (verify_function_arguments): Same.
31573 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31574 * varasm.c (get_section): Same.
31577 2017-05-04 Martin Sebor <msebor@redhat.com>
31579 PR translation/80280
31580 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31582 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31584 * config/aarch64/aarch64.c (generic_addrcost_table):
31585 Change HI/TI mode setting.
31587 2017-05-04 Martin Jambor <mjambor@suse.cz>
31589 PR tree-optimization/80622
31590 * tree-sra.c (comes_initialized_p): New function.
31591 (build_accesses_from_assign): Only set write lazily when
31592 comes_initialized_p is false.
31593 (analyze_access_subtree): Use comes_initialized_p.
31594 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31595 instead of testing for PARM_DECL.
31597 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31599 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31600 constraint on operand 0 to allow more general addressing modes.
31601 Adjust output template.
31602 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31604 * config/aarch64/aarch64-protos.h
31605 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31606 * config/aarch64/constraints.md (Dp): New address constraint.
31607 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31610 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31612 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31613 update use of estimate_ipcp_clone_size_and_time.
31614 (estimate_local_effects): Update use of
31615 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31616 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31617 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31618 Return nonspecialized time.
31620 2017-05-04 Richard Biener <rguenther@suse.de>
31622 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31623 for the last VUSE which def dominates the PHI. Directly call
31625 (get_continuation_for_phi_1): Remove.
31627 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31629 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31630 to explain the use of truncating division. Cap the number of
31631 iterations to the maximum given by nb_iterations_upper_bound,
31634 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31636 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31637 * configure: Regenerate.
31638 * config.in: Regenerate.
31639 * config/i386/driver-mingw32.c: new file.
31640 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31641 * config.host: Link driver-mingw32.o on MinGW host.
31642 * doc/install.texi: Document new --enable-mingw-wildcard configure
31645 2017-05-04 Marek Polacek <polacek@redhat.com>
31647 PR tree-optimization/80612
31648 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31650 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31651 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31653 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31654 (movt splitter): Likewise.
31655 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31656 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31657 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31658 block for Thumb-1 with MOVT.
31659 (thumb2_legitimate_address_p): Move code block ...
31660 (can_avoid_literal_pool_for_label_p): ... into this new function.
31661 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31663 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31664 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31665 "M-profile targets with the MOVT instruction".
31667 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31669 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31670 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31671 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31673 2017-05-04 Martin Liska <mliska@suse.cz>
31675 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31676 variable cond_code.
31678 2017-05-04 Richard Biener <rguenther@suse.de>
31680 * tree.c (array_at_struct_end_p): Handle arrays at struct
31681 end with flexarrays more conservatively. Refactor and treat
31682 arrays of arrays or aggregates more strict. Fix
31683 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31684 * tree.c (array_at_struct_end_p): Adjust prototype.
31685 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31686 * gimple-fold.c (get_range_strlen): Likewise.
31687 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31689 2017-05-04 Richard Biener <rguenther@suse.de>
31691 PR tree-optimization/31130
31692 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31694 (supports_overflow_infinity): Likewise.
31695 (is_negative_overflow_infinity): Likewise.
31696 (is_positive_overflow_infinity): Likewise.
31697 (is_overflow_infinity): Likewise.
31698 (stmt_overflow_infinity): Likewise.
31699 (overflow_infinity_range_p): Likewise.
31700 (usable_range_p): Remove as always returning true.
31701 (make_overflow_infinity): Remove.
31702 (negative_overflow_infinity): Likewise.
31703 (positive_overflow_infinity): Likewise.
31704 (avoid_overflow_infinity): Likewise.
31705 (set_value_range): Adjust accordingly.
31706 (set_value_range_to_nonnegative): Likewise, remove now unused
31707 overflow_infinity arg.
31708 (vrp_operand_equal_p): Adjust.
31709 (update_value_range): Likewise.
31710 (range_int_cst_singleton_p): Likewise.
31711 (operand_less_p): Likewise.
31712 (compare_values_warnv): Likewise.
31713 (extract_range_for_var_from_comparison_expr): Likewise.
31714 (vrp_int_const_binop): Likewise.
31715 (zero_nonzero_bits_from_vr): Likewise.
31716 (extract_range_from_multiplicative_op_1): Likewise.
31717 (extract_range_from_binary_expr_1): Likewise.
31718 (extract_range_from_unary_expr): Likewise.
31719 (extract_range_from_comparison): Likewise.
31720 (extract_range_basic): Likewise.
31721 (adjust_range_with_scev): Likewise.
31722 (compare_ranges): Likewise.
31723 (compare_range_with_value): Likewise.
31724 (dump_value_range): Likewise.
31725 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31727 (simplify_cond_using_ranges): Adjust.
31729 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31731 * brig-builtins.def: Added a builtin for class_f64.
31732 * builtin-types.def: Added a builtin type needed by class_f64.
31734 2017-05-03 Jason Merrill <jason@redhat.com>
31736 * timevar.def: Add TV_CONSTEXPR.
31738 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31740 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31742 2017-05-03 Martin Jambor <mjambor@suse.cz>
31744 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31745 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31746 * ipa-cp.c (ipcp_propagate_stage): Do not call
31747 ipa_update_after_lto_read.
31748 * ipa-inline.c (ipa_inline): Likewise.
31750 2017-05-03 Martin Jambor <mjambor@suse.cz>
31752 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31753 tag. Added a default constructor and a destructor.
31754 (ipa_edge_args_sum_t): New class;
31755 (ipa_edge_args_sum): Declare.
31756 (ipa_edge_args_vector): Remove declaration.
31757 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31758 (ipa_free_edge_args_substructures): Remove declaration.
31759 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31760 (ipa_edge_args_info_available_for_edge_p): Likewise.
31761 * ipa-prop.c (ipa_edge_args_vector): Removed.
31762 (edge_removal_hook_holder): Likewise.
31763 (edge_duplication_hook_holder): Likewise.
31764 (ipa_edge_args_sum): New variable.
31765 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31766 ipa_edge_args_vector.
31767 (ipa_free_edge_args_substructures): Likewise.
31768 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31769 ipa_edge_args_vector.
31770 (ipa_edge_removal_hook): Turned into method
31771 ipa_edge_args_sum_t::remove.
31772 (ipa_edge_duplication_hook): Turned into method
31773 ipa_edge_args_sum_t::duplicate.
31774 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31775 registering edge hooks.
31776 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31777 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31778 ipa_edge_args_sum instead of ipa_edge_args_vector.
31779 * ipa-profile.c (ipa_profile): Likewise.
31781 2017-05-03 Martin Jambor <mjambor@suse.cz>
31783 * symbol-summary.h (function_summary): New method exists.
31784 (function_summary::symtab_removal): Deallocate through release.
31785 (call_summary): New class.
31786 (gt_ggc_mx): New overload.
31787 (gt_pch_nx): Likewise.
31788 (gt_pch_nx): Likewise.
31790 2017-05-03 Jeff Law <law@redhat.com>
31792 PR tree-optimization/78496
31793 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31794 from simplify_cond_using_ranges. Split off code to walk
31795 backwards through casts into ...
31796 (simplify_cond_using_ranges_2): New function.
31797 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31798 (execute_vrp): After identifying jump threads, call
31799 simplify_cond_using_ranges_2.
31801 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31804 * ipa-inline.h (inline_summary): Add ctor.
31805 (create_ggc): Do not use ggc_cleared_alloc.
31807 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31808 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31810 * gcc.c (handle_braces): Support escaping in switch matching
31812 * doc/invoke.texi (Spec Files): Document it.
31813 Remove superfluous @code markup in items.
31815 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31817 * diagnostic-show-locus.c (struct column_range): New struct.
31818 (get_affected_columns): New function.
31819 (get_printed_columns): New function.
31820 (struct correction): New struct.
31821 (correction::ensure_capacity): New function.
31822 (correction::ensure_terminated): New function.
31823 (struct line_corrections): New struct.
31824 (line_corrections::~line_corrections): New dtor.
31825 (line_corrections::add_hint): New function.
31826 (layout::print_trailing_fixits): Reimplement in terms of the new
31828 (selftest::test_overlapped_fixit_printing): New function.
31829 (selftest::diagnostic_show_locus_c_tests): Call it.
31831 2017-05-03 Nathan Sidwell <nathan@acm.org>
31833 Canonicalize canonical type hashing
31834 * tree.h (type_hash_canon_hash): Declare.
31835 * tree.c (type_hash_list, attribute_hash_list): Move into
31836 type_hash_canon_hash.
31837 (build_type_attribute_qual_variant): Break out hash code calc into
31838 type_hash_canon_hash.
31839 (type_hash_canon_hash): New. Generic type hash computation.
31840 (build_range_type_1, build_array_type_1, build_function_type,
31841 build_method_type_directly, build_offset_type, build_complex_type,
31842 make_vector_type): Call it.
31844 2017-05-03 Richard Biener <rguenther@suse.de>
31846 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31847 When all DRs have unknown misaligned do not always peel
31848 when there is a store but apply the same costing model as if
31849 there were only loads.
31851 2017-05-03 Richard Biener <rguenther@suse.de>
31854 PR tree-optimization/80492
31855 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31856 compare_base_decls returning dont-know properly.
31858 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
31860 * config/arm/iterators.md (CCSI): New mode iterator.
31861 (arch): New mode attribute.
31862 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
31863 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
31864 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
31865 code iterator for success result mode.
31866 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
31867 the corresponding new insn generators.
31869 2017-05-03 Bin Cheng <bin.cheng@arm.com>
31872 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31873 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31875 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
31877 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
31878 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
31879 (DDR_A): Wrap DDR argument in brackets.
31880 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
31881 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
31882 (DDR_REVERSED_P): Likewise.
31884 2017-05-03 Jakub Jelinek <jakub@redhat.com>
31886 PR tree-optimization/79472
31887 * tree-switch-conversion.c (struct switch_conv_info): Add
31888 contiguous_range and default_case_nonstandard fields.
31889 (collect_switch_conv_info): Compute contiguous_range and
31890 default_case_nonstandard fields, don't clear final_bb if
31891 contiguous_range and only the default case doesn't have the required
31893 (check_all_empty_except_final): Set default_case_nonstandard instead
31894 of failing if contiguous_range and the default case doesn't have empty
31896 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
31897 and only the default case doesn't have the required constants. Skip
31899 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
31900 if default_case_nonstandard.
31901 (build_constructors): Build constant 1 just once. Assert that default
31902 values aren't inserted in between cases if contiguous_range. Skip
31904 (build_arrays): Skip virtual phis.
31905 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
31906 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
31907 Handle virtual phis.
31908 (gen_inbound_check): Handle default_case_nonstandard case.
31909 (process_switch): Adjust check_final_bb caller. Call
31910 gather_default_values with the first non-default case instead of
31911 default case if default_case_nonstandard.
31913 2017-05-02 Nathan Sidwell <nathan@acm.org>
31915 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
31916 check. Fix formatting.
31918 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
31920 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
31921 errors when comparing specialized and unspecialized times.
31923 2017-05-02 David Malcolm <dmalcolm@redhat.com>
31925 * diagnostic-show-locus.c
31926 (layout::should_print_annotation_line_p): Make private.
31927 (layout::print_annotation_line): Make private.
31928 (layout::annotation_line_showed_range_p): Make private.
31929 (layout::show_ruler): Make private.
31930 (layout::print_source_line): Make private. Pass in line and
31931 line_width, rather than calling location_get_source_line. Drop
31933 (layout::print_leading_fixits): New method.
31934 (layout::print_any_fixits): Rename to...
31935 (layout::print_trailing_fixits): ...this, and make private.
31936 Don't print newline fixits.
31937 (diagnostic_show_locus): Move logic for printing one row into...
31938 (layout::print_line): ...this new function. Move the
31939 location_get_source_line call and error-handling from
31940 print_source_line to here. Call print_leading_fixits, and rename
31941 print_any_fixits to print_trailing_fixits.
31942 (selftest::test_fixit_insert_containing_newline): Update now that
31943 newlines are partially supported.
31944 (selftest::test_fixit_insert_containing_newline_2): New test.
31945 (selftest::test_fixit_replace_containing_newline): Update comments.
31946 (selftest::diagnostic_show_locus_c_tests): Call the new test.
31947 * edit-context.c (class added_line): New class.
31948 (class edited_line): Describe newline handling in comment.
31949 (edited_line::actually_edited_p): New method.
31950 (edited_line::print_content): Delete redundant decl.
31951 (edited_line::m_predecessors): New field.
31952 (edited_file::print_content): Call edited_line::print_content.
31953 (edited_file::print_diff): Update to support newlines.
31954 (edited_file::print_diff_hunk): Likewise.
31955 (edited_file::print_run_of_changed_lines): New function.
31956 (edited_file::print_diff_line): Convert to...
31957 (print_diff_line): ...this.
31958 (edited_file::get_effective_line_count): New function.
31959 (edited_line::edited_line): Initialize new field m_predecessors.
31960 (edited_line::~edited_line): Clean up m_predecessors.
31961 (edited_line::apply_fixit): Handle newlines.
31962 (edited_line::get_effective_line_count): New function.
31963 (edited_line::print_content): New function.
31964 (edited_line::print_diff_lines): New function.
31965 (selftest::test_applying_fixits_insert_containing_newline): New
31967 (selftest::test_applying_fixits_replace_containing_newline): New
31969 (selftest::insert_line): New function.
31970 (selftest::test_applying_fixits_multiple_lines): Add example of
31972 (selftest::edit_context_c_tests): Call the new tests.
31974 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31976 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
31977 parameter cand. Update dump information.
31978 (get_computation_cost): Update uses.
31980 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31982 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
31983 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
31984 (get_computation_at, rewrite_use_address): Update use of
31985 get_computation_aff.
31987 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31989 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
31990 (get_computation): Delete.
31991 (get_computation_cost): Implement like get_computation_cost_at.
31992 Use get_computation_at.
31993 (get_computation_cost_at): Delete.
31994 (rewrite_use_nonlinear_expr): Use get_computation_at.
31995 (rewrite_use_compare, remove_unused_ivs): Ditto.
31997 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31999 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32001 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32003 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32004 (ivopts_global_cost_for_size): Rename parameter and update uses.
32005 (iv_ca_recount_cost): Update uses.
32006 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32007 candidates seperately in n_invs and n_cands.
32008 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32010 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32012 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32013 (find_inv_vars_cb): New.
32014 (find_depends): Renamed to ...
32015 (find_inv_vars): ... this.
32016 (add_candidate_1, force_var_cost): Call find_inv_vars.
32017 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32019 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32021 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32022 inv_vars. Add inv_exprs.
32023 (struct iv_cand): Rename depends_on to inv_vars.
32024 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32025 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32026 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32027 (dump_cand): Dump inv_vars.
32028 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32029 (record_invariant, find_depends, add_candidate_1): Ditto.
32030 (set_group_iv_cost, force_var_cost): Ditto.
32031 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32032 (get_computation_cost_at, get_computation_cost): Ditto.
32033 (determine_group_iv_cost_generic): Ditto.
32034 (determine_group_iv_cost_address): Ditto.
32035 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32036 (determine_group_iv_costs): Ditto.
32037 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32038 (iv_ca_set_remove_invariants): Renamed to ...
32039 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32040 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32041 (iv_ca_set_add_invariants): Renamed to ...
32042 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32043 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32044 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32045 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32046 (create_new_ivs): Remove useless dump.
32048 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32050 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32052 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32053 (iv_ca_set_no_cp, create_new_iv): Ditto.
32055 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32057 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32059 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32061 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32062 function tree_check2.
32064 2017-05-02 Martin Liska <mliska@suse.cz>
32066 * doc/gcov.texi: Add missing preposition.
32067 * gcov.c (function_info::function_info): Properly fill up
32068 all member variables.
32070 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32072 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32074 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32076 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32078 2017-05-02 Martin Liska <mliska@suse.cz>
32081 * lto-streamer-in.c (lto_read_tree_1): Remove
32082 LTO_STREAMER_DEBUG.
32083 * lto-streamer.c (struct tree_hash_entry): Likewise.
32084 (struct tree_entry_hasher): Likewise.
32085 (tree_entry_hasher::hash): Likewise.
32086 (tree_entry_hasher::equal): Likewise.
32087 (lto_streamer_init): Likewise.
32088 (lto_orig_address_map): Likewise.
32089 (lto_orig_address_get): Likewise.
32090 (lto_orig_address_remove): Likewise.
32091 * lto-streamer.h: Likewise.
32092 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32093 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32095 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32097 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32098 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32099 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32100 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32101 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32102 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32103 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32104 (mm_maskz_sub_ss): New intrinsics.
32105 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32106 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32107 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32108 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32109 (__builtin_ia32_subss_mask_round): New builtins.
32110 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32111 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32112 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32114 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32115 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32117 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32120 2017-05-02 Martin Jambor <mjambor@suse.cz>
32122 PR tree-optimization/78687
32123 * tree-sra.c (access): New field parent.
32124 (process_subtree_disqualification): New function.
32125 (disqualify_candidate): Call it.
32126 (build_accesses_from_assign): Reset write flag if creating an
32128 (build_access_subtree): Fill in parent field and also prpagate
32129 down grp_write flag.
32130 (create_artificial_child_access): New parameter set_grp_write, set
32131 grp_write to its value.
32132 (propagate_subaccesses_across_link): Also propagate grp_write flag
32134 (propagate_all_subaccesses): Push the closest parent back to work
32135 queue if add_access_to_work_queue returned true.
32137 2017-05-02 Richard Biener <rguenther@suse.de>
32139 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32140 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32141 -fstrict-overflow documentation.
32142 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32143 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32144 flag_strict_overflow.
32145 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32146 * lto-opts.c (lto_write_options): Do not stream it.
32147 * lto-wrapper.c (merge_and_complain): Do not handle it.
32148 * opts.c (default_options_table): Do not set -fstrict-overflow.
32149 (finish_options): Likewise do not clear it when sanitizing.
32150 * simplify-rtx.c (simplify_const_relational_operation): Do not
32151 test flag_strict_overflow.
32153 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32155 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32156 using enabled attribute.
32157 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32158 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32159 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32160 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32161 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32162 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32163 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32164 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32165 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32166 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32168 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32170 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32172 2017-05-02 Richard Biener <rguenther@suse.de>
32174 PR tree-optimization/80591
32176 2017-04-10 Richard Biener <rguenther@suse.de>
32178 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32181 2017-05-02 Richard Biener <rguenther@suse.de>
32183 PR tree-optimization/80549
32184 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32185 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32186 headers if they do not have a preheader.
32188 2017-05-02 Martin Liska <mliska@suse.cz>
32191 * common.opt: Fix typo.
32192 * doc/invoke.texi: Likewise.
32194 2017-05-01 Jan Beulich <jbeulich@suse.com>
32196 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32197 swapping, add (x,x,m,x,n) alternative.
32199 2017-05-01 Nathan Sidwell <nathan@acm.org>
32201 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32202 unnecessary unadjusted_alignment check.
32204 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32207 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32209 * gimplify.c (gimplify_cilk_detach): New function.
32210 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32211 * tree-core.h: Document EXPR_CILK_SPAWN.
32212 * tree.h (EXPR_CILK_SPAWN): Define.
32214 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32216 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32217 to use new fixit_hint representation, using the "replace" logic.
32218 (get_line_span_for_fixit_hint): Likewise.
32219 (layout::print_any_fixits): Likewise.
32220 (selftest::test_one_liner_many_fixits): Rename to...
32221 (selftest::test_one_liner_many_fixits_1): ...this, and update
32222 comment and expected output to reflect that the multiple fix-it
32223 hints are now consolidated into one insertion.
32224 (selftest::test_one_liner_many_fixits_2): New test.
32225 (selftest::test_diagnostic_show_locus_one_liner): Update for
32227 (selftest::test_fixit_consolidation): Update for fix-it API
32229 * diagnostic.c (print_parseable_fixits): Likewise.
32230 * edit-context.c (edited_line::m_line_events): Convert from
32231 auto_vec <line_event *> to auto_vec <line_event>.
32232 (class line_event): Convert from abstract base class to a concrete
32233 class, taking over the role of replace_event.
32234 (class insert_event): Delete.
32235 (class replace_event): Rename to class line_event. Convert to
32237 (edit_context::add_fixits): Reimplement.
32238 (edit_context::apply_insert): Delete.
32239 (edit_context::apply_replace): Rename to...
32240 (edit_context::apply_fixit): ...this. Convert to half-open range.
32241 (edited_file::apply_insert): Delete.
32242 (edited_file::apply_replace): Rename to...
32243 (edited_file::apply_fixit): ...this.
32244 (edited_line::~edited_line): Drop deletion of events.
32245 (edited_line::apply_insert): Delete.
32246 (edited_line::apply_replace): Rename to...
32247 (edited_line::apply_fixit): ...this. Convert to half-open range.
32248 Update for change to type of m_line_events.
32249 * edit-context.h (edit_context::apply_insert): Delete.
32250 (edit_context::apply_replace): Rename to...
32251 (edit_context::apply_fixit): ...this.
32253 2017-05-01 Martin Sebor <msebor@redhat.com>
32255 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32258 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32261 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32262 __get_cpuid_max returns 0.
32263 (__get_cpuid_count): Ditto.
32265 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32267 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32268 replacement expression is another instance of one of its arguments.
32270 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32273 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32274 check for stack push/pop autoinc.
32275 * config/i386/i386.c (ix86_agi_dependent): Return false
32276 if the only reason why modified_in_p returned true is that
32277 addr is SP based and set_insn is a push or pop.
32279 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32281 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32284 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32287 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32288 (account_size_time): Use two predicates - exec_pred and
32290 (evaluate_conditions_for_known_args): Compute both clause and
32292 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32293 (inline_summary_t::duplicate): Update.
32294 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32296 (compute_inline_parameters): Likewise.
32297 (estimate_edge_size_and_time): Update caluclation of time.
32298 (estimate_node_size_and_time): Compute both time and nonspecialized
32300 (estimate_ipcp_clone_size_and_time): Update.
32301 (inline_merge_summary): Update.
32302 (do_estimate_edge_time): Update.
32303 (do_estimate_edge_size): Update.
32304 (do_estimate_edge_hints): Update.
32305 (inline_read_section, inline_write_summary): Stream both new predicates.
32306 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32308 (compute_inlined_call_time): Cleanup.
32309 (big_speedup_p): Update.
32310 (edge_badness): Update.
32311 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32312 (size_time_entry): Replace predicate by exec_predicate and
32313 nonconst_predicate.
32314 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32315 (estimate_edge_time): Return also nonspec_time.
32316 (reset_edge_growth_cache): Update.
32318 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32320 PR rtl-optimization/80491
32321 * ifcvt.c (noce_process_if_block): When looking for x setter
32322 with missing else_bb, don't check only the insn right before
32323 cond_earliest, but look for the last insn that x is modified in
32324 within the same bb.
32326 PR rtl-optimization/80491
32327 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32329 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32331 PR tree-optimization/80487
32332 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32334 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32336 PR tree-optimization/79697
32337 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32338 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32339 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32341 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32342 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32344 2017-04-28 Martin Sebor <msebor@redhat.com>
32346 PR tree-optimization/80523
32347 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32348 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32350 (maybe_warn, format_directive, parse_directive): Use new functions.
32351 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32353 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32355 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32357 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32359 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32360 target_header_dir): Set correctly.
32361 * configure: Regenerated.
32362 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32363 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32364 instead of SYSTEM_HEADER_DIR.
32366 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32368 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32369 (estimate_local_effects): Likewise.
32370 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32371 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32372 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32373 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32374 do_estimate_edge_time, estimate_edge_time): Likewise.
32375 * ipa-inline-analysis.c (estimate_node_size_and_time,
32376 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32377 (estimate_time_after_inlining): Remove.
32379 2017-04-28 Martin Liska <mliska@suse.cz>
32381 * doc/gcov.texi: Enhance documentation of gcov.
32383 2017-04-28 Martin Liska <mliska@suse.cz>
32385 * doc/gcov.texi: Sort options in alphabetic order.
32386 * doc/gcov-dump.texi: Likewise.
32387 * doc/gcov-tool.texi: Likewise.
32388 * gcov.c (print_usage): Likewise.
32389 * gcov-dump.c (print_usage): Likewise.
32390 * gcov-tool.c (print_merge_usage_message): Likewise.
32391 (print_rewrite_usage_message): Likewise.
32392 (print_overlap_usage_message): Likewise.
32394 2017-04-28 Martin Liska <mliska@suse.cz>
32396 PR gcov-profile/53915
32397 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32399 2017-04-28 Martin Liska <mliska@suse.cz>
32401 PR gcov-profile/79891
32402 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32403 is marked by compiler as living on a line.
32404 (get_cycles_count): Remove usage of the union.
32405 (output_intermediate_file): Likewise.
32406 (find_source): Fix GNU coding style.
32407 (accumulate_line_counts): Remove old non-all block mode.
32408 (output_lines): Remove usage of the union.
32409 * profile.c (output_location): Include all BBs, even if
32410 belonging to a same line (and file) as a previous BB.
32412 2017-04-28 Martin Liska <mliska@suse.cz>
32414 * gcov.c (process_args): Handle new argument 'w'.
32415 (read_graph_file): Assign ID to BBs.
32416 (output_branch_count): Display BB # if verbose flag is set.
32417 (output_lines): Likewise for arcs.
32418 (print_usage): Add '--verbose' option help.
32419 * doc/gcov.texi: Document --verbose (-w) option.
32421 2017-04-28 Martin Liska <mliska@suse.cz>
32423 * gcov.c (struct block_location_info): New struct.
32424 (process_file): Fill up the new structure.
32425 (read_graph_file): Replace usage of encoding by the newly added
32427 (add_line_counts): Likewise.
32428 (accumulate_line_counts): Remove usage of the union.
32429 (function_info::function_info): New function.
32430 (function_info::~function_info): Likewise.
32431 (process_file): Call delete instead of release_function.
32432 (release_function): Release the function.
32433 (release_structures): Call delete instead of release_function.
32434 (solve_flow_graph): Replace usage of num_blocks.
32435 (find_exception_blocks): Likewise.
32436 (output_lines): Fix GNU coding style.
32438 2017-04-28 Martin Liska <mliska@suse.cz>
32441 * coverage.c (coverage_remove_note_file): New function.
32442 * coverage.h: Declare the function.
32443 * toplev.c (finalize): Clean if an error has been seen.
32445 2017-04-28 Martin Liska <mliska@suse.cz>
32447 PR gcov-profile/80031
32448 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32449 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32450 * gcov.c (read_graph_file): Read just number of blocks.
32451 * profile.c (branch_prob): Do not stream 0 flags per a basic
32454 2017-04-28 Martin Liska <mliska@suse.cz>
32456 * gcov-dump.c (tag_*): Add new argument to declarations.
32457 (dump_gcov_file): Likewise.
32458 (tag_blocks): Add and use new argument depth.
32459 (tag_arcs): Likewise.
32460 (tag_lines): Likewise.
32461 (tag_counters): Likewise.
32462 (tag_summary): Likewise.
32463 (dump_working_sets): Use depth to do a proper indentation.
32465 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32468 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32469 * symtab.c (symtab_node::debug_symtab): Move definition here.
32471 2017-04-28 Richard Biener <rguenther@suse.de>
32473 * lto-streamer.h (LTO_major_version): Bump to 7.
32475 2017-04-28 Richard Biener <rguenther@suse.de>
32477 * tree-vrp.c (assert_info): New struct.
32478 (add_assert_info): New helper.
32479 (register_edge_assert_for_2): Refactor to add asserts to a vector
32481 (register_edge_assert_for_1): Likewise.
32482 (register_edge_assert_for): Likewise.
32483 (finish_register_edge_assert_for): New helper actually registering
32484 asserts where live on edge.
32485 (find_conditional_asserts): Adjust.
32486 (find_switch_asserts): Likewise.
32487 (evrp_dom_walker::try_find_new_range): Generalize.
32488 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32490 2017-04-27 Marek Polacek <polacek@redhat.com>
32493 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32494 arg10 and arg11 to itype.
32496 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32498 * doc/extend.texi (Object Size Checking): Improve grammar.
32500 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32503 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32504 that the logic for permitting reciprocal estimates matches that
32507 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32510 * tree.c (type_cache_hasher::equal): Only compare
32511 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32512 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32513 non-aggregate element types.
32514 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32515 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32517 2017-04-27 Richard Biener <rguenther@suse.de>
32519 PR middle-end/80533
32520 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32521 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32522 keeping a reference to a trailing array.
32524 2017-04-27 Richard Biener <rguenther@suse.de>
32526 PR middle-end/80539
32527 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32528 being in loop-closed SSA form conservatively.
32529 (chrec_fold_multiply_poly_poly): Likewise.
32531 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32533 PR middle-end/79665
32534 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32535 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32537 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32540 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32541 (aarch64_function_arg_alignment): Return unsigned int again, but still
32542 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32543 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32544 Don't emit -Wpsabi note.
32545 (aarch64_function_arg_boundary): Likewise.
32546 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32549 2017-04-26 Nathan Sidwell <nathan@acm.org>
32551 * tree.h (crc32_unsigned_n): Declare.
32552 (crc32_unsigned, crc32_unsigned): Make inline.
32553 * tree.c (crc32_unsigned_bits): Replace with ...
32554 (crc32_unsigned_n): ... this.
32555 (crc32_unsigned, crc32_byte): Remove.
32556 (crc32_string): Remove unnecessary braces.
32558 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32560 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32561 * ipa-inline-analysis.c (MAX_TIME): Remove.
32562 (account_size_time): Use sreal for time.
32563 (dump_inline_summary): Update.
32564 (estimate_function_body_sizes): Update.
32565 (estimate_edge_size_and_time): Update.
32566 (estimate_calls_size_and_time): Update.
32567 (estimate_node_size_and_time): Update.
32568 (inline_merge_summary): Update.
32569 (inline_update_overall_summary): Update.
32570 (estimate_time_after_inlining): Update.
32571 (inline_read_section): Update.
32572 (inline_write_summary): Update.
32573 * ipa-inline.c (compute_uninlined_call_time): Update.
32574 (compute_inlined_call_time): Update.
32575 (recursive_inlining): Update.
32576 (inline_small_functions): Update.
32577 (dump_overall_stats): Update.
32578 * ipa-inline.h: Include sreal.h.
32579 (size_time_entry): Turn time to sreal.
32580 (inline_summary): Turn self_time nad time to sreal.
32582 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32584 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32586 (sreal::stream_out, sreal::stream_in): New.
32587 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32589 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32591 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32594 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32597 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32599 (dimode_scalar_chain::compute_convert_gain): Ditto.
32600 (dimode_scalar_chain::make_vector_copies): Ditto.
32601 (dimode_scalar_chain::convert_reg): Ditto.
32602 (dimode_scalar_chain::convert_insn): Ditto.
32603 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32604 (VI248_AVX512BW_1): New mode iterator.
32605 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32606 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32609 2017-04-25 Martin Sebor <msebor@redhat.com>
32611 PR tree-optimization/80497
32612 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32613 constants are representable in HOST_WIDE_INT.
32614 (parse_directive): Ditto.
32616 2017-04-25 Martin Sebor <msebor@redhat.com>
32619 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32620 (new_zero_array): Adjust signature.
32621 (dom_info::dom_init): Used unsigned rather that size_t.
32622 (dom_info::dom_info): Same.
32624 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32625 Jakub Jelinek <jakub@redhat.com>
32628 * config/arm/arm.c: Include gimple.h.
32629 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32630 returns negative, increment ncrn only if it returned positive.
32631 (arm_needs_doubleword_align): Return int instead of bool,
32632 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32633 members, but if there is any such non-FIELD_DECL
32634 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32635 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32636 returns negative, increment nregs only if it returned positive.
32637 (arm_setup_incoming_varargs): Likewise.
32638 (arm_function_arg_boundary): Emit -Wpsabi note if
32639 arm_needs_doubleword_align returns negative, return
32640 DOUBLEWORD_ALIGNMENT only if it returned positive.
32642 2017-04-25 Marek Polacek <polacek@redhat.com>
32645 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32646 first argument to type.
32648 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32651 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32652 type checks to test for compatibility instead of equality.
32654 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32655 Jakub Jelinek <jakub@redhat.com>
32658 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32660 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32661 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32662 the alignment computation, but return their maximum in warn_alignment.
32663 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32664 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32666 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32667 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32670 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32672 * config/arc/simdext.md (dmpyh): Fix typo.
32674 2017-04-25 Richard Biener <rguenther@suse.de>
32676 PR tree-optimization/80492
32677 * alias.c (compare_base_decls): Handle registers with asm
32678 specification conservatively.
32679 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32680 compare_base_decls returning dont-know properly.
32682 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32684 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32685 (legitimate_offset_address_p): New function.
32686 (arc_legitimate_address_p): Use above function.
32688 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32690 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32692 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32694 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32695 ACCH registers whenever they are available.
32697 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32699 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32700 double regs fix when not used.
32702 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32704 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32706 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32707 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32709 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32711 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32712 when emitting short ADD instructions.
32714 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32716 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32718 (cmpsi_cc_c_insn): Likewise.
32719 (cbranchsi4_scratch): Compute proper instruction length using
32720 compact_hreg_operand.
32721 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32723 2017-04-25 Richard Biener <rguenther@suse.de>
32725 PR middle-end/80509
32726 * passes.c (pass_manager::pass_manager): Initialize
32727 m_name_to_pass_map.
32729 2017-04-25 Richard Biener <rguenther@suse.de>
32731 PR tree-optimization/79201
32732 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32734 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32737 * config/s390/vector.md: Split MEM->GPR vector moves for
32738 non-s_operand addresses.
32740 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32743 * config/s390/predicates.md (reload_const_wide_int_operand): New
32745 * config/s390/s390.md ("movti"): Remove d/P alternative.
32746 ("movti_bigconst"): New pattern definition.
32748 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32751 * s390-protos.h (s390_expand_cs_hqi): Removed.
32752 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32753 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32754 modes as well as CCZ1mode and CCZmode.
32755 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32756 signature of s390_emit_compare_and_swap.
32757 (s390_expand_cs_hqi): Likewise, make static.
32758 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32759 compare-and-swap, in some cases.
32760 (s390_expand_cs): Wrapper function.
32761 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32763 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32764 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32765 patterns for small and large integers. Forbid symref memory operands.
32766 Move expander to s390.c. Require cc register.
32767 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32768 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32769 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32770 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32771 symref memory operands. Remove CC mode and call s390_match_ccmode
32773 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32775 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32777 * config/s390/s390.md (define_peephole2): New peephole to help
32778 combining the load-and-test pattern with volatile memory.
32780 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32782 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32783 with CCZmode for TARGET_Z196.
32785 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32787 PR rtl-optimization/80501
32788 * combine.c (make_compound_operation_int): Set subreg_code to SET
32789 even for AND with mask of the sign bit of mode.
32791 PR rtl-optimization/80500
32792 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32793 sum's initial value.
32795 2017-04-25 Julian Brown <julian@codesourcery.com>
32796 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32798 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32800 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32802 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32804 2017-04-25 Julian Brown <julian@codesourcery.com>
32805 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32807 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32808 (thunderx2t99_sha): New Reservation.
32810 2017-04-25 Julian Brown <julian@codesourcery.com>
32811 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32813 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32814 type for 1-element load.
32816 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32818 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32820 2017-04-24 Martin Jambor <mjambor@suse.cz>
32822 PR tree-optimization/80293
32823 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32824 char arrays not totally scalarizable if it is false.
32825 (analyze_all_variable_accesses): Pass correct value in the new
32826 parameter. Add a statistics counter.
32828 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32830 PR middle-end/79931
32831 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32833 2017-04-24 Richard Biener <rguenther@suse.de>
32835 PR tree-optimization/80494
32836 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32837 out for complex types.
32839 2017-04-24 Richard Biener <rguenther@suse.de>
32841 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32842 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32843 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32844 (DFS): Adjust and never fail.
32845 (sccvn_dom_walker::fail): Remove.
32846 (sccvn_dom_walker::before_dom_children): Adjust.
32847 (run_scc_vn): Likewise and never fail.
32848 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32849 (pass_fre::execute): Likewise.
32851 2017-04-24 Richard Biener <rguenther@suse.de>
32853 PR tree-optimization/79725
32854 * tree-ssa-sink.c (statement_sink_location): Return whether
32855 failure reason was zero uses. Move that check later.
32856 (sink_code_in_bb): Deal with zero uses by removing the stmt
32859 2017-04-24 Richard Biener <rguenther@suse.de>
32862 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
32863 pointer-based references.
32865 2017-04-24 Richard Biener <rguenther@suse.de>
32868 * pass_manager.h (pass_manager::operator new): Remove.
32869 (pass_manager::operator delete): Likewise.
32870 * passes.c (pass_manager::operator new): Remove.
32871 (pass_manager::operator delete): Likewise.
32872 (pass_manager::pass_manager): Zero individual pass members.
32874 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
32877 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
32878 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
32879 Check "XEXP (src, 1)" operand here.
32880 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
32881 Check "XEXP (src, 1)" operand here.
32882 (dimode_scalar_chain::make_vector_copies): Detect count register
32883 of a shift instruction. Zero extend count register from QImode
32884 to DImode to satisfy vector shift pattern count operand predicate.
32885 Substitute vector shift count operand with a DImode copy.
32886 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
32889 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
32891 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
32892 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32893 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
32894 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32895 (UNSPEC_NOREX_MEM): Remove definition.
32897 2017-04-21 Richard Biener <rguenther@suse.de>
32899 PR tree-optimization/79547
32900 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
32901 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
32902 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
32903 without any constraints.
32905 2017-04-21 Richard Biener <rguenther@suse.de>
32907 PR tree-optimization/78847
32908 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
32910 2017-04-21 Richard Biener <rguenther@suse.de>
32912 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
32913 (build_distinct_type_copy): Likewise.
32914 (build_variant_type_copy): Likewise.
32915 * tree.c (build_qualified_type): Pass down mem-stat info.
32916 (build_distinct_type_copy): Likewise.
32917 (build_variant_type_copy): Likewise.
32919 2017-04-21 Richard Biener <rguenther@suse.de>
32921 PR tree-optimization/80237
32922 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
32924 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
32925 for a simplified result.
32927 2016-04-21 Richard Biener <rguenther@suse.de>
32929 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
32930 sth as strict as a simple_iv but a chrec without symbols and an
32931 operand defined in the loop we are peeling (and not some subloop).
32932 (propagate_constants_for_unrolling): Propagate all constants.
32934 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32937 * config/i386/i386.c (print_reg): Remove assert for disalowed
32938 regno values, call output_operand_lossage instead.
32940 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32943 * config/i386/constraints.md (Yc): New register constraint.
32944 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
32945 Use Yc constraint for alternative 2 of operand 0. Remove
32946 preferred_for_speed attribute.
32948 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32950 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
32951 lastprivate clauses in SIMT case.
32953 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
32955 * doc/invoke.texi (-Wextra-semi): Document new warning option.
32957 2017-04-20 Richard Biener <rguenther@suse.de>
32959 PR tree-optimization/57796
32960 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
32961 as N scalar stores.
32962 (vect_model_load_cost): Cost gathers as N scalar loads.
32964 2017-04-20 Richard Biener <rguenther@suse.de>
32966 * ggc-page.c (ggc_allocated_p): Rename to ...
32967 (safe_lookup_page_table_entry): ... this and return the lookup
32969 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
32971 2017-04-20 Richard Biener <rguenther@suse.de>
32973 PR tree-optimization/80453
32974 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
32975 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
32976 from the conditions.
32977 (vn_phi_eq): Pass them down.
32978 (vn_phi_lookup): Record them.
32979 (vn_phi_insert): Likewise.
32981 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
32983 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
32984 uninitialized variable warning to avoid buffer overrun.
32986 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32989 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
32990 is suppressed for '{ 0 }' in C.
32992 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32994 * BASE-VER: Set to 8.0.0.
32996 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
32998 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
32999 priority .init_array and .fini_array section with SECTION_NOTYPE
33002 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33004 PR middle-end/80423
33005 * tree.h (build_array_type): Add typeless_storage default argument.
33006 * tree.c (type_cache_hasher::equal): Also compare
33007 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33008 (build_array_type): Add typeless_storage argument, set
33009 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33011 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33012 (build_array_type): Likewise. Add typeless_storage argument.
33014 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33015 Jakub Jelinek <jakub@redhat.com>
33017 PR tree-optimization/80426
33018 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33019 operation on symbolic operands, also compute the overflow for the
33020 invariant part when the operation degenerates into a negation.
33022 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33025 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33026 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33029 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33031 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33034 * config/avr/avr.c (tree.h): Include it.
33035 (cgraph.h): Include it.
33036 (avr_encode_section_info): Don't warn for uninitialized progmem
33037 variable if it's just an alias.
33039 2017-04-19 Richard Biener <rguenther@suse.de>
33042 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33043 when needed by AutoPGO.
33045 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33048 * doc/lto.texi: Remove an extra 'that'.
33050 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33052 PR rtl-optimization/80429
33053 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33054 are only used in debug insns.
33056 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33057 Vladimir Makarov <vmakarov@redhat.com>
33059 * config/sparc/predicates.md (input_operand): Add comment. Return
33060 true for any memory operand when LRA is in progress.
33061 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33063 2017-04-18 Jeff Law <law@redhat.com>
33066 * mips.md ({return,simple_return}_internal): Do not overwrite
33069 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33071 PR tree-optimization/80443
33072 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33073 instead of adding 1, subtract -1 and similarly instead of subtracting
33076 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33078 PR rtl-optimization/80357
33079 * haifa-sched.c (tmp_bitmap): New variable.
33080 (model_recompute): Handle duplicate use records.
33081 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33082 (free_global_sched_pressure_data): Free it.
33084 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33087 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33088 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33089 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33090 instead of SYSTEM_HEADER_DIR.
33092 2017-04-18 Jeff Law <law@redhat.com>
33094 PR middle-end/80422
33095 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33096 predecessors after walking up the insn chain.
33098 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33101 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33102 sizetype type into debug info.
33104 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33107 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33108 unneeded test for TARGET_UPPER_REGS_SF.
33109 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33111 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33114 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33115 instead of gsi_after_labels.
33117 2017-04-18 Jeff Law <law@redhat.com>
33119 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33123 2017-04-13 Jeff Law <law@redhat.com>
33124 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33125 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33127 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33130 * config/avr/avr.c (intl.h): Include it.
33131 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33133 2017-04-18 Martin Liska <mliska@suse.cz>
33135 PR gcov-profile/78783
33136 * gcov-tool.c (gcov_output_files): Validate that destination
33137 file is either removed by the tool or by a user.
33139 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33140 Guy Benyei <guybe@mellanox.com>
33142 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33143 block, and do not negate it, the stored id is already negative.
33145 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33147 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33149 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33152 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33153 masks of options that should be turned off if the VSX vector
33154 options are turned off.
33155 (OTHER_P8_VECTOR_MASKS): Likewise.
33156 (OTHER_VSX_VECTOR_MASKS): Likewise.
33157 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33158 rs6000_disable_incompatible_switches to validate no type switches
33160 (rs6000_incompatible_switch): New function to disallow turning on
33161 other vector options if -mno-vsx, -mno-power8-vector, or
33162 -mno-power9-vector are specified.
33164 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33166 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33168 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33170 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33171 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33172 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33173 (ARG_POINTER_CFA_OFFSET): Likewise.
33175 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33177 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33178 conditions to take advantage of various optimizations.
33180 2017-04-13 Jeff Law <law@redhat.com>
33182 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33183 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33184 (zero_extendsidi2_dext): Likewise.
33186 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33189 * fold-const.c (fold_ternary_loc): Revert
33190 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33193 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33195 PR rtl-optimization/80343
33196 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33197 new scratch pseudo.
33199 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33202 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33203 to ubsan_encode_value.
33205 2017-04-13 Jeff Law <law@redhat.com>
33207 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33208 appearing in DEBUG_INSNs.
33210 2017-04-13 Martin Liska <mliska@suse.cz>
33212 PR gcov-profile/80413
33213 * gcov-io.c (gcov_write_string): Copy to buffer just when
33214 allocated size is greater than zero.
33216 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33219 * dwarf2out.c (decls_for_scope): Ignore declarations of
33220 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33222 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33225 * ipa-visibility.c (non_local_p): Fix typos.
33226 (localize_node): When localizing symbol in same comdat group,
33227 dissolve the group only when we know external symbols are going
33229 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33231 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33233 PR tree-optimization/79390
33234 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33235 order does not result in usable sequence, retry with reversed operand
33241 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33242 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33243 op0 instead of fold_convert_loc (loc, type, arg0).
33245 2017-04-12 Jeff Law <law@redhat.com>
33247 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33248 has a delay slot in the generated code.
33250 * config/cris/cris.md (cris_preferred_reload_class): Return
33251 GENNONACR_REGS rather than GENERAL_REGS.
33253 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33256 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33257 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33258 signedness of the result type.
33260 2017-04-12 Richard Biener <rguenther@suse.de>
33261 Jeff Law <law@redhat.com>
33263 PR tree-optimization/80359
33264 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33265 trim stores to TARGET_MEM_REFs.
33267 2017-04-12 Richard Biener <rguenther@suse.de>
33269 PR tree-optimization/79390
33270 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33271 threading case even more.
33273 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33276 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33277 for quad_address_p for TImode, instead of just not indexed_address.
33279 2017-04-12 Richard Biener <rguenther@suse.de>
33280 Bernd Edlinger <bernd.edlinger@hotmail.de>
33282 PR middle-end/79671
33283 * alias.c (component_uses_parent_alias_set_from): Handle
33284 TYPE_TYPELESS_STORAGE.
33285 (get_alias_set): Likewise.
33286 * tree-core.h (tree_type_common): Add typeless_storage flag.
33287 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33288 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33289 for types containing members with TYPE_TYPELESS_STORAGE.
33290 (place_field): Likewise.
33291 (layout_type): Likewise for ARRAY_TYPE.
33292 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33293 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33294 TYPE_TYPELESS_STORAGE.
33295 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33297 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33300 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33301 first argument to type.
33303 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33307 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33308 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33309 (rs6000_expand_binop_builtin): Likewise.
33310 (rs6000_expand_ternop_builtin): Likewise; also add missing
33311 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33312 vshasigma built-ins.
33313 * doc/extend.texi: Document that vec_xxpermdi's third argument
33314 must be a constant.
33316 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33318 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33319 Use shift_const cost parameter when calculating gain of STV shifts.
33321 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33323 PR rtl-optimization/70478
33324 * lra-constraints.c (process_alt_operands): Check memory for
33325 disfavoring memory insn operand.
33327 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33329 PR middle-end/80100
33330 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33331 left shift in unsigned HOST_WIDE_INT type.
33333 PR rtl-optimization/80385
33334 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33335 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33338 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33339 if they have any depend clauses.
33341 2017-04-11 Martin Liska <mliska@suse.cz>
33344 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33345 * ipa-split.c (split_function): Create a local comdat symbol
33346 if caller is in a comdat group.
33348 2017-04-11 Martin Liska <mliska@suse.cz>
33351 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33354 2017-04-11 Martin Sebor <msebor@redhat.com>
33356 PR middle-end/80364
33357 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33358 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33360 (directive::set_width, directive::set_precision, format_character):
33362 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33365 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33368 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33369 conflict, set target->arch_name instead of target->cpu_name.
33371 2017-04-11 Richard Biener <rguenther@suse.de>
33373 PR tree-optimization/80374
33374 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33375 build_zero_cst, remove fold_convertible_p check again.
33377 2017-04-11 Martin Liska <mliska@suse.cz>
33380 * ubsan.c (instrument_object_size): Do not instrument register
33383 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33386 * config/i386/i386-builtin-types.def
33387 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33388 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33389 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33390 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33391 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33392 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33393 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33394 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33395 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33396 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33397 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33398 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33399 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33400 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33401 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33402 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33403 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33404 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33405 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33406 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33407 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33408 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33409 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33410 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33411 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33412 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33413 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33414 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33415 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33416 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33417 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33418 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33419 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33420 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33421 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33422 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33423 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33424 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33425 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33426 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33427 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33428 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33429 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33430 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33431 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33432 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33434 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33435 flag to second_arg_count, handle 4 argument function type _COUNT
33436 aliases, handle second_arg_count on second argument rather than last.
33438 2017-04-10 Jeff Law <law@redhat.com>
33440 PR tree-optimization/80374
33441 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33442 record anything if we can not convert integer_zero_node to the
33445 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33448 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33449 Enhance special handling given to the TARGET_P9_MINMAX option in
33450 relation to certain other options.
33452 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33454 PR tree-optimization/80153
33455 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33456 remove POINTER_PLUS_EXPR's base part directly, rather than through
33459 2017-04-10 Richard Biener <rguenther@suse.de>
33460 Bin Cheng <bin.cheng@arm.com>
33462 PR tree-optimization/80153
33463 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33464 the first element of pointer type aff_tree. Build result expr in
33466 (add_elt_to_tree): Convert to type unconditionally. Remove other
33467 fold_convert calls.
33468 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33469 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33471 2017-04-10 Richard Biener <rguenther@suse.de>
33473 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33476 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33478 PR rtl-optimization/70478
33479 * lra-constraints.c (curr_small_class_check): New.
33480 (update_and_check_small_class_inputs): New.
33481 (process_alt_operands): Update curr_small_class_check. Disfavor
33482 alternative insn memory operands. Check available regs for small
33485 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33488 * config/mips/mips.opt (-mvirt): Update description.
33489 * doc/invoke.texi (-mvirt): Likewise.
33491 2017-04-10 Richard Biener <rguenther@suse.de>
33493 PR middle-end/80362
33494 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33495 looking for NEGATE_EXPR in -A / -B to A / B folding.
33497 2017-04-10 Martin Liska <mliska@suse.cz>
33499 PR gcov-profile/80224
33500 * gcov.c (print_usage): Fix usage string.
33501 (get_gcov_intermediate_filename): Remove.
33502 (output_gcov_file): Use both for normal and intermediate format.
33503 (generate_results): Do not initialize special file for
33504 intermediate format.
33506 2017-04-10 Richard Biener <rguenther@suse.de>
33508 PR tree-optimization/80304
33509 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33512 2017-04-10 Nathan Sidwell <nathan@acm.org>
33515 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33516 (rs6000_init_builtins): Use it.
33518 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33520 * config/arm/arm.md (<mrc>): Add mode to SET source.
33521 (<mrrc>): Likewise.
33523 2017-04-10 Richard Biener <rguenther@suse.de>
33525 PR middle-end/80344
33526 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33528 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33531 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33532 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33533 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33534 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33535 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33536 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33537 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33538 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33539 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33540 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33541 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33542 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33543 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33544 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33545 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33546 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33547 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33548 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33549 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33550 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33551 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33552 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33553 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33555 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33557 PR rtl-optimization/70478
33558 * lra-constraints.c: Reverse the last patch.
33560 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33562 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33563 Add comment for WCHAR_T.
33565 2017-04-08 Martin Liska <mliska@suse.cz>
33568 2017-04-07 Martin Liska <mliska@suse.cz>
33571 * ipa-split.c (split_function): Add function part to a same comdat
33574 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33577 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33579 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33581 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33582 (vec_pairing): ...to this.
33583 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33584 (rs6000_sched_init): Adjust for name change.
33585 (struct rs6000_sched_context): Likewise.
33586 (rs6000_init_sched_context): Likewise.
33587 (rs6000_set_sched_context): Likewise.
33589 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33595 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33597 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33598 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33599 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33601 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33603 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33605 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33607 PR rtl-optimization/70703
33608 * ira-color.c (update_conflict_hard_regno_costs): Use
33609 int64_t instead of HOST_WIDE_INT.
33611 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33613 PR rtl-optimization/70478
33614 * lra-constraints.c (process_alt_operands): Disfavor alternative
33615 insn memory operands.
33617 2017-04-07 Jeff Law <law@redhat.com>
33619 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33620 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33622 2017-04-07 Martin Liska <mliska@suse.cz>
33625 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33626 Show error message instead of an ICE.
33628 2017-04-07 Martin Liska <mliska@suse.cz>
33631 * ipa-split.c (split_function): Add function part to a same comdat
33634 2017-04-07 Richard Biener <rguenther@suse.de>
33636 PR middle-end/80341
33637 * tree.c (get_unwidened): Also handle ! for_type case for
33639 * convert.c (do_narrow): Split out from ...
33640 (convert_to_integer_1): ... here. Do not pass final truncation
33641 type to get_unwidened for TRUNC_DIV_EXPR.
33643 2017-04-07 Richard Biener <rguenther@suse.de>
33645 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33647 (aff_combination_const): Adjust.
33648 (aff_combination_scale): Likewise.
33649 (aff_combination_add_elt): Likewise.
33650 (aff_combination_add_cst): Likewise.
33651 (aff_combination_convert): Likewise.
33652 (add_elt_to_tree): Likewise. Remove unused argument.
33653 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33655 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33657 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33659 * config/arm/arm.c (arm_default_short_enums): Use
33660 ARM_DEFAULT_SHORT_ENUMS.
33661 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33663 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33666 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33667 members with redundant out-of-class redeclaration.
33669 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33672 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33673 * config/i386/i386.md (*zero_extendsidi2):
33674 Add (?*x,*x) and (?*v,*v) alternatives.
33676 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33679 * config/i386/i386.c (ix86_expand_builtin)
33680 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33681 mode from insn data. Convert operands to insn operand mode.
33682 Copy operands that don't satisfy insn predicate to a register.
33684 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33686 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33689 2017-04-06 Richard Biener <rguenther@suse.de>
33691 PR tree-optimization/80334
33692 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33693 preserve alignment of accesses.
33695 2017-04-06 Richard Biener <rguenther@suse.de>
33697 PR tree-optimization/80262
33698 * tree-sra.c (build_ref_for_offset): Preserve address-space
33700 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33701 Drop useless address-space information on MEM_REF offsets.
33703 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33705 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33707 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33709 PR rtl-optimization/70703
33710 * ira-color.c (update_conflict_hard_regno_costs): Use
33711 HOST_WIDE_INT instead of long.
33713 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33716 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33717 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33719 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33720 for x86_64 target. Handle -m3dnowa option.
33722 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33724 PR rtl-optimization/70703
33725 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33726 (update_conflict_hard_regno_costs): Use long instead of unsigned
33727 arithmetic for cost calculation.
33729 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33730 Bernd Edlinger <bernd.edlinger@hotmail.de>
33733 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33736 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33739 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33740 ptr_mode with Pmode throughout.
33741 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33742 into probe_stack_range and use DImode.
33744 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33747 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33748 call_eh_return is true.
33750 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33752 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33753 Initialize last_match_fntype_index.
33755 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33758 * tree-nvr.c: Include internal-fn.h.
33759 (pass_return_slot::execute): Ignore internal calls without
33762 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33763 Richard Biener <rguenther@suse.de>
33766 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33767 captures used multiple times, except for the last use.
33768 * generic-match-head.c: Include gimplify.h.
33770 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33772 PR tree-optimization/79390
33773 * target.h (struct noce_if_info): Declare.
33774 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33775 * target.def (noce_conversion_profitable_p): New target hook.
33776 * ifcvt.h (struct noce_if_info): New type, moved from ...
33777 * ifcvt.c (struct noce_if_info): ... here.
33778 (noce_conversion_profitable_p): Renamed to ...
33779 (default_noce_conversion_profitable_p): ... this. No longer
33781 (noce_try_store_flag_constants, noce_try_addcc,
33782 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33783 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33784 instead of noce_conversion_profitable_p.
33785 * config/i386/i386.c: Include ifcvt.h.
33786 (ix86_option_override_internal): Don't override
33787 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33788 (ix86_noce_conversion_profitable_p): New function.
33789 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33790 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33791 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33792 * doc/tm.texi: Regenerated.
33794 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33796 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33799 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33802 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33803 instructions for small multiply cores.
33805 2017-04-04 Jeff Law <law@redhat.com>
33807 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33809 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33810 that are not set by the loop over the elements.
33812 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33815 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33816 int mode, convert_modes it to mode as unsigned, otherwise use
33817 lowpart_subreg to mode rather than SImode.
33818 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33819 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33820 Use DImode instead of SImode for the shift count operand.
33821 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33824 2017-04-04 Richard Biener <rguenther@suse.de>
33826 PR middle-end/80281
33827 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33828 arithmetic done for the negate or the plus. Simplify.
33829 (A - (-B) -> A + B): Likewise.
33830 * fold-const.c (split_tree): Make sure to not negate pointers.
33832 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33834 PR rtl-optimization/60818
33835 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33836 a compare of comparisons with the thing compared if this results
33837 in a different machine mode.
33839 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33841 * alias.c (base_alias_check): Fix typo in comment.
33842 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33843 * cgraphunit.c (symbol_table::compile): Likewise.
33844 * collect2.c (maybe_run_lto_and_relink): Likewise.
33845 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33846 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33847 * config/avr/avr.c (avr_map_op_t): Likewise.
33848 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33849 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33850 * config/epiphany/epiphany.md (movcc): Likewise.
33851 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33852 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33854 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33855 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33856 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33857 * config/sh/sh.c (sh_rtx_costs): Likewise.
33858 * fold-const.c (fold_truth_andor): Likewise.
33859 * genautomata.c (collapse_flag): Likewise.
33860 * gengtype.h (struct type::u::s): Likewise.
33861 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
33862 * input.c (FORMAT_AMOUNT): Likewise.
33863 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
33864 (known_aggs_to_agg_replacement_list): Likewise.
33865 * ipa-inline-analysis.c: Likewise.
33866 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
33867 * ipa-polymorphic-call.c
33868 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
33869 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
33870 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
33872 * modulo-sched.c (apply_reg_moves): Likewise.
33873 * omp-expand.c (build_omp_regions_1): Likewise.
33874 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
33875 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
33876 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
33877 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33878 * value-prof.c: Likewise.
33879 * var-tracking.c (val_reset): Likewise.
33881 2017-04-03 Richard Biener <rguenther@suse.de>
33883 PR tree-optimization/80275
33884 * fold-const.c (split_address_to_core_and_offset): Handle
33887 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
33889 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
33890 descriptors is at least equal to that of functions.
33892 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33894 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
33896 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33899 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
33900 (mov<IMOD4:mode>): New expander.
33901 (*mov<IMOD4:mode>_internal): New insn and split pattern.
33903 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
33905 PR rtl-optimization/79405
33906 * fwprop.c (propagations_left): New variable.
33907 (forward_propagate_into): Decrement it.
33908 (fwprop_init): Initialize it.
33909 (fw_prop): If the variable has reached zero, stop propagating.
33910 (fwprop_addr): Ditto.
33912 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33915 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
33916 a FUNCTION_DECL, pass it as decl instead of origin to
33919 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
33921 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
33924 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
33927 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
33928 TARGET_VSX_SMALL_INTEGER.
33930 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33932 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
33933 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
33935 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33937 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
33938 extraction from odd-numbered MSA register.
33940 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33942 PR middle-end/80173
33943 * expmed.c (store_bit_field_1): Don't attempt to create
33944 a word subreg out of hard registers wider than word if they
33945 have HARD_REGNO_NREGS of 1 for their mode.
33947 PR middle-end/80163
33948 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
33949 conversions to integer types wider than word and pointer.
33952 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
33953 (rtx_equal_for_cselib_p): Pass 0 to it.
33954 * cselib.c (cselib_hasher::equal): Likewise.
33955 (rtx_equal_for_cselib_1): Add depth argument. If depth
33956 is 128, don't look up VALUE locs and punt. Increment
33957 depth in recursive calls when walking VALUE locs.
33959 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
33961 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
33962 (make_gcov_file_name): Use the canonical path name for generating
33964 (read_line): Fix handling of files with ascii null bytes.
33966 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
33968 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
33969 to initialise a vector register instead
33970 of using a const_int.
33972 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33974 PR translation/80189
33975 * gimplify.c (omp_default_clause): Use %qs instead of %s in
33976 diagnostic messages.
33978 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
33981 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
33982 (dfp_diex_<mode>): Update mode of operand 1.
33983 * doc/extend.texi (dxex, dxexq): Document change to return type.
33984 (diex, diexq): Document change to argument type.
33986 2017-03-30 Martin Jambor <mjambor@suse.cz>
33989 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
33990 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
33991 it reflects the signature changes performed at the callee side.
33992 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
33993 to cgraph_build_function_type_skip_args.
33994 (build_function_decl_skip_args): Adjust call to the above function.
33996 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33999 * config/i386/sse.md
34000 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34001 register as dest whenever it is a MEM not rtx_equal_p to the
34002 corresponding dup operand, and when forcing into reg move the
34003 reg into the memory afterwards.
34004 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34005 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34006 for the force_reg mode.
34007 (avx512vl_vextractf128<mode>): Use register as dest either
34008 always when a MEM, or when it is a MEM not rtx_equal_p to the
34009 corresponding dup operand, or even not when it is a CONST_VECTOR
34010 depending on the mode and lo vs. hi.
34011 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34013 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34014 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34015 Likewise. Require that operands[2] is even.
34016 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34017 Remove extraneous parens. Require that operands[2] is a multiple
34019 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34020 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34021 disallow memory then.
34023 2017-03-30 Richard Biener <rguenther@suse.de>
34025 PR tree-optimization/77498
34026 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34027 to non-constants over backedges.
34029 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34031 PR rtl-optimization/80233
34032 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34033 as last_combined_insn. Do not test for BARRIER_P separately.
34035 2017-03-29 Andreas Schwab <schwab@suse.de>
34038 * calls.c (prepare_call_address): Convert funexp to Pmode before
34039 copying to temp reg.
34041 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34043 PR tree-optimization/80158
34044 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34045 Handle possible future case of more than one alternate
34047 (replace_rhs_if_not_dup): Likewise.
34048 (replace_one_candidate): Likewise.
34050 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34052 PR rtl-optimization/80193
34053 * ira.c (ira): Do not check allocation for LRA.
34055 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34057 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34058 (nvptx_output_simt_exit): Declare.
34059 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34060 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34061 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34062 (nvptx_declare_function_name): ...here. Emit declaration of local
34063 memory space buffer for omp_simt_enter insn.
34064 (nvptx_output_unisimt_switch): New.
34065 (nvptx_output_softstack_switch): New.
34066 (nvptx_output_simt_enter): New.
34067 (nvptx_output_simt_exit): New.
34068 * config/nvptx/nvptx.h (struct machine_function): New fields
34069 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34070 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34071 (UNSPECV_SIMT_EXIT): Ditto.
34072 (omp_simt_enter_insn): New insn.
34073 (omp_simt_enter): New expansion.
34074 (omp_simt_exit): New insn.
34075 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34077 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34078 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34079 (expand_GOMP_SIMT_EXIT): New.
34080 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34081 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34082 (GOMP_SIMT_EXIT): Ditto.
34083 * target-insns.def (omp_simt_enter): New insn.
34084 (omp_simt_exit): Ditto.
34085 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34087 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34088 (lower_rec_input_clauses): Likewise.
34089 (lower_lastprivate_clauses): Handle SIMT privatization.
34091 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34092 (ompdevlow_adjust_simt_enter): New.
34093 (find_simtpriv_var_op): New.
34094 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34095 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34097 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34098 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34099 (copy_decl_for_dup_finish): Ditto.
34101 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34103 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34106 * config/i386/i386.c (ix86_option_override_internal): Always
34107 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34109 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34111 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34113 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34115 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34116 mark new edge's irreducible flag accordign to it.
34117 (vect_do_peeling): Check loop preheader edge's irreducible flag
34118 and pass it to function slpeel_add_loop_guard.
34120 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34122 PR tree-optimization/80218
34123 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34124 Update block frequencies and counts.
34126 2017-03-28 Richard Biener <rguenther@suse.de>
34128 PR tree-optimization/78644
34129 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34130 of a simplification result we may not use it at all.
34132 2017-03-28 Richard Biener <rguenther@suse.de>
34135 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34136 without arguments, generate default definition of a SSA name.
34138 2017-03-28 Richard Biener <rguenther@suse.de>
34140 PR middle-end/80222
34141 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34142 TYPE_REF_CAN_ALIAS_ALL references.
34143 * fold-const.c (fold_indirect_ref_1): Likewise.
34145 2017-03-28 Martin Liska <mliska@suse.cz>
34148 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34149 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34151 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34152 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34154 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34155 (EXTRA_SPECS): Define.
34156 (SUBTARGET_EXTRA_SPECS): Likewise.
34157 (SUBTARGET_CPP_SPEC): Likewise.
34158 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34159 SUBTARGET_EXTRA_SPECS.
34160 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34162 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34164 * config/arc/simdext.md (vst64_insn): Update pattern.
34165 (vld32wh_insn): Likewise.
34166 (vld32wl_insn): Likewise.
34167 (vld64_insn): Likewise.
34168 (vld32_insn): Likewise.
34170 2017-03-28 Marek Polacek <polacek@redhat.com>
34173 * fold-const.c (fold_comparison): Use protected_set_expr_location
34174 instead of SET_EXPR_LOCATION.
34176 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34178 * tree.c (add_expr): Avoid name lookup warning.
34180 2017-03-27 Jeff Law <law@redhat.com>
34182 PR tree-optimization/80216
34183 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34184 function name. Limit recursion depth.
34185 (record_temporary_equivalences): Corresponding changes.
34187 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34189 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34192 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34195 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34197 * cfgcleanup.c (reg_note_cfa_p): New array.
34198 (insns_have_identical_cfa_notes): New function.
34199 (old_insns_match_p): Don't cross-jump in between /f
34200 and non-/f instructions. If both i1 and i2 are frame related,
34201 verify all CFA notes, their order and content.
34203 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34206 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34207 HImode and SImode with zero extend to DImode to one insn.
34208 (bswap<mode>2_extenddi): Likewise.
34209 (bswapsi2_extenddi): Likewise.
34210 (bswaphi2_extendsi): Likewise.
34211 (bswaphi2): Combine bswap HImode and SImode into one insn.
34212 Separate memory insns from swapping register.
34213 (bswapsi2): Likewise.
34214 (bswap<mode>2): Likewise.
34215 (bswaphi2_internal): Delete, no longer used.
34216 (bswapsi2_internal): Likewise.
34217 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34218 store, and gpr<-gpr swap insns.
34219 (bswap<mode>2_store): Likewise.
34220 (bswaphi2_reg): Register only splitter, combine with the splitter.
34221 (bswaphi2 splitter): Likewise.
34222 (bswapsi2_reg): Likewise.
34223 (bswapsi2 splitter): Likewise.
34224 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34225 the insns into load, store, and register/register insns.
34226 (bswapdi2_ldbrx): Likewise.
34227 (bswapdi2_load): Likewise.
34228 (bswapdi2_store): Likewise.
34229 (bswapdi2_reg): Likewise.
34231 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34233 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34234 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34236 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34239 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34241 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34242 special handling for target option conflicts between dform
34243 options (-mpower9-dform, -mpower9-dform-vector,
34244 -mpower9-dform-scalar) and -mno-direct-move.
34246 2017-03-27 Richard Biener <rguenther@suse.de>
34248 PR tree-optimization/80181
34249 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34251 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34253 * config/arc/predicates.md (move_double_src_operand): Replace the
34254 call to move_double_src_operand with a call to address_operand.
34256 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34258 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34259 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34260 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34262 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34264 * config/arc/predicates.md (long_immediate_loadstore_operand):
34265 Consider scaled addresses cases.
34267 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34269 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34270 restored when in interrupt.
34271 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34272 doesn't have delay slot.
34274 2017-03-27 Richard Biener <rguenther@suse.de>
34277 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34278 inlined thunk clones.
34280 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34283 * asan.c (instrument_derefs): Copy over last operand from
34284 original COMPONENT_REF to the new COMPONENT_REF with
34285 DECL_BIT_FIELD_REPRESENTATIVE.
34286 * ubsan.c (instrument_object_size): Likewise.
34288 2017-03-27 Richard Biener <rguenther@suse.de>
34290 PR tree-optimization/80170
34291 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34292 sure DR/SCEV didnt fold in constants we do not see when looking
34293 at the reference base alignment.
34295 2017-03-27 Richard Biener <rguenther@suse.de>
34297 PR middle-end/80171
34298 * gimple-fold.c (fold_ctor_reference): Properly guard against
34299 NULL return value from canonicalize_constructor_val.
34301 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34304 * config/i386/i386.c (ix86_expand_builtin)
34305 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34306 flags reg setting and flags reg using instructions.
34307 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34308 clobbering instructions to zero extend op2.
34310 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34312 * doc/install.texi (Configuration) <--with-aix-soname>:
34313 Update link to AIX ld.
34315 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34317 PR rtl-optimization/80160
34318 PR rtl-optimization/80159
34319 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34320 reg_alternate_class into account.
34322 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34325 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34326 to consider in curr_insn_transform.
34328 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34330 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34331 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34332 and emit_mode_inner.
34334 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34336 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34337 argument to the overloaded builtin variants. Use the new flag to
34338 deprecate certain builtin variants.
34339 * config/s390/s390-builtin-types.def: Add new builtin types.
34340 * config/s390/s390-builtins.h: Support new flags field for
34341 overloaded builtins.
34342 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34343 (s390_macro_to_expand): Enable vector float data type.
34344 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34345 builtins by incrementing the __VEC__ version number.
34346 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34348 (s390_resolve_overloaded_builtin): Emit error messages depending
34349 on the builtin flags.
34350 * config/s390/s390.c (s390_expand_builtin): Support additional
34351 flags argument. Change error message to match the messages
34352 emitted in s390-c.c.
34353 * config/s390/s390.md: New UNSPEC_* constants.
34354 (op_type): Add new instruction types.
34355 * config/s390/vecintrin.h: Add new builtins and test data class
34357 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34358 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34359 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34360 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34361 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34362 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34364 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34365 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34366 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34367 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34369 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34370 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34371 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34372 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34374 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34375 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34376 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34377 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34378 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34379 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34380 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34382 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34383 ("vec_scatter_element<V_HW_4:mode>_DI")
34384 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34385 ("vec_fpint<mode>", "vflls")
34386 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34387 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34388 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34389 ("*vec_cmphe<mode>_cc"): ... these.
34391 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34392 mode constant instead of magic value.
34394 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34396 * config/s390/s390.c (s390_expand_vec_compare): Support other
34397 vector floating point modes than just V2DF.
34398 (s390_expand_vcond): Likewise.
34399 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34400 (s390_cannot_change_mode_class): Prevent mode changes between TF
34401 and V1TF in vector registers.
34402 * config/s390/s390.md (DF, SF): New mode attributes.
34403 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34404 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34405 SFmode support for VRs.
34406 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34408 (VFT, VF_HW): New mode iterators.
34409 (vw, sdx): New mode attributes.
34410 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34411 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34412 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34413 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34414 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34415 also the new vector floating point modes. Renaming to ...
34417 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34418 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34419 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34420 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34421 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34422 ("vec_unordered<mode>"): ... these.
34424 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34425 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34426 ("*vec_extendv2df"): New insn definitions.
34428 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34430 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34431 ("mulditi3_2", "*muldi3_sign"): New patterns.
34432 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34433 rename the pattern definition.
34435 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34437 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34439 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34441 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34443 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34444 instruction if possible.
34445 * config/s390/vector.md (vec_halfnumelts): New mode
34447 ("*vec_vllezlf<mode>"): New pattern.
34449 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34451 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34452 ("popcountv4si2", "popcountv2di2"): Rename to ...
34453 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34454 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34456 ("popcount<mode>2_vxe"): New pattern.
34458 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34460 * common/config/s390/s390-common.c (processor_flags_table): Add
34462 * config.gcc: Add arch12.
34463 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34464 Default to arch12 for unknown CPU model numbers.
34465 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34466 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34467 PROCESSOR_max sanity check.
34468 * config/s390/s390-opts.h (enum processor_type): Add
34470 * config/s390/s390.c (processor_table): Add arch12.
34471 (s390_expand_builtin): Add check for B_VXE flag.
34472 (s390_issue_rate): Add PROCESSOR_ARCH12.
34473 (s390_get_sched_attrmask): Likewise.
34474 (s390_get_unit_mask): Likewise.
34475 (s390_sched_score): Enable z13 scheduling for arch12.
34476 (s390_sched_reorder): Likewise.
34477 (s390_sched_variable_issue): Likewise.
34478 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34480 (s390_tune_attr): Use z13 scheduling also for arch12.
34481 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34482 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34483 (TARGET_VXE_P): New macros.
34484 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34485 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34486 * config/s390/s390.opt: Add arch12 as processor_type.
34488 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34490 * config/s390/s390.md
34491 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34492 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34493 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34495 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34496 Rename expanders to ...
34498 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34499 ("fixuns_truncdddi2_emu"): ... these.
34501 ("fixuns_trunc<mode>si2_emu"): New expander.
34503 ("*fixuns_truncdfdi2_z13"): Rename to ...
34504 ("*fixuns_truncdfdi2_vx"): ... this.
34506 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34508 * config/s390/2964.md: Remove the single element vector compare
34509 instructions which are no longer used.
34510 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34512 (s390_canonicalize_comparison): Remove handling of DFmode
34514 (s390_expand_vec_compare_scalar): Remove function.
34515 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34516 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34518 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34520 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34522 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34524 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34525 will anyway by matched by mov<mode>_64dfp.
34527 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34529 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34530 vlef/vstef. Add missing operand to vleif.
34532 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34534 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34535 pair for all vector types with 64 bit elements.
34536 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34537 * config/s390/vector.md (V_HW_64): ... here.
34538 (V_128_NOSINGLE): New mode iterator.
34539 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34540 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34541 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34542 ("*vec_load_pairv2di"): Change to ...
34543 ("*vec_load_pair<mode>"): ... this one.
34545 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34547 * config/s390/constraints.md: Add comments.
34548 (jKK): Reject element sizes > 8 bytes.
34549 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34551 * config/s390/s390.md: Add the s_operand checks formerly in
34552 s390_split_ok_p to various splitters where they are still
34554 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34555 for 128 bit vectors. Plus two splitters.
34557 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34559 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34562 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34565 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34566 error if the boundary argument is not constant.
34568 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34570 PR rtl-optimization/80112
34571 * loop-doloop.c (doloop_condition_get): Don't check condition
34572 if cmp isn't SET with IF_THEN_ELSE src.
34574 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34576 PR tree-optimization/80158
34577 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34578 replacing a candidate statement, also replace it for the
34579 candidate's alternate interpretation.
34580 (replace_rhs_if_not_dup): Likewise.
34581 (replace_one_candidate): Likewise.
34583 2017-03-24 Richard Biener <rguenther@suse.de>
34585 PR tree-optimization/80167
34586 * graphite-isl-ast-to-gimple.c
34587 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34589 (translate_isl_ast_to_gimple::get_rename): Likewise.
34591 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34593 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34594 handling of certain combinations of target options, including the
34595 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34596 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34598 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34601 * config/arm/arm.md (*load_multiple): Add reload_completed to
34602 matching condition.
34604 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34605 Richard Biener <rguenth@suse.de>
34607 PR tree-optimization/79908
34608 PR tree-optimization/80136
34609 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34610 been cast away, gimplify_and_add suffices.
34612 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34614 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34616 2017-03-23 Richard Biener <rguenther@suse.de>
34618 PR tree-optimization/80032
34619 * gimplify.c (gimple_push_cleanup): Forced unconditional
34620 cleanups still have to go to the conditional_cleanups
34623 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34625 PR tree-optimization/80072
34626 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34628 (next_operand_entry_id): Change type to unsigned int.
34629 (sort_by_operand_rank): Make sure to return the right return value
34630 even if unsigned fields are bigger than INT_MAX.
34631 (struct oecount): Change cnt and id type to unsigned int.
34632 (oecount_hasher::equal): Formatting fix.
34633 (oecount_cmp): Make sure to return the right return value
34634 even if unsigned fields are bigger than INT_MAX.
34635 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34638 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34639 TREE_READONLY on result if writing it more than once.
34642 * doc/invoke.texi (-fsanitize=thread): Document that with
34643 -fnon-call-exceptions atomics are not able to throw
34647 * tsan.c: Include tree-eh.h.
34648 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34649 maybe_clean_or_replace_eh_stmt where needed.
34650 (instrument_memory_accesses): Add cfg_changed argument.
34651 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34652 if it returned true.
34653 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34655 PR rtl-optimization/63191
34656 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34657 wrapper function, moved the whole old content into ...
34658 (ix86_delegitimize_address_1): ... this. New inline function.
34659 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34660 true as last argument instead of ix86_delegitimize_address.
34662 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34664 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34665 cortexa57_branch_cost.
34667 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34669 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34671 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34674 * doc/md.texi (Constraints): Document wA constraint.
34675 * config/rs6000/constraints.md (wA): New.
34676 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34677 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34678 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34679 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34681 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34684 * gimplify.c (is_oacc_declared): New function.
34685 (oacc_default_clause): Use it to set default flags for acc declared
34686 variables inside parallel regions.
34687 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34688 declared variables.
34689 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34690 declare attribute to any decl as necessary.
34692 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34695 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34696 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34697 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34698 (arm_arch_lpae): This.
34699 * config/arm/arm.c (arm_arch7ve): Rename into ...
34700 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34701 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34704 2017-03-22 Martin Liska <mliska@suse.cz>
34707 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34708 error message instead of an ICE.
34710 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34712 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34714 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34716 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34718 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34721 2017-03-21 Martin Sebor <msebor@redhat.com>
34723 * doc/extend.texi: Use "cannot" instead of "can't."
34724 * doc/hostconfig.texi: Same.
34725 * doc/install.texi: Same.
34726 * doc/invoke.texi: Same.
34727 * doc/loop.texi: Same.
34728 * doc/md.texi: Same.
34729 * doc/objc.texi: Same.
34730 * doc/rtl.texi: Same.
34731 * doc/tm.texi: Same.
34732 * doc/tm.texi.in: Same.
34733 * doc/trouble.texi: Same.
34735 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34738 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34739 (collect_checksum_attributes): Set it.
34740 (die_checksum_ordered): Use it.
34742 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34744 PR tree-optimization/79908
34745 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34746 change: For a VA_ARG whose LHS has been cast away, use
34747 force_gimple_operand to construct the side effects.
34749 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34751 PR translation/80001
34752 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34753 more amenable to translation.
34754 (oacc_loop_auto_partitions): Likewise.
34756 2017-03-21 Marek Polacek <polacek@redhat.com>
34757 Martin Sebor <msebor@redhat.com>
34759 PR tree-optimization/80109
34760 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34761 on INTEGRAL_TYPE_P.
34763 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34764 Segher Boessenkool <segher@kernel.crashing.org>
34767 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34768 check reg_used_between_p between insn and one of succ or succ2
34769 depending on if succ is artificial insn not inserted into insn
34772 2017-03-21 Martin Liska <mliska@suse.cz>
34774 PR gcov-profile/80081
34775 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34776 * doc/gcc.texi: Include gcov-dump stuff.
34777 * doc/gcov-dump.texi: New file.
34779 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34781 PR rtl-optimization/79150
34782 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34783 conditional jump, if the jump is the last insn of the loop.
34785 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34786 Richard Biener <rguenth@suse.de>
34788 PR tree-optimization/79908
34789 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34790 been cast away, use force_gimple_operand to construct the side
34793 2017-03-21 Martin Liska <mliska@suse.cz>
34795 PR libfortran/79956
34796 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34799 2017-03-21 Brad Spengler <spender@grsecurity.net>
34802 * plugin.c (htab_hash_plugin): New function.
34803 (add_new_plugin): Use it and adjust.
34804 (parse_plugin_arg_opt): Adjust.
34805 (init_one_plugin): Likewise.
34807 2017-03-21 Richard Biener <rguenther@suse.de>
34809 PR tree-optimization/80032
34810 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34811 if set force the cleanup to happen unconditionally.
34812 (gimplify_target_expr): Push inserted clobbers with force_uncond
34813 to avoid them being removed by control-dependent DCE.
34815 2017-03-21 Richard Biener <rguenther@suse.de>
34817 PR tree-optimization/80122
34818 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34819 va_arg_pack_len when the inlined call stmt requires pack
34821 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34823 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34826 * tsan.c (instrument_builtin_call): If the memory model argument
34827 is not a constant, assume it is valid.
34830 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34833 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34835 PR rtl-optimization/79910
34836 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34837 if its dest is used by an insn before I2 (other than the combined
34838 insns themselves, which are properly handled already).
34840 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34843 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34845 * combine.c (record_used_regs): New static function.
34846 (try_combine): Handle situations where there is an additional
34847 instruction between I2 and I3 which needs to have a LOG_LINK
34851 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34853 * combine.c (try_combine): Delete redundant i1 test. Call
34854 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34856 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34859 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
34860 alternatives 13/14.
34862 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34864 PR tree-optimization/80054
34865 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
34866 the optimization if a PHI or any of its arguments is not dominated
34867 by the candidate's basis. Use gphi* rather than gimple* as
34869 (replace_profitable_candidates): Clean up a gimple* variable that
34870 should be a gphi* variable.
34872 2017-03-20 Martin Sebor <msebor@redhat.com>
34875 * doc/extend.texi (attribute constructor): Document present limitation.
34877 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
34880 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
34881 __POWER9_VECTOR__ #ifdef control, change template definition to
34882 use Power9-specific built-in function.
34883 (vec_any_eq): Likewise.
34884 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
34885 to control outcomes from this test.
34886 (vector_ae_<mode>p): For VEC_F modes, likewise.
34888 2017-03-20 Ian Lance Taylor <iant@google.com>
34890 * config/i386/i386.c (ix86_function_regparm): Save an extra
34891 register for -fsplit-stack with DECL_STATIC_CHAIN.
34893 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34896 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
34897 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
34899 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34901 * config/riscv/riscv.c (riscv_print_operand): Use "fence
34903 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
34906 2017-03-20 Marek Polacek <polacek@redhat.com>
34909 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
34911 2017-03-20 Richard Biener <rguenther@suse.de>
34913 PR tree-optimization/80113
34914 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
34915 allocate extra SSA name for PHI def.
34916 (add_close_phis_to_outer_loops): Likewise.
34917 (add_close_phis_to_merge_points): Likewise.
34918 (copy_loop_close_phi_args): Likewise.
34919 (copy_cond_phi_nodes): Likewise.
34921 2017-03-20 Martin Liska <mliska@suse.cz>
34923 PR middle-end/79753
34924 * tree-chkp.c (chkp_build_returned_bound): Do not build
34925 returned bounds for a LHS that's not a BOUNDED_P type.
34927 2017-03-20 Martin Liska <mliska@suse.cz>
34931 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
34932 COMPLEX_CST and VECTOR_CST.
34934 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34937 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
34938 target operand. A new splitter adds the clobber statement in case
34939 the target operand is dead anyway.
34941 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
34943 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
34944 to age-old versions of binutils and glibc.
34946 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
34948 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
34950 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34952 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
34954 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34956 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
34957 requirement for binutils 2.13.
34959 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34961 * combine.c (try_combine): Delete redundant i1 test. Call
34962 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34964 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
34966 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
34967 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
34969 <riscv64-*-elf>: Re-arrange section
34970 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
34971 <riscv32-*-linux>: Likewise.
34972 <riscv64-*-elf>: Likewise
34973 <riscv64-*-linux>: Likewise.
34975 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
34978 * aarch64.opt(verbose-cost-dump): Fix typo.
34980 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
34983 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
34984 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
34986 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34988 * reload.c (find_reloads): When reloading a nonoffsettable address,
34989 use RELOAD_OTHER for it and its address reloads.
34991 PR rtl-optimization/79910
34992 * combine.c (record_used_regs): New static function.
34993 (try_combine): Handle situations where there is an additional
34994 instruction between I2 and I3 which needs to have a LOG_LINK
34997 2017-03-17 Jeff Law <law@redhat.com>
34999 PR tree-optimization/71437
35000 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35001 conditional in the hash table first.
35002 (vrp_dom_walker::before_dom_children): Extract condition from
35003 ASSERT_EXPR. Record condition, its inverion and any implied
35004 conditions as well.
35006 2017-03-17 Marek Polacek <polacek@redhat.com>
35007 Markus Trippelsdorf <markus@trippelsdorf.de>
35009 PR tree-optimization/80079
35010 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35013 2017-03-17 Richard Biener <rguenther@suse.de>
35015 PR middle-end/80075
35016 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35017 Properly verify the LHS before the RHS possibly claims to be
35019 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35022 2017-03-17 Martin Jambor <mjambor@suse.cz>
35024 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35025 (List of -O2 options): Likewise.
35026 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35029 2017-03-17 Tom de Vries <tom@codesourcery.com>
35031 * gcov-dump.c (print_usage): Print bug_report_url.
35033 2017-03-17 Richard Biener <rguenther@suse.de>
35035 PR middle-end/80050
35036 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35037 (parser::peek): Likewise.
35039 2017-03-17 Richard Biener <rguenther@suse.de>
35041 PR tree-optimization/80048
35042 * sese.c (free_sese_info): Properly release rename_map and
35043 copied_bb_map elements.
35045 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35047 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35048 Add linked-list forward and backlinks. Insert on
35049 construction, remove on destruction.
35050 (class pass_store_merging): Add m_stores_head field.
35051 (pass_store_merging::terminate_and_process_all_chains):
35052 Iterate over m_stores_head list.
35053 (pass_store_merging::terminate_all_aliasing_chains):
35055 (pass_store_merging::execute): Check for debug stmts first.
35056 Push new chains onto the m_stores_head stack.
35058 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35061 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35062 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35063 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35065 2017-03-16 Jeff Law <law@redhat.com>
35067 PR tree-optimization/71437
35068 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35069 member function. Implementation moved into after_dom_children
35070 member function and into the threader's thread_outgoing_edges
35072 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35073 some code into new thread_outgoing_edges.
35074 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35075 definition. Simplify marker handling (do it here). Assume we always
35076 have the available expression and the const/copies tables.
35077 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35079 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35080 * tree-vrp.c (equiv_stack): No longer file scoped.
35081 (vrp_dom_walker): New class.
35082 (vrp_dom_walker::before_dom_children): New member function.
35083 (vrp_dom_walker::after_dom_children): Likewise.
35084 (identify_jump_threads): Setup domwalker. Use it rather than
35085 walking edges in a random order by hand. Simplify setup/finalization.
35086 (finalize_jump_threads): Remove.
35087 (vrp_finalize): Do not call identify_jump_threads here.
35088 (execute_vrp): Do it here instead and call thread_through_all_blocks
35091 PR tree-optimization/71437
35092 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35094 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35096 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35097 (dom_opt_dom_walker::thread_across_edge): Remove
35098 handle_dominating_asserts argument. All callers changed.
35099 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35100 changes. Remove calls to lhs_of_dominating_assert. Other
35101 uses of handle_dominating_asserts turn into unconditional code
35102 (simplify_control_stmt_condition_1): Likewise.
35103 (simplify_control_stmt_condition): Likewise.
35104 (thread_through_normal_block, thread_across_edge): Likewise.
35105 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35106 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35107 object if it is not an SSA_NAME.
35108 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35109 before calling into the VRP specific simplifiers.
35110 (identify_jump_threads): Remove handle_dominating_asserts
35113 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35116 * tree-diagnostic.c (default_tree_printer): No longer static.
35117 * tree-diagnostic.h (default_tree_printer): New prototype.
35119 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35121 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35122 Change ins into fmov.
35124 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35126 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35127 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35128 Use h_con constraint for operand 1.
35129 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35130 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35132 2017-03-15 Jeff Law <law@redhat.com>
35134 PR tree-optimization/71437
35135 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35136 (record_temporary_equivalences): Use it.
35138 PR tree-optimization/71437
35139 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35140 tree-ssa-scopedtables.
35141 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35142 (record_conditions, record_cond, vuse_eq): Likewise.
35143 (record_edge_info): Adjust to API tweak of record_conditions.
35144 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35145 (record_temporary_equivalences, optimize_stmt): Likewise.
35146 (eliminate_redundant_computations): Likewise.
35147 (record_equivalences_from_stmt): Likewise.
35148 * tree-ssa-scopedtables.c: Include options.h and params.h.
35149 (vuse_eq): New function, moved from tree-ssa-dom.c
35150 (build_and_record_new_cond): Likewise.
35151 (record_conditions): Likewise. Accept vector of conditions rather
35152 than edge_equivalence structure for first argument.
35153 for the first argument.
35154 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35155 from tree-ssa-dom.c.
35156 (avail_exprs_stack::record_cond): Likewise.
35157 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35158 from tree-ssa-dom.c.
35159 (avail_exprs_stack): Add new member functions lookup_avail_expr
35161 (record_conditions): Declare.
35163 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35166 * lra-constraints.c (process_alt_operands): Increase reject for
35167 reloading an input/output operand.
35169 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35172 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35173 insns to convert from signed/unsigned char/short to IEEE 128-bit
35175 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35177 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35180 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35181 subreg of inner mode for values already in registers.
35183 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35185 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35186 iteration reg is used after the loop.
35188 2017-03-14 Martin Sebor <msebor@redhat.com>
35190 PR tree-optimization/79800
35191 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35192 precision in negative-positive range.
35193 (format_floating): Call non-const overload with adjusted precision.
35195 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35198 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35201 2017-03-14 Martin Sebor <msebor@redhat.com>
35203 PR middle-end/80020
35204 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35205 * builtins.def (aligned_alloc): Use it.
35208 * Makefile.in (GTFILES): Add calls.c.
35209 * calls.c: Include "gt-calls.h".
35211 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35213 PR rtl-optimization/79728
35214 * regs.h (struct target_regs): New field
35215 x_contains_allocatable_regs_of_mode.
35216 (contains_allocatable_regs_of_mode): New macro.
35217 * reginfo.c (init_reg_sets_1): Initialize it, and change
35218 contains_reg_of_mode so it includes global regs as well.
35219 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35220 rather than contains_regs_of_mode.
35222 2017-03-14 Martin Liska <mliska@suse.cz>
35224 * doc/invoke.texi: Document options that can't be combined with
35225 -fcheck-pointer-bounds.
35227 2017-03-14 Martin Liska <mliska@suse.cz>
35229 PR middle-end/79831
35230 * doc/invoke.texi (-Wchkp): Document the option.
35232 2017-03-14 Martin Liska <mliska@suse.cz>
35234 * Makefile.in: Install gcov-dump.
35236 2017-03-14 Martin Liska <mliska@suse.cz>
35238 * multiple_target.c (expand_target_clones): Bail out for
35239 an invalid attribute.
35241 2017-03-14 Richard Biener <rguenther@suse.de>
35243 * alias.c (struct alias_set_entry): Pack properly.
35244 * cfgloop.h (struct loop): Likewise.
35245 * cse.c (struct set): Likewise.
35246 * ipa-utils.c (struct searchc_env): Likewise.
35247 * loop-invariant.c (struct invariant): Likewise.
35248 * lra-remat.c (struct cand): Likewise.
35249 * recog.c (struct change_t): Likewise.
35250 * rtl.h (struct address_info): Likewise.
35251 * symbol-summary.h (function_summary): Likewise.
35252 * tree-loop-distribution.c (struct partition): Likewise.
35253 * tree-object-size.c (struct object_size_info): Likewise.
35254 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35255 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35256 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35257 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35258 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35259 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35260 (struct _stmt_vec_info): Likewise.
35262 2017-03-14 Martin Liska <mliska@suse.cz>
35265 * multiple_target.c (create_dispatcher_calls): Check that
35266 a target can create a function dispatcher.
35268 2017-03-14 Martin Liska <mliska@suse.cz>
35271 * multiple_target.c (expand_target_clones): Drop local.local
35272 flag for default implementation.
35274 2017-03-14 Richard Biener <rguenther@suse.de>
35276 PR tree-optimization/80030
35277 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35279 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35281 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35282 gcc_fallthrough() instead of __attribute__((fallthrough));
35284 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35286 * doc/gcc.texi: Remove "up" link to (DIR).
35287 * doc/gccint.texi: Ditto.
35289 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35291 * doc/install.texi (Specific) <avr>: Remove reference to
35294 2017-03-13 Jeff Law <law@redhat.com>
35296 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35297 attribute rather than comments.
35299 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35300 match_scratch operand is highest.
35302 2017-03-13 Martin Liska <mliska@suse.cz>
35304 PR middle-end/78339
35305 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35306 is a CHKP clone, use original declaration.
35308 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35310 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35311 (arc_conditional_register_usage): Use a different allocation order
35312 when optimizing for size.
35313 * common/config/arc/arc-common.c (arc_option_optimization_table):
35314 Section anchors default on when optimizing for size.
35316 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35318 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35320 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35322 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35323 * config/arc/arc.md (cpu_facility): Add cd variant.
35324 (*movqi_insn): Add code density variant.
35325 (*movhi_insn): Likewise.
35326 (*movqi_insn): Likewise.
35327 (*addsi3_mixed): Likewise.
35328 (subsi3_insn): Likewise.
35330 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35332 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35334 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35336 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35337 expressions with MINUS and UNARY ops.
35339 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35342 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35344 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35345 between vec_select and vector argument.
35346 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35347 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35348 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35349 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35350 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35351 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35353 2017-03-13 Richard Biener <rguenther@suse.de>
35356 * params.def (vect-max-peeling-for-alignment): Fix typo.
35358 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35360 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35361 issue that only occurred with binutils below 2.18.
35363 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35365 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35366 refer to binutils 2.11/2.12 minimum.
35368 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35370 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35371 ftp.kernel.org and simplify binutils requirement.
35373 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35375 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35377 (Optimize Options): Ditto. Also remove redundancy.
35379 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35381 PR translation/79848
35382 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35384 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35385 to G_ to avoid double translation.
35387 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35389 PR translation/79923
35390 * auto-profile.c (get_combined_location): Convert leading
35391 character of diagnostics to lower case and remove trailing period.
35392 (read_profile): Likewise for various diagnostics.
35393 * config/arm/arm.c (arm_option_override): Remove trailing period
35394 from various diagnostics.
35395 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35396 (msp430_expand_delay_cycles): Likewise.
35398 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35401 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35402 full command-line argument, rather than just "str".
35403 (aarch64_validate_march): Likewise.
35404 (aarch64_validate_mtune): Likewise.
35406 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35408 PR rtl-optimization/78911
35409 * lra-assigns.c (must_not_spill_p): New function.
35410 (spill_for): Use it.
35412 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35414 PR tree-optimization/79981
35415 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35416 ATOMIC_COMPARE_EXCHANGE ifn result.
35417 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35418 IFN_ATOMIC_COMPARE_EXCHANGE.
35420 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35423 * opts.c (parse_sanitizer_options): Add missing question mark to
35424 "did you mean" message.
35426 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35428 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35430 (VMULEUH_UNS): Likewise.
35431 (VMULOUB_UNS): Likewise.
35432 (VMULOUH_UNS): Likewise.
35433 * config/rs6000/rs6000.c (builtin_function_type): Remove
35434 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35436 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35439 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35440 x with result of extra_parsing_for_operand_code_0.
35441 (function_reader::extra_parsing_for_operand_code_0): Convert
35442 return type from void to rtx, returning x. When reading
35443 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35444 larger size containing struct block_symbol.
35446 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35448 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35449 -mfloat128-hardware without -m64.
35451 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35454 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35455 entries to the case statement that marks unsigned arguments to
35456 overloaded functions.
35458 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35460 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35461 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35463 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35466 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35467 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35469 2017-03-10 Martin Liska <mliska@suse.cz>
35473 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35474 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35477 2017-03-10 Olivier Hainque <hainque@adacore.com>
35479 * tree-switch-conversion (array_value_type): Start by resetting
35480 candidate type to it's main variant.
35482 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35484 PR rtl-optimization/79909
35485 * combine.c (try_combine): Use simplify_replace_rtx on individual
35486 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35487 of the whole CALL_INSN_FUNCTION_USAGE.
35489 PR tree-optimization/79972
35490 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35491 get_range_info on SSA_NAMEs. Formatting fixes.
35493 2017-03-10 Richard Biener <rguenther@suse.de>
35494 Jakub Jelinek <jakub@redhat.com>
35496 PR tree-optimization/77975
35497 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35498 edge to be constant.
35499 (get_val_for): For constant x return it. Formatting fix.
35500 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35501 would use the same bases as the current one.
35503 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35505 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35506 instead of vec_select for V1TImode.
35507 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35509 (VSX_LE_128): Add V1TI to this mode iterator.
35510 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35511 (*vsx_le_perm_store_<mode>): Likewise.
35512 (pre-reload splitter for VSX stores): Likewise.
35513 (post-reload splitter for VSX stores): Likewise.
35514 (*vsx_xxpermdi2_le_<mode>): Likewise.
35515 (*vsx_lxvd2x2_le_<mode>): Likewise.
35516 (*vsx_stxvd2x2_le_<mode>): Likewise.
35518 2017-03-09 Michael Eager <eager@eagercon.com>
35520 Correct failures with --enable-checking=yes,rtl.
35522 * config/microblaze/microblaze.c (microblaze_expand_shift):
35523 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35524 test for const0_rtx.
35525 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35526 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35528 2017-03-09 Richard Biener <rguenther@suse.de>
35530 PR tree-optimization/79977
35531 * graphite-scop-detection.c (scop_detection::merge_sese):
35532 Handle the case of extra exits to blocks dominating the entry.
35534 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35536 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35539 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35541 PR rtl-optimization/79949
35542 * lra-constraints.c (process_alt_operands): Check memory when
35543 trying to predict a cycle. Print about the overall increase.
35545 2017-03-09 Richard Biener <rguenther@suse.de>
35547 PR middle-end/79971
35548 * gimple-expr.c (useless_type_conversion_p): Preserve
35549 TYPE_SATURATING for fixed-point types.
35551 2017-03-09 Richard Biener <rguenther@suse.de>
35554 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35555 alignment of BLKmode params.
35557 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35560 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35561 (VALL_NO_V2Q): Likewise.
35563 * config/aarch64/aarch64-simd.md
35564 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35566 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35567 VALL_NO_V2Q mode iterator.
35568 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35570 2017-03-09 Martin Liska <mliska@suse.cz>
35572 PR tree-optimization/79631
35573 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35574 tree_int_cst_sign_bit just for INTEGER constants.
35576 2017-03-09 Martin Liska <mliska@suse.cz>
35580 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35583 2017-03-09 Marek Polacek <polacek@redhat.com>
35586 * tree.c (inchash::add_expr): Handle TREE_VEC.
35588 2017-03-09 Martin Liska <mliska@suse.cz>
35591 (chkp_narrow_size_and_offset): New function.
35592 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35593 (void chkp_parse_bit_field_ref): New function.
35594 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35595 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35597 2017-03-09 Martin Liska <mliska@suse.cz>
35600 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35601 (chkp_find_bounds_1): Remove gcc_unreachable.
35603 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35606 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35607 BUILT_IN_SYNC*, determine the access type from the size suffix and
35608 always build a MEM_REF with that type. Handle forgotten
35609 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35612 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35613 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35614 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35615 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35616 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35617 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35618 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35619 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35620 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35621 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35622 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35623 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35624 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35625 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35626 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35627 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35628 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35629 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35630 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35631 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35632 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35633 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35634 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35635 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35636 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35637 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35638 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35639 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35640 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35641 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35642 definitions outside of __OPTIMIZE__ guarded section.
35645 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35646 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35647 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35648 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35651 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35653 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35654 ("vfenez<mode>"): Add missing constraints.
35656 2017-03-08 Martin Sebor <msebor@redhat.com>
35659 * config/nds32/nds32.c (nds32_option_override):
35660 Fix misspelled diagnostic.
35662 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35665 * gimplify.c (gimplify_omp_for): Replace index var in outer
35666 taskloop statement with an artificial variable and add
35667 OMP_CLAUSE_PRIVATE clause for it.
35669 2017-03-08 Richard Biener <rguenther@suse.de>
35671 PR tree-optimization/79955
35672 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35673 for accesses that are completely outside of the variable.
35675 2017-03-08 Andrew Haley <aph@redhat.com>
35677 PR tree-optimization/79943
35678 * tree-ssa-loop-split.c (compute_new_first_bound): When
35679 calculating the new upper bound, (END-BEG) should be added, not
35682 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35684 * config/avr/avr.md (setmemhi): Make sure match_dup
35685 operand number comes before match_scratch.
35687 2017-03-08 Richard Biener <rguenther@suse.de>
35689 PR tree-optimization/79920
35690 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35691 with ncopies == 1 to ...
35692 (vect_transform_slp_perm_load): ... here. Properly compute
35693 all element loads by iterating VF times over the group. Do
35694 not handle ncopies (computed in a broken way) in
35695 vect_create_mask_and_perm.
35697 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35700 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35701 is a uniform vector, use uniform_vector_p return value instead of
35702 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35704 2017-03-07 Marek Polacek <polacek@redhat.com>
35706 PR middle-end/79809
35707 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35708 (alloca_call_type): Likewise.
35710 2017-03-07 Martin Liska <mliska@suse.cz>
35712 * gcov.c (process_args): Put comment to correct location.
35714 2017-03-07 Martin Liska <mliska@suse.cz>
35716 PR middle-end/68270
35717 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35718 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35719 (chkp_narrow_bounds_for_field): Likewise.
35720 (chkp_parse_array_and_component_ref): Pass one more argument to
35723 2017-03-07 Richard Biener <rguenther@suse.de>
35725 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35728 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35730 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35731 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35733 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35736 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35737 to end of description.
35738 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35740 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35742 PR rtl-optimization/79901
35743 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35745 (*avx512f_<code><mode>3<mask_name>): ... this.
35746 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35747 iterator instead of VI8_AVX2_AVX512BW.
35749 PR rtl-optimization/79901
35750 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35751 min/max expander, expand it using expand_vec_cond_expr.
35754 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35757 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35760 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35761 to void * for PCH reasons.
35762 * dwarf2out.c (output_loc_operands, output_die): Cast
35763 v.val_vec.array to unsigned char *.
35765 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35768 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35771 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35773 PR rtl-optimization/79571
35774 * lra-constraints.c (process_alt_operands): Calculate static
35775 reject and subtract it from overall when only addresses will be
35778 2017-03-06 Julia Koval <julia.koval@intel.com>
35781 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35782 incoming stack boundary to 128 for 64-bit targets.
35784 2017-03-06 Richard Biener <rguenther@suse.de>
35786 PR tree-optimization/79894
35787 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35788 to NULL after folding it.
35790 2017-03-06 Richard Biener <rguenther@suse.de>
35792 PR tree-optimization/79824
35793 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35794 check disabling peeling for gaps.
35796 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35798 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35799 attributes): Document gettimeofday.
35801 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35803 * config/s390/s390.c (s390_option_override_internal): Set
35804 PARAM_MIN_VECT_LOOP_BOUND
35806 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35808 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35809 * config/s390/s390.md: Likewise.
35811 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35814 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35815 (<avx2_avx512>_perm<mode>): Rename to ...
35816 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35818 (<avx512>_perm<mode>_mask): Rename to ...
35819 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35821 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35822 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35823 instead of VI8F_256_512.
35824 (avx512f_perm<mode>): New define_expand.
35825 (avx512f_perm<mode>_mask): Likewise.
35826 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35827 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35829 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35831 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35832 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35834 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35836 2017-03-06 Martin Liska <mliska@suse.cz>
35839 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35840 when having a SSA NAME w/o VAR_DECL assigned to it.
35842 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35844 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35845 msa_dpsub_<su>_d): Fix MODE for vec_select.
35847 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35849 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35851 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35853 2017-03-06 Richard Biener <rguenther@suse.de>
35855 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35856 * plugin.c (register_plugin_info): Likewise.
35857 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
35859 2017-03-05 Jakub Jelinek <jakub@redhat.com>
35861 * config/i386/sse.md (sse_storehps, sse_storelps,
35862 avx_<castmode><avxsizesuffix>_<castmode>,
35863 avx512f_<castmode><avxsizesuffix>_<castmode>,
35864 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
35865 in condition that at least one operand is not a MEM.
35867 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35869 PR middle-end/79805
35870 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
35871 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
35873 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
35874 gimple_call_nothrow_p flag based on whether original builtin can throw.
35875 If it can, emit following stmts on the fallthrough edge.
35876 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
35877 don't create new bb if inserting just debug stmts on the edge, try to
35878 insert them on the fallthru bb or just reset debug stmts.
35880 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
35883 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
35884 restore recog_data (including the operand rtxes inside it) around
35885 the call to get_insn_template.
35887 2017-03-03 Martin Sebor <msebor@redhat.com>
35889 PR tree-optimization/79699
35890 * context.c (context::~context): Free MPFR caches to avoid
35891 a memory leak on program exit.
35893 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35895 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
35896 Use wide_int::ulow () instead of .elt (0).
35898 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35900 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
35901 (*pushxf): Limit oF constraint to 32bit targets and add oC
35902 constraint for 64bit targets.
35903 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
35904 (*pushdf): Change rmF constraint to rmC.
35906 2017-03-03 Martin Liska <mliska@suse.cz>
35908 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
35909 Remove unused variable.
35911 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35914 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
35915 is a memory operand, increase num_memory.
35916 (ix86_expand_args_builtin): Likewise.
35918 2017-03-03 Jan Hubicka <jh@suse.cz>
35921 * ipa-devirt.c (maybe_record_node): Properly handle
35922 __cxa_pure_virtual visibility.
35924 2017-03-03 Martin Liska <mliska@suse.cz>
35926 PR tree-optimization/79803
35927 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
35929 (pass_loop_prefetch::execute): Disabled optimization if an
35930 assumption about L1 cache size is not met.
35932 2017-03-03 Martin Liska <mliska@suse.cz>
35934 PR rtl-optimization/79574
35935 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
35936 (hash_scan_set): Likewise.
35937 (dump_hash_table): Likewise.
35938 (hoist_code): Likewise.
35940 2017-03-03 Richard Biener <rguenther@suse.de>
35942 * fixed-value.c (fixed_from_string): Restore use of elt (1)
35943 in place of uhigh ().
35944 (fixed_convert_from_real): Likewise.
35946 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35949 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
35951 2017-03-03 Richard Biener <rguenther@suse.de>
35953 PR middle-end/79818
35954 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
35955 TYPE_OVERFLOW_UNDEFINED check.
35957 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35959 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
35961 (vector_ae_<mode>_p): Likewise.
35962 (vector_nez_<mode>_p): Likewise.
35963 (vector_ne_v2di_p): Likewise.
35964 (vector_ae_v2di_p): Likewise.
35965 (vector_ne_<mode>_p): Likewise.
35966 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
35968 (vsx_tsqrt<mode>2_fe): Likewise.
35970 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
35973 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
35975 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35977 PR rtl-optimization/79780
35978 * cprop.c (one_cprop_pass): When second and further conditional trap
35979 in a single basic block is turned into an unconditional trap, turn it
35980 into a deleted note to avoid RTL verification failures.
35982 2017-03-02 Richard Biener <rguenther@suse.de>
35984 * fold-const.c (const_binop): Use ulow () instead of elt (0).
35986 2017-03-02 Richard Biener <rguenther@suse.de>
35988 PR tree-optimization/79345
35990 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
35991 param and abort the walk, returning -1 if it is hit.
35992 (walk_aliased_vdefs): Take a limit param and pass it on.
35993 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
35994 defaulting to 0 and return a signed int.
35995 * tree-ssa-uninit.c (struct check_defs_data): New struct.
35996 (check_defs): New helper.
35997 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
35998 about uninitialized memory.
35999 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36000 bogus uninitialized warning.
36001 (fixed_convert_from_real): Likewise.
36003 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36005 PR tree-optimization/66768
36006 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36007 iv_use if base object can't be determined.
36009 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36011 PR tree-optimization/79345
36012 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36013 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36014 (get_pattern_stats): Initialize it.
36015 * genemit.c (gen_expand): Verify match_scratch numbers come after
36016 match_operand/match_dup numbers.
36017 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36018 match_scratch numbers.
36019 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36021 * config/s390/s390.md (trunctdsd2): Likewise.
36023 2017-03-02 Richard Biener <rguenther@suse.de>
36025 * wide-int.h (wide_int_storage::operator=): Implement in terms
36028 2017-03-02 Richard Biener <rguenther@suse.de>
36030 PR tree-optimization/79777
36031 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36032 the to insert expression to sth existing.
36034 2017-03-01 Martin Sebor <msebor@redhat.com>
36036 PR middle-end/79692
36037 * gimple-ssa-sprintf.c
36038 (directive::known_width_and_precision): New function.
36039 (format_integer): Use it.
36040 (get_mpfr_format_length): Consider the full range of precision
36041 when computing %g output with the # flag. Set the likely byte
36042 count to 3 rather than 1 when precision is indeterminate.
36043 (format_floating): Correct the lower bound of precision.
36045 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36047 * doc/invoke.texi: Document default code model for 64-bit Linux.
36049 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36052 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36053 udiv rather than div since input pattern is unsigned.
36055 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36057 * config/i386/i386.c (print_reg): Warn for values of
36058 unsupported size in integer register.
36060 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36063 * config/rs6000/predicates.md (current_file_function_operand): Do
36064 not allow self calls to be local if the function is replaceable.
36066 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36069 * config/rs6000/altivec.h (vec_ctz and others): Change the
36070 preprocessor macro that controls conditional compilation from
36071 _ARCH_PWR9 to __POWER9_VECTOR__.
36072 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36073 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36074 control (instead of _ARCH_PWR9 control) so that template
36075 definition uses power9-specific function.
36076 (vec_any_eq): Likewise.
36077 (vec_all_ne): Change macro definition to use a power9-specific
36078 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36079 _ARCH_PWR9 control).
36080 (vec_any_eq) Likewise.
36081 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36082 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36083 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36084 support for xvcmpnedp instruction.
36085 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36086 macro expansion so that Power9 implementation of vec_all_ne does
36087 not use the AltiVec predicate framework.
36088 (VCMPNEH_P): Likewise.
36089 (VCMPNEW_P): Likewise.
36090 (VCMPNED_P): Likewise.
36091 (VCMPNEFP_P): Likewise.
36092 (VCMPNEDP_P): Likewise.
36093 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36094 implementation of vec_any_eq to not use AltiVec predicate
36096 (VCMPAEH_P): Likewise.
36097 (VCMPAEW_P): Likewise.
36098 (VCMPAED_P): Likewise.
36099 (VCMPAEFP_P): Likewise.
36100 (VCMPAEDP_P): Likewise.
36101 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36102 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36103 not use the AltiVec predicate framework.
36104 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36105 of vec_any_eq to not use AltiVec predicate framework.
36106 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36107 support for predefined __POWER9_VECTOR__ macro to indicate that
36108 Power9 instruction selection is enabled.
36109 (altivec_overloaded_builtins): Remove extraneous
36110 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36111 function argument types RS6000_BTI_bool_V16QI and
36112 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36113 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36114 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36115 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36116 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36117 Power9 for implementations of vec_cmpne. Change the signature for
36118 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36119 (representing vec_all_ne) to remove the previously described first
36120 argument of type RS6000_BTI_INTSI, as this was an artifact of
36121 reliance on the AltiVec predicate framework, which is no longer
36122 used in the implementation of these functions. Add
36123 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36124 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36125 since, unlike the AltiVec predicate framework implementation, we
36126 do not share function descriptors between vec_alle and vec_anyeq.
36127 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36128 set of modes that receive special treatment even when
36129 TARGET_P9_VECTOR is true. The special treatment emits code that
36130 does not depend on Power9 instructions.
36131 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36132 define_expand to not rely on AltiVec predicate framework.
36133 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36135 (vector_ne_v2di_p): Change this define_expand to not rely on
36136 AltiVec predicate framework.
36137 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36139 (vector_ne_<mode>_p): Change this define_expand to not rely on
36140 AltiVec predicate framework.
36141 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36143 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36144 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36145 define_insn pattern.
36146 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36147 define_insn pattern because the xvcmpne<VSs>. instruction is not
36149 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36150 instruction is not supported.
36152 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36154 * config/nvptx/nvptx.c: Include intl.h.
36156 2017-03-01 Martin Jambor <mjambor@suse.cz>
36159 * ipa-prop.h (ipa_bits): Removed field known.
36160 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36161 to pointers. Adjusted their comments to warn about their sharing.
36162 (ipcp_transformation_summary): Change bits to a vector of pointers.
36163 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36164 (ipa_get_ipa_bits_for_value): Declare.
36165 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36166 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36167 (ipa_bits_hash_table): Likewise.
36168 (ipa_vr_ggc_hash_traits): Likewise.
36169 (ipa_vr_hash_table): Likewise.
36170 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36171 being pointers and vr_known being removed.
36172 (ipa_set_jf_unknown): Likewise.
36173 (ipa_get_ipa_bits_for_value): New function.
36174 (ipa_set_jfunc_bits): Likewise.
36175 (ipa_get_value_range): New overloaded functions.
36176 (ipa_set_jfunc_vr): Likewise.
36177 (ipa_compute_jump_functions_for_edge): Use the above functions to
36178 construct bits and vr parts of jump functions.
36179 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36180 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36182 (ipcp_grow_transformations_if_necessary): Also allocate
36183 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36185 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36186 them. Fix too long lines.
36187 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36188 vr_known being removed.
36189 (ipa_read_jump_function): Use new setter functions to construct bits
36190 and vr parts of jump functions or set them to NULL.
36191 (write_ipcp_transformation_info): Adjust for bits being pointers.
36192 (read_ipcp_transformation_info): Likewise.
36193 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36195 Include gt-ipa-prop.h.
36196 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36198 (ipcp_store_bits_results): Likewise.
36199 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36200 Do not write to existing jump functions but use a temporary instead.
36202 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36205 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36206 attempt to use its first operand as BIT_FIELD_REF base.
36208 2017-03-01 Richard Biener <rguenther@suse.de>
36210 PR middle-end/79721
36211 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36212 interpolating formula in wrapping arithmetic.
36213 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36215 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36217 PR tree-optimization/79734
36218 * tree-vect-generic.c (expand_vector_condition): Optimize
36219 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36220 Handle VEC_COND_EXPR where comparison has different inner width from
36221 type's inner width.
36223 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36225 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36226 markup, and similar issues. Remove @opindex entries for things
36227 that aren't options. Add missing -mmpy-option entries.
36229 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36231 PR tree-optimization/79737
36232 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36233 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36234 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36235 instead of byte_size. Formatting fix.
36236 (shift_bytes_in_array_right): Formatting fix.
36238 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36241 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36242 condition on optimize for the leaf function test.
36244 2017-02-28 Martin Liska <mliska@suse.cz>
36247 * read-rtl-function.c (function_reader::handle_unknown_directive):
36248 Bail out when one uses -flto.
36250 2017-02-28 Martin Liska <mliska@suse.cz>
36252 * common.opt: Replace space with tabular for options of <number>
36254 * config/i386/i386.opt: Show <number> value for
36255 -mlarge-data-threshold.
36256 * opts.c (print_filtered_help): Do not display number in hexadecimal
36259 2017-02-28 Martin Liska <mliska@suse.cz>
36261 * common.opt: Fix --help=option -Q for options which are of
36264 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36266 * config/i386/i386.c (print_reg): Error out for values
36267 of 8-bit size in invalid integer register.
36269 2017-02-28 Martin Sebor <msebor@redhat.com>
36271 PR tree-optimization/79691
36272 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36274 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36277 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36278 gcc_unreachable with output_operand_lossage.
36280 2017-02-28 Richard Biener <rguenther@suse.de>
36282 PR tree-optimization/79740
36283 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36285 (visit_nary_op): Insert the nary into the hashtable if we
36286 pattern-matched sth.
36287 * tree-ssa-pre.c (eliminate_insert): Robustify.
36289 2017-02-28 Richard Biener <rguenther@suse.de>
36291 PR middle-end/79731
36292 * fold-const.c (decode_field_reference): Reject out-of-bound
36295 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36297 * config/i386/i386.c: Include intl.h.
36298 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36299 instead of just cond ? "..." : "...".
36300 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36301 * coverage.c (read_counts_file): Likewise.
36302 * omp-offload.c: Include intl.h.
36303 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36304 of just cond ? "..." : "...".
36305 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36306 of just cond ? "..." : "...".
36308 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36311 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36313 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36314 'tune for' CPU name.
36315 * config/arm/arm-cpu-data.h: Regenerated.
36317 2017-02-28 Richard Biener <rguenther@suse.de>
36319 PR tree-optimization/79732
36320 * tree-inline.c (expand_call_inline): Do not shadow var.
36322 2017-02-28 Richard Biener <rguenther@suse.de>
36324 PR tree-optimization/79723
36325 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36326 address-space properly.
36328 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36330 * doc/optinfo.texi (Optimization groups): Fix option used for
36332 * doc/invoke.texi (-fopt-info): Document "omp".
36333 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36334 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36335 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36336 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36337 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36339 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36341 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36342 OPTGROUP_OMP with "omp".
36344 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36347 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36348 for arithmetic shift of unsigned V2DI.
36350 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36352 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36353 arc/linux.h headers.
36354 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36355 (LINK_SPEC): Likewise.
36356 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36357 (EXTRA_SPECS): Likewise.
36358 (STARTFILE_SPEC): Likewise.
36359 (ENDFILE_SPEC): Likewise.
36360 (LIB_SPEC): Likewise.
36361 (TARGET_SDATA_DEFAULT): Likewise.
36362 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36363 (MULTILIB_DEFAULTS): Likewise.
36364 (DWARF2_UNWIND_INFO): Likewise.
36365 * config/arc/big.h: New file.
36366 * config/arc/elf.h: Likewise.
36367 * config/arc/linux.h: Likewise.
36368 * config/arc/t-uClibc: Remove.
36370 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36372 PR tree-optimization/77536
36373 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36374 (tree_transform_and_unroll_loop): Use above function to compute the
36375 estimated niter of unrolled loop and use it when scaling profile.
36376 Also use count info rather than frequency if it's non-zero.
36377 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36378 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36379 (vect_transform_loop): Call above function.
36381 2017-02-27 Richard Biener <rguenther@suse.de>
36383 PR tree-optimization/45397
36384 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36385 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36386 (visit_nary_op): Add pattern matching for CSEing sign-changed
36387 or truncated operations with wider ones.
36389 2017-02-27 Richard Biener <rguenther@suse.de>
36391 PR tree-optimization/79690
36392 * tree-vect-stmts.c (vectorizable_store): Use vector type
36393 built from the DR with address-space.
36395 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36397 * doc/invoke.texi (Optimize Options): Refine the description
36398 of asan-use-after-return.
36400 2017-02-25 Alan Modra <amodra@gmail.com>
36402 PR rtl-optimization/79584
36403 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36404 base, not ad->base_term, the reg within base. Remove assertion
36405 that ad->base == ad->base_term. Replace gen_int_mode using
36406 bogus mode with const0_rtx.
36408 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36410 PR middle-end/79396
36411 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36412 FMA_EXPR like tcc_binary or tcc_unary.
36414 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36417 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36419 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36420 (output_loc_operands): Handle DW_OP_call_ref and
36421 DW_OP_GNU_variable_value.
36422 (struct variable_value_struct): New type.
36423 (struct variable_value_hasher): Likewise.
36424 (variable_value_hash): New variable.
36425 (string_types): Remove.
36426 (copy_loc_descr): New function.
36427 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36428 (prepend_loc_descr_to_each): New function.
36429 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36430 instead of add_loc_descr_to_each if the first argument is single
36431 location list and the second has multiple.
36432 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36433 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36434 when looking for variable value which doesn't have other location info.
36435 (loc_list_from_tree): Formatting fix.
36436 (gen_array_type_die): Simplify DW_AT_string_length handling.
36437 (adjust_string_types): Remove.
36438 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36439 string_types. Call resolve_variable_values.
36440 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36441 (resolve_addr_in_expr): Likewise. Add A argument.
36442 (copy_deref_exprloc): Remove deref argument. Adjust for the
36443 original expression being DW_OP_GNU_variable_value with optionally
36444 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36445 optionally after it.
36446 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36447 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36448 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36449 (variable_value_hasher::hash, variable_value_hasher::equal): New
36451 (resolve_variable_value_in_expr, resolve_variable_value,
36452 resolve_variable_values, note_variable_value_in_expr,
36453 note_variable_value): New functions.
36454 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36457 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36460 * opts.h (handle_generated_option): Add GENERATED_P argument.
36461 * opts-common.c (handle_option): Adjust function comment.
36462 (handle_generated_option): Add GENERATED_P argument, pass it to
36464 (control_warning_option): Pass false to handle_generated_option
36466 * opts.c (maybe_default_option): Pass true to handle_generated_option
36468 * optc-gen.awk: Likewise.
36470 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36472 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36473 a REG, look at the REG it is a SUBREG of.
36474 (splitter for cmpeqsi_t): Ditto.
36476 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36478 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36479 the special USEs with the pattern of the insn, not the insn itself.
36481 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36484 * doc/invoke.texi: Document -mload-store-pairs.
36486 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36487 Sandra Loosemore <sandra@codesourcery.com>
36489 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36490 argument isn't a CONST_INT.
36491 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36492 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36493 (nios2_validate_compare): Bypass alternate compare logic if *op2
36494 is not a CONST_INT.
36495 (ldstwm_operation_p): Return false if first_base is not a REG or
36496 if first_offset is not a CONST_INT.
36498 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36500 * config/cris/cris.md: Use correct operand in a define_peephole2.
36502 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36504 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36506 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36508 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36509 this_insn if it is an INSN or JUMP_INSN.
36510 (force_offsettable): Look at base, not at addr.
36511 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36512 on things that aren't necessarily CONST_INTs.
36514 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36516 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36517 -mfpmath=sse is the default also for x86-32 targets with SSE2
36518 instruction set when @option{-ffast-math} is enabled
36520 2017-02-24 Jeff Law <law@redhat.com>
36522 PR rtl-optimizatoin/79286
36523 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36526 2017-02-24 Richard Biener <rguenther@suse.de>
36528 PR tree-optimization/79389
36529 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36532 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36534 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36535 function comment to reflect reality.
36536 (loop_exits_before_overflow): Fix typo in function description.
36538 2017-02-24 Richard Biener <rguenther@suse.de>
36540 PR tree-optimization/79389
36541 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36542 properly that a threading opportunity exists. Detect conditional
36543 copy/constant propagation opportunities.
36545 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36547 * config/visium/visium.md (type): Add trap.
36548 (b): New mode attribute.
36549 (*btst): Rename into...
36550 (*btst<mode>): ...this and adjust.
36551 (*cbranchsi4_btst_insn): Rename into...
36552 (*cbranch<mode>4_btst_insn): ...this and adjust.
36553 (trap): New define_insn.
36555 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36557 PR tree-optimization/79389
36558 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36559 (noce_reversed_cond_code): New function.
36560 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36561 reversed_comparison_code. Formatting fix.
36562 (noce_try_store_flag): Test rev_cond != NULL in addition to
36563 reversed_comparison_code.
36564 (noce_try_store_flag_constants): Likewise.
36565 (noce_try_store_flag_mask): Likewise.
36566 (noce_try_addcc): Use rev_cond if non-NULL instead of
36567 reversed_comparison_code.
36568 (noce_try_cmove_arith): Likewise. Formatting fixes.
36569 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36570 (noce_find_if_block): Initialize rev_cond.
36571 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36572 instead of false as last argument never attempt to reverse it
36575 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36577 PR tree-optimization/79663
36578 * tree-predcom.c (combine_chains): Process refs in reverse order
36579 only for ZERO length chains, and add explaining comment.
36581 2017-02-23 Jeff Law <law@redhat.com>
36583 PR tree-optimization/79578
36584 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36585 in call to operand_equal_p.
36587 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36590 * config/i386/cpuid.h: Fix another undefined behavior.
36592 2017-02-23 Richard Biener <rguenther@suse.de>
36594 PR tree-optimization/79683
36595 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36596 vector types for data-refs.
36598 2017-02-23 Martin Liska <mliska@suse.cz>
36600 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36602 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36604 PR middle-end/79665
36605 * internal-fn.c (get_range_pos_neg): Moved to ...
36606 * tree.c (get_range_pos_neg): ... here. No longer static.
36607 * tree.h (get_range_pos_neg): New prototype.
36608 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36609 are known to be in between 0 and signed maximum inclusive, try to
36610 expand both unsigned and signed divmod and use the cheaper one from
36613 2017-02-22 Jeff Law <law@redhat.com>
36615 PR tree-optimization/79578
36616 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36617 to compare base operands.
36619 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36622 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36623 gpc_reg_operand instead of fpr_reg_operand.
36625 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36627 * config/mips/mips.c (mips_return_in_memory): Force FP
36628 vector types to be returned in memory for o32 ABI.
36630 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36632 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36633 instead of DW_TAG_member for static data member declarations and don't
36634 set no_linkage_name for static inline data members.
36635 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36638 2017-02-22 Martin Liska <mliska@suse.cz>
36640 * doc/invoke.texi: Replace inequality signs with square brackets
36643 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36646 * lra-constraints.c (simplify_operand_subreg): Handle
36647 WORD_REGISTER_OPERATIONS targets.
36649 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36652 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36653 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36654 elimination by swapping fld*.
36656 2017-02-22 Richard Biener <rguenther@suse.de>
36658 PR tree-optimization/79673
36659 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36660 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36661 irrelevant address-space qualifiers and avoiding a
36662 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36664 2017-02-22 Richard Biener <rguenther@suse.de>
36666 PR tree-optimization/79666
36667 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36668 to not symbolically negate if that may introduce undefined
36671 2017-02-22 Martin Liska <mliska@suse.cz>
36674 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36675 * data-streamer-out.c (streamer_write_gcov_count_stream):
36677 * value-prof.c (stream_out_histogram_value): Make assert more
36678 precise based on type of counter.
36680 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36683 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36684 Use nonimmediate_operand instead of memory_operand for operand 1.
36685 (float-extend standard_x87sse_constant_load splitter): Ditto.
36687 2017-02-21 Jeff Law <law@redhat.com>
36689 PR tree-optimization/79621
36690 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36691 blocks with edges to themselves.
36693 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36696 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36697 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36698 Use gimple_call_builtin_p.
36701 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36702 on temporarily removed DEBUG_INSNs.
36704 PR tree-optimization/79649
36705 * tree-loop-distribution.c (classify_partition): Give up on
36706 non-generic address space loads/stores.
36708 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36710 * doc/loop.texi (Loop manipulation): Remove nonexistent
36711 tree_ssa_loop_version from the documentation.
36712 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36714 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36717 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36718 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36719 * config/rs6000/rs6000.c: Include except.h.
36720 (rs6000_expand_split_stack_prologue): Call
36721 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36723 2017-02-21 Martin Jambor <mjambor@suse.cz>
36726 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36727 have been analyzed.
36729 2017-02-21 Martin Jambor <mjambor@suse.cz>
36731 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36732 for backward compatibility only.
36733 * doc/invoke.texi (Option Summary): Remove all references to
36734 -fipa-cp-alignment.
36736 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36740 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36742 * lra-constraints.c (curr_insn_transform): Handle
36743 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36745 2017-02-21 Martin Liska <mliska@suse.cz>
36747 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36749 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36751 PR translation/79638
36752 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36754 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36757 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36758 (arm_function_ok_for_sibcall): Return false for an indirect call by
36759 descriptor if all the argument registers are used.
36760 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36761 alignment of the function.
36763 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36765 PR tree-optimization/61441
36766 * simplify-rtx.c (simplify_const_unary_operation): For
36767 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36768 the sNaN unmodified.
36770 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36772 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36773 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36774 instead of SYSTEM_HEADER_DIR.
36776 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36777 Martin Liška <mliska@suse.cz>
36779 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36780 Fix typos and grammar, use active voice, and clarify.
36782 2017-02-20 Marek Polacek <polacek@redhat.com>
36784 PR middle-end/79537
36785 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36788 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36790 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36793 * config/i386/i386.c (ix86_expand_builtin): Handle
36794 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36795 ix86_builtins_isa[fcode].isa as a requirement of those
36796 flags and any other flag in the bitmask.
36797 (ix86_init_mmx_sse_builtins): Use 0 instead of
36798 ~OPTION_MASK_ISA_64BIT as mask.
36799 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36800 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36801 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36802 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36804 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36807 * lra-constraints.c (split_reg): Check requested split mode
36808 is supported by the register.
36810 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36812 * lra-constraints.c (simplify_operand_subreg): Remove early
36815 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36818 * lra-constraints.c (curr_insn_transform): Tighten condition
36819 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36821 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36824 * lra-constraints.c (curr_insn_transform): Handle
36825 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36827 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36830 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36832 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36834 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36837 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36840 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36842 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36845 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36847 * final.c (last_columnnum, override_columnnum): New variables.
36848 (final_start_function): Set last_columnnum, pass it to begin_prologue
36849 hook and pass 0 to dwarf2out_begin_prologue.
36850 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36851 to source_line debug hook.
36852 (notice_source_line): Compute last_columnnum and for debug_column_info
36853 return true on column changes.
36854 * debug.h (struct gcc_debug_hooks): Add column argument to
36855 source_line and begin_prologue hooks.
36856 (debug_nothing_int_charstar_int_bool): Remove prototype.
36857 (debug_nothing_int_int_charstar,
36858 debug_nothing_int_int_charstar_int_bool): New prototypes.
36859 (dwarf2out_begin_prologue): Add column argument.
36860 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36861 begin_prologue hooks.
36862 (debug_nothing_int_charstar_int_bool): Remove.
36863 (debug_nothing_int_int_charstar,
36864 debug_nothing_int_int_charstar_int_bool): New functions.
36865 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
36866 through to dwarf2out_source_line.
36867 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
36868 (dwarf2out_source_line): Add column argument, emit it if requested.
36869 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
36871 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36872 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36873 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
36874 through to dwarf2out_begin_prologue.
36875 (vmsdbgout_source_line): Add column argument, pass it through to
36876 dwarf2out_source_line.
36877 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
36878 dbxout_source_line caller.
36879 (dbxout_source_line): Add column argument.
36881 * common.opt (gno-column-info, gcolumn-info): New options.
36882 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
36883 (check_die): Also test for multiple DW_AT_decl_column attributes.
36884 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
36885 DW_AT_decl_column if requested.
36886 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
36888 (gen_variable_die): Likewise.
36889 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
36890 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
36893 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
36894 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
36895 (ix86_handle_option): Handle OPT_m3dnowa.
36896 * doc/invoke.texi (-m3dnowa): Document.
36897 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
36898 -m3dnowa instead of -m3dnow -march=athlon.
36901 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
36902 instead of gcc_assert for K, r and R code checks. Formatting fixes.
36904 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36907 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
36908 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
36909 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
36910 generator for vsx_xxpermdi_<mode>_be.
36911 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
36912 force big-endian semantics.
36913 (vsx_xxpermdi_<mode>_be): New define_expand with same
36914 implementation as previous version of vsx_xxpermdi_<mode>.
36916 2017-02-17 Jakub Jelinek <jakub@redhat.com>
36918 PR tree-optimization/79327
36919 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
36920 variable, its initialization and use.
36922 2017-02-17 Julia Koval <julia.koval@intel.com>
36924 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
36925 (OPTION_MASK_ISA_PKU_UNSET): New.
36926 (ix86_handle_option): Handle -mrdpid.
36927 * config/i386/cpuid.h (bit_RDPID): New.
36928 * config/i386/driver-i386.c (host_detect_local_cpu):
36929 Detect RDPID feature.
36930 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
36931 * config/i386/i386-c.c (ix86_target_macros_internal):
36933 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
36934 (ix86_valid_target_attribute_inner_p): Add "rdpid".
36935 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
36936 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
36937 * config/i386/i386.md (define_insn "rdpid"): New.
36938 * config/i386/i386.opt Add -mrdpid.
36939 * config/i386/immintrin.h (_rdpid_u32): New.
36941 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
36943 PR rtl-optimization/79541
36944 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
36945 instead of transforming it into USE.
36947 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
36949 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
36950 If HONOR_SNANS (SFmode) force the input to a register.
36951 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
36952 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
36953 an frsp or similar insn.
36955 2017-02-17 Martin Liska <mliska@suse.cz>
36957 PR rtl-optimization/79577
36958 * params.def (selsched-max-sched-times): Increase minimum to 1.
36960 2017-02-17 Martin Liska <mliska@suse.cz>
36962 PR rtl-optimization/79574
36963 * gcse.c (want_to_gcse_p): Prevent integer overflow.
36965 2017-02-17 Martin Liska <mliska@suse.cz>
36967 PR tree-optimization/79529
36968 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
36969 ssa_defined_default_def_p to handle cases which are implicitly
36971 * tree-ssa.c (ssa_defined_default_def_p): New function.
36972 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
36973 which are implicitly defined.
36974 * tree-ssa.h (ssa_defined_default_def_p): Declare.
36976 2017-02-17 Richard Biener <rguenther@suse.de>
36978 PR middle-end/79576
36979 * params.def (max-ssa-name-query-depth): Limit to 10.
36981 2017-02-17 Richard Biener <rguenther@suse.de>
36983 PR tree-optimization/79552
36984 * tree-ssa-structalias.c (visit_loadstore): Properly verify
36987 2017-02-17 Richard Biener <rguenther@suse.de>
36990 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
36992 2017-02-17 Marek Polacek <polacek@redhat.com>
36994 PR middle-end/79536
36995 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
36996 (fold_negate_expr): New wrapper.
36998 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37000 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37001 Correct terminology and de-emphasize pre-standard behavior.
37003 2017-02-16 Alan Modra <amodra@gmail.com>
37005 PR rtl-optimization/79286
37006 * ira.c (def_dominates_uses): New function.
37007 (update_equiv_regs): Don't create an equivalence for insns that
37008 may trap where the register def does not dominate the use.
37010 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37012 PR rtl-optimization/78127
37013 * lra.c (lra): Call lra_eliminate before finish the loop after
37016 2017-02-16 Richard Biener <rguenther@suse.de>
37018 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37020 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37021 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37022 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37023 (isl_val_int_from_wi): New function.
37024 (extract_affine_gmp): Rename to ...
37025 (extract_affine_wi): ... this, take a widest_int.
37026 (extract_affine_int): Just wrap extract_affine_wi.
37027 (add_param_constraints): Use isl_val_int_from_wi.
37028 (add_loop_constraints): Likewise, and extract_affine_wi.
37030 2017-02-15 Jeff Law <law@redhat.com>
37032 PR middle-end/79521
37033 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37034 ira_init_register_move_cost_if_necessary.
37036 2017-02-15 Martin Sebor <msebor@redhat.com>
37038 PR middle-end/32003
37039 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37040 removed in a prior commit.
37042 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37044 PR tree-optimization/79347
37045 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37046 counters during peeling.
37048 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37050 * Makefile.in (site.exp): Remove "set ISLVER".
37052 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37055 * real.c (real_from_integer): Call real_convert even for decimal.
37057 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37060 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37062 2017-02-14 Andrew Pinski <apinski@cavium.com>
37064 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37065 cores and change the partno/implementer to be correct.
37066 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37067 the 'B" as the implementer.
37068 * config/aarch64/aarch64-tune.md: Regenerate.
37070 2017-02-14 Carl Love <cel@us.ibm.com>
37072 * config/rs6000/rs6000.c: Add case statement entry to make the
37073 xvcvuxdsp built-in argument unsigned.
37074 * config/rs6000/vsx.md: Fix the source and return operand types so they
37075 match the instruction definitions from the ISA document. Fix typo
37076 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37079 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37082 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37083 member early_clobber_alts.
37084 * lra-lives.c (reg_early_clobber_p): New.
37085 (process_bb_lives): Use it.
37086 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37087 (debug_operand_data): Initialize early_clobber_alts.
37088 (setup_operand_alternative): Set up early_clobber_alts.
37089 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37090 alternatives to new_insn_reg.
37091 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37093 (lra_update_insn_regno_info): Pass the new arg.
37095 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37097 PR middle-end/79505
37098 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37099 (new_oacc_loop_raw): Don't clear already cleared fields.
37102 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37103 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37104 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37106 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37109 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37111 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37114 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37115 the extra instruction to the right place to store 128-bit constant
37118 2017-02-14 Martin Sebor <msebor@redhat.com>
37120 PR middle-end/79448
37121 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37122 warning for strings of unknown length.
37124 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37126 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37128 2017-02-14 Jeff Law <law@redhat.com>
37131 * ira-costs.c (scan_one_insn): Initialize register move costs
37132 for pseudos seen in USE/CLOBBER insns.
37134 PR tree-optimization/79095
37135 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37136 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37137 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37138 if the operands are known to be not equal, then the resulting range
37140 (intersect_ranges): If the new range is ~[0,0] and the old range is
37141 wide, then prefer ~[0,0].
37142 * tree-vrp.c (overflow_comparison_p_1): New function.
37143 (overflow_comparison_p): New function.
37144 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37145 if NAME is used in an overflow test.
37146 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37147 overflow check that can be expressed as an equality test, then adjust
37148 ops to be that equality test.
37150 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37152 * config/s390/s390-builtin-types.def: Remove flags argument.
37153 * config/s390/s390.c (s390_init_builtins): Likewise.
37155 2017-02-14 Martin Liska <mliska@suse.cz>
37157 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37158 vector. Fix trailing white spaces.
37160 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37162 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37165 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37167 PR rtl-optimization/68664
37168 * config/arm/arm.c (arm_sched_can_speculate_insn):
37169 New function. Declare prototype.
37170 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37172 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37174 PR rtl-optimization/68664
37175 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37177 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37179 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37181 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37182 max skip bytes for function, loop and jump.
37184 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37186 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37187 ABS_EXPR for gimple dump.
37189 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37192 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37194 PR tree-optimization/79408
37195 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37196 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37197 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37198 also if rhs1 is INTEGER_CST.
37200 2017-02-14 Richard Biener <rguenther@suse.de>
37202 PR middle-end/79432
37203 * tree-into-ssa.c (insert_phi_nodes): When the function can
37204 have abnormal edges rewrite SSA names with broken use-def
37205 dominance out of SSA and register them for PHI insertion.
37207 2017-02-13 Martin Sebor <msebor@redhat.com>
37209 PR middle-end/79496
37210 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37211 clearing info.nowrite flag when snprintf size argument is a range.
37213 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37215 * cprop.c (cprop_jump): Add missing space in string literal.
37216 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37217 (get_constraint_for_component_ref): Likewise.
37218 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37219 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37220 * lra-constraints.c (process_alt_operands): Likewise.
37221 * ipa-inline.c (inline_small_functions): Likewise.
37222 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37223 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37224 * trans-mem.c (diagnose_tm_1_op): Likewise.
37225 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37226 (grid_parallel_clauses_gridifiable): Likewise.
37228 * config/nvptx/mkoffload.c (process): Add space in between
37231 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37232 "MOD4_SSE_REGS" and "ALL_REGS".
37234 * spellcheck.c (test_data): Add , in between "foo" and "food".
37236 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37239 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37240 boundary crossing check and subsequent code generation agree.
37242 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37244 * config/aarch64/aarch64.c (has_memory_op): Delete.
37245 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37248 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37250 PR rtl-optimization/79388
37251 PR rtl-optimization/79450
37252 * combine.c (distribute_notes): When removing TEM_INSN for which
37253 corresponding dest has last value recorded, invalidate that last
37256 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37258 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37259 of explicit '@'. Add missing assembly comment marker on branch costs
37262 2017-02-13 Nathan Sidwell <nathan@acm.org>
37264 * gengtype-lex.l (<in_struct>): Add '/'.
37266 2017-02-13 Martin Liska <mliska@suse.cz>
37269 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37271 2017-02-13 Richard Biener <rguenther@suse.de>
37273 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37275 * configure: Re-generate.
37276 * config.in: Likewise.
37277 * graphite-dependences.c: Simplify as if
37278 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37279 * graphite-isl-ast-to-gimple.c: Likewise.
37280 * graphite-optimize-isl.c: Likewise.
37281 * graphite-poly.c: Likewise.
37282 * graphite-sese-to-poly.c: Likewise.
37283 * graphite.h: Likewise.
37284 * toplev.c: Include isl/version.h and use isl_version () for
37285 printing the ISL version.
37286 * doc/install.texi: Update ISL requirement.
37288 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37290 * doc/standards.texi (Standards): Update reference to
37293 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37295 * doc/extend.texi (Named Address Spaces): sourceware.org now
37297 * doc/install.texi (Binaries): Ditto.
37300 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37302 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37303 standard terminology "stringize"/"stringizing" throughout.
37304 * doc/cppinternals.texi: Likewise.
37306 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37308 * doc/extend.texi: Fix some spelling mistakes and typos.
37309 * doc/invoke.texi: Likewise.
37311 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37314 * params.def (inline-min-speedup) Change from 10 to 8.
37316 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37318 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37321 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37324 * ipa-inline-analysis.c (get_minimal_bb): New function.
37325 (record_modified): Use it.
37326 (remap_edge_change_prob): Handle also ancestor functions.
37328 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37330 * doc/contrib.texi (Contributors): Remove broken link into
37331 the Mauve CVS repository.
37333 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37335 PR middle-end/79454
37336 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37337 result computation whenever lhs doesn't have vector mode, not
37338 just when it has BLKmode.
37340 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37342 * doc/makefile.texi (profiledbootstrap): Refer to the
37343 installation instructions only in textual form.
37345 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37348 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37350 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37352 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37353 (Specific): Update mingw-w64 reference.
37355 (Specific): Remove broken link to Renesas RX processor.
37357 2017-02-10 Richard Biener <rguenther@suse.de>
37359 * toplev.c (process_options): Do not mention obsolete graphite
37360 options when printing sorry message about missing graphite support.
37361 Mention -floop-nest-optimize.
37363 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37365 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37366 (vtst_p16): Likewise.
37367 (vtstq_p8): Likewise.
37368 (vtstq_p16): Likewise.
37370 (vtstq_p64): Likewise.
37371 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37372 (vset_lane_p64): New.
37373 (vsetq_lane_p64): New.
37375 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37377 PR tree-optimization/79411
37378 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37379 stmt operands are SSA_NAMEs used in abnormal phis.
37380 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37383 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37386 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37389 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37391 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37393 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37395 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37399 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37400 not arbitrary TREE_CONSTANT.
37403 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37404 "omp declare target link" attribute unless is_global_var.
37405 * omp-offload.c (find_link_var_op): Likewise.
37407 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37408 Chung-Lin Tang <cltang@codesourcery.com>
37410 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37412 (gimplify_adjust_omp_clauses): Don't delete TILE.
37413 (gimplify_omp_for): Deal with TILE.
37414 * internal-fn.c (expand_GOACC_TILE): New function.
37415 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37417 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37418 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37420 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37421 avoid DIV for outermost collapse var.
37422 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37423 Remove out of date comments, fix whitespace.
37424 * omp-general.c (omp_extract_for_data): Deal with tiling.
37425 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37426 adjust OLF_DIM_BASE value.
37427 (struct omp_for_data): Add tiling field.
37428 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37429 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37430 for auto loops. Remove default auto determining, moved to
37431 oacc_loop_fixed_partitions.
37432 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37433 stmts, add e_mask field.
37434 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37435 (oacc_thread_numbers): Use oacc_dim_call.
37436 (oacc_xform_tile): New.
37437 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37438 (finish_oacc_loop): Adjust for ifns vector.
37439 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37440 add case for GOACC_TILE fns.
37441 (oacc_loop_xform_loop): Delete.
37442 (oacc_loop_process): Iterate over call list directly, and add
37443 handling for GOACC_TILE fns.
37444 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37446 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37447 vector partitioning to outer loops. Assign 2 partitions to loops
37448 when available. Add TILE handling.
37449 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37450 (execite_oacc_device_lower): Process GOACC_TILE fns,
37451 ignore unknown specs.
37452 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37453 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37454 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37456 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37458 * configure.ac (ACX_BUGURL): Update.
37459 * configure: Regenerate.
37461 2017-02-09 Richard Biener <rguenther@suse.de>
37463 PR tree-optimization/69823
37464 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37465 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37467 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37469 * config/arc/arc-c.def: Add __NPS400__ definition.
37470 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37471 (TARGET_NPS400): Define.
37473 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37475 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37477 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37478 pointer, arch_info.
37479 (arc_cpu_types): Fill the arch_info field with a pointer into the
37480 arc_arch_types table.
37481 (arc_selected_cpu): Declare.
37482 * config/arc/arc.c (arc_selected_cpu): Make global.
37483 (arc_selected_arch): Delete.
37484 (arc_base_cpu): Delete.
37485 (arc_override_options): Remove references to deleted variables,
37486 update access to arch information.
37487 (ARC_OPT): Update access to arch information.
37488 (ARC_OPTX): Likewise.
37489 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37490 (TARGET_ARC600): Update access to arch information.
37491 (TARGET_ARC601): Likewise.
37492 (TARGET_ARC700): Likewise.
37493 (TARGET_EM): Likewise.
37494 (TARGET_HS): Likewise.
37495 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37498 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37501 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37502 condition/operands for integer GE/LE/GEU/LEU operations.
37504 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37506 PR translation/79397
37507 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37510 2017-02-08 Martin Jambor <mjambor@suse.cz>
37513 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37514 whether allocation happened.
37515 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37516 nothing was allocated.
37518 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37520 PR tree-optimization/79408
37521 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37522 constant, but SSA_NAME with a known integer range, use the minimum
37523 of that range instead of op1 to determine if modulo can be replaced
37524 with its first operand.
37526 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37528 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37530 2017-02-08 Richard Biener <rguenther@suse.de>
37532 PR tree-optimization/71824
37533 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37534 Check all loops contained in the merged region.
37536 2017-02-07 Andrew Pinski <apinski@cavium.com>
37538 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37540 2017-02-07 Andrew Pinski <apinski@cavium.com>
37542 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37543 (thunderxt88): Likewise.
37544 (thunderxt81): Disable LSE and change v8.1 to v8.
37545 (thunderxt83): Likewise.
37547 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37548 Richard Biener <rguenther@suse.de>
37550 PR middle-end/79399
37551 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37552 type from int to size_t.
37553 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37555 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37557 PR rtl-optimization/79386
37558 * cprop.c (bypass_conditional_jumps): Initialize
37559 bypass_last_basic_block already before splitting bbs after
37560 unconditional traps...
37561 (bypass_conditional_jumps): ... rather than here.
37564 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37565 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37566 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37567 fix -masm=intel patterns.
37569 2017-02-07 Richard Biener <rguenther@suse.de>
37571 PR tree-optimization/79256
37572 PR middle-end/79278
37573 * builtins.c (get_object_alignment_2): Use min_align_of_type
37574 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37575 and ADJUST_FIELD_ALIGN.
37577 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37579 * doc/tm.texi: Regenerate.
37580 * stor-layout.c (layout_decl): Adjust.
37581 (update_alignment_for_field): Likewise.
37582 (place_field): Likewise.
37583 (min_align_of_type): Likewise.
37584 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37585 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37586 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37587 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37588 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37589 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37590 * config/i386/i386.c (x86_field_alignment): Likewise.
37591 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37592 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37593 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37594 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37595 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37596 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37600 2017-01-30 Richard Biener <rguenther@suse.de>
37602 PR tree-optimization/79256
37603 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37604 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37607 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37609 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37610 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37611 builtins to SImode and emit a zero-extend, if necessary.
37613 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37615 * docs/invoke.texi (RISC-V Options): Alphabetize.
37617 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37619 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37622 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37624 * config/riscv/riscv.c: New file.
37625 * common/config/riscv/riscv-common.c: Likewise.
37626 * config.gcc: Likewise.
37627 * config/riscv/constraints.md: Likewise.
37628 * config/riscv/elf.h: Likewise.
37629 * config/riscv/generic.md: Likewise.
37630 * config/riscv/linux.h: Likewise.
37631 * config/riscv/multilib-generator: Likewise.
37632 * config/riscv/peephole.md: Likewise.
37633 * config/riscv/pic.md: Likewise.
37634 * config/riscv/predicates.md: Likewise.
37635 * config/riscv/riscv-builtins.c: Likewise.
37636 * config/riscv/riscv-c.c: Likewise.
37637 * config/riscv/riscv-ftypes.def: Likewise.
37638 * config/riscv/riscv-modes.def: Likewise.
37639 * config/riscv/riscv-opts.h: Likewise.
37640 * config/riscv/riscv-protos.h: Likewise.
37641 * config/riscv/riscv.h: Likewise.
37642 * config/riscv/riscv.md: Likewise.
37643 * config/riscv/riscv.opt: Likewise.
37644 * config/riscv/sync.md: Likewise.
37645 * config/riscv/t-elf-multilib: Likewise.
37646 * config/riscv/t-linux: Likewise.
37647 * config/riscv/t-linux-multilib: Likewise.
37648 * config/riscv/t-riscv: Likewise.
37649 * configure.ac: Likewise.
37650 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37651 Waterman as RISC-V maintainers.
37652 * doc/install.texi: Add RISC-V entries.
37653 * doc/invoke.texi: Add RISC-V options section.
37654 * doc/md.texi: Add RISC-V constraints section.
37655 * configure: Regenerated.
37657 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37660 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37661 false values to be constant vectors with all 0 or all 1 bits set.
37662 (vcondu<mode><mode>): Likewise.
37663 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37665 (fpmask_comparison_operator): Update comment.
37666 (vecint_comparison_operator): New predicate.
37667 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37668 vector conditionals when the true and false values are constant
37669 vectors with all 0 bits or all 1 bits set.
37671 2017-02-06 Martin Sebor <msebor@redhat.com>
37673 PR tree-optimization/79376
37674 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37676 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37678 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37679 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37680 to simplify split condition.
37682 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37684 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37685 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37688 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37690 PR rtl-optimization/68664
37691 * target.def (can_speculate_insn): New hook.
37692 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37693 * doc/tm.texi: Regenerate.
37694 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37695 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37696 (rs6000_sched_can_speculate_insn): New function.
37698 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37700 PR tree-optimization/79284
37701 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37702 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37703 vectorizable_mask_load_store, vectorizable_operation,
37704 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37705 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37706 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37707 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37708 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37709 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37710 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37711 is_gimple_assign (stmt). Replace another such test with
37712 is_gimple_assign (stmt).
37714 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37717 * config/avr/avr.c (rtl-iter.h): Include it.
37718 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37719 (avr_legitimate_combined_insn): ...and implementation.
37721 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37723 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37724 * config/s390/s390.c (s390_const_operand_ok)
37725 (s390_canonicalize_comparison, s390_extract_part)
37726 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37727 (s390_contiguous_bitmask_p, s390_rtx_costs)
37728 (legitimize_pic_address): Likewise.
37729 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37730 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37731 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37732 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37733 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37735 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37737 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37738 REGNO($0) == REGNO($1).
37740 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37742 * config/s390/linux.h(SIZE_TYPE): Add comment.
37744 2017-02-06 Julian Brown <julian@codesourcery.com>
37745 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37746 Virendra Pathak <virendra.pathak@broadcom.com>
37748 * config/aarch64/aarch64-cores.def: Change the scheduler
37750 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37751 * config/aarch64/thunderx2t99.md: New file.
37753 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37755 * doc/standards.texi (Go Language): Update link to language
37758 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37760 * tree-eh.c (lower_resx): Sanitize profile.
37761 (cleanup_empty_eh_move_lp): Likewise.
37763 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37766 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37768 * cfgloopmanip.h (loop_version): Update prototype.
37769 * modulo-sched.c (sms_schedule): Update call of loop_version.
37770 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37771 * tree-parloops.c (gen_parallel_loop): Likewise.
37772 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37773 * tree-ssa-loop-split.c (split_loop): Likewise.
37774 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37775 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37777 2017-02-05 Martin Liska <mliska@suse.cz>
37780 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37782 (print_operand_address): Initialize a struct to zero.
37784 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37786 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37787 garbage collector only in textual form.
37789 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37791 * doc/extend.texi (x86 specific memory model extensions for
37792 transactional memory): Simplify a phrase.
37794 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37797 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37798 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37799 (atomic_storedi_1): Likewise.
37801 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37803 PR tree-optimization/79338
37804 * tree-parloops.c (gather_scalar_reductions): Don't call
37805 vect_analyze_loop_form for loop->inner before destroying loop's
37808 2017-02-03 Martin Sebor <msebor@redhat.com>
37810 PR tree-optimization/79327
37811 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37812 when precision has resulted in leading zeros.
37813 (format_integer): Adjust the likely counter to assume an unknown
37814 argument that may be zero is non-zero.
37816 2017-02-03 Jason Merrill <jason@redhat.com>
37819 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37820 avoid copying non-taken branch.
37822 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37824 PR tree-optimization/79340
37825 * tree-vect-loop.c (vectorizable_reduction): Release
37826 vec_defs elements after safe_splicing them into other vectors.
37829 PR tree-optimization/79327
37830 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37831 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37833 (format_integer): Use wide_int_to_tree instead of build_int_cst
37834 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37835 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37836 of shortest and longest sequence.
37838 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37840 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37841 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37843 2017-02-03 Walter Lee <walt@tilera.com>
37846 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37847 after initial stackframe link reg save.
37848 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37850 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37853 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37854 wu for stxssp alternative.
37856 2017-02-03 Martin Sebor <msebor@redhat.com>
37858 PR tree-optimization/79352
37859 * gimple-fold.c (get_range_strlen): Add argument.
37860 (get_range_strlen): Change return type to bool.
37861 (get_maxval_strlen): Pass in a dummy argument.
37862 * gimple-fold.h (get_range_strlen): Change return type to bool.
37863 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
37864 * tree.h (array_at_struct_end_p): Add argument.
37865 * tree.c (array_at_struct_end_p): Handle it.
37867 2017-02-03 Martin Liska <mliska@suse.cz>
37870 * multiple_target.c (create_dispatcher_calls): Redirect edge
37871 from a caller of a dispatcher.
37872 (expand_target_clones): Make the clones local.
37873 (ipa_target_clone): Do both target clones and resolvers.
37874 (ipa_dispatcher_calls): Remove the pass.
37875 (pass_dispatcher_calls::gate): Likewise.
37876 (make_pass_dispatcher_calls): Likewise.
37877 * passes.def (pass_target_clone): Put as very first IPA early
37880 2017-02-03 Martin Liska <mliska@suse.cz>
37882 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
37883 in case of a function with ifunc attribute.
37885 2017-02-03 Martin Liska <mliska@suse.cz>
37887 * cgraph.c (cgraph_node::dump): Dump function version info.
37888 * symtab.c (symtab_node::dump_base): Add missing new line.
37890 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37892 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
37893 (ifcombine_ifandif): Use it.
37895 2017-02-03 Martin Liska <mliska@suse.cz>
37897 * doc/invoke.texi: Document default value for
37898 use-after-scope-direct-emission-threshold.
37900 2017-02-03 Martin Liska <mliska@suse.cz>
37902 PR tree-optimization/79339
37903 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
37904 (format_floating): Likewise.
37906 2017-02-03 Martin Liska <mliska@suse.cz>
37909 * ipa-prop.c (ipa_node_params_t::insert): Remove current
37911 (ipa_node_params_t::remove): Likewise.
37912 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
37913 initialization from removed ipa_node_params_t::insert.
37914 (ipa_node_params::~ipa_node_params): Move from removed
37915 ipa_node_params_t::release.
37916 * symbol-summary.h (symbol_summary::m_released): New member.
37917 Do not release a summary twice. Do not allow to call finalizer
37918 for types of a summary that live in GGC memory.
37920 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37922 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
37925 2017-02-02 Martin Sebor <msebor@redhat.com>
37927 PR middle-end/79275
37928 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
37929 (format_string): Tighten up the range of output for non-constant
37930 strings and correct the expected range for wide non-constant strings.
37932 2017-02-02 Martin Sebor <msebor@redhat.com>
37934 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
37936 PR middle-end/32003
37937 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
37939 (-fdump-tree-@var): Add to index and document how to come up
37940 with pass-specific option and dump file names.
37941 (-fdump-passes): Clarify where to look for output.
37943 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37945 PR middle-end/77445
37946 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
37947 statistics of the analyzed path; allow threading for speed when
37948 any of BBs along the path are optimized for speed.
37950 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
37952 PR middle-end/78468
37953 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
37954 settings of the virtual registers.
37957 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
37959 * explow.c (get_dynamic_stack_size): Take known alignment of stack
37960 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
37963 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37965 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
37966 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
37968 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37970 * config/s390/s390.md: Add missing comments with the expanded
37972 * config/s390/vector.md: Likewise.
37973 * config/s390/vx-builtins.md: Likewise.
37975 2017-02-02 Jakub Jelinek <jakub@redhat.com>
37978 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
37979 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
37980 conditions on a single line.
37982 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37984 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37985 __S390_VX__ to __VX__.
37987 2017-02-01 Andrew Pinski <apinski@cavium.com>
37989 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
37990 stmt_info to record_stmt_cost.
37991 (vect_get_known_peeling_cost): Pass stmt_info if known to
37993 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
37994 cpu_vector_cost field into
37995 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
37996 field into vec_int_stmt_cost and vec_fp_stmt_cost.
37997 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
37998 splitting of scalar_stmt_cost and vec_stmt_cost.
37999 (thunderx_vector_cost): Likewise.
38000 (cortexa57_vector_cost): LIkewise.
38001 (exynosm1_vector_cost): Likewise.
38002 (xgene1_vector_cost): Likewise.
38003 (thunderx2t99_vector_cost): Improve after the splitting of the two
38005 (aarch64_builtin_vectorization_cost): Update for the splitting of
38006 scalar_stmt_cost and vec_stmt_cost.
38008 2017-02-01 Torvald Riegel <triegel@redhat.com>
38009 Richard Henderson <rth@redhat.com>
38011 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38012 conditional on existance of a fast atomic load.
38013 * optabs-query.c (can_atomic_load_p): New function.
38014 * optabs-query.h (can_atomic_load_p): Declare it.
38015 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38016 no fast atomic load is available for the particular size of access.
38017 (expand_atomic_compare_and_swap): Likewise.
38018 (expand_atomic_load): Likewise.
38019 (expand_atomic_store): Likewise.
38020 (expand_atomic_fetch_op): Likewise.
38021 * testsuite/lib/target-supports.exp
38022 (check_effective_target_sync_int_128): Remove x86 because it provides
38023 no fast atomic load.
38024 (check_effective_target_sync_int_128_runtime): Likewise.
38026 2017-02-01 Richard Biener <rguenther@suse.de>
38028 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38029 (graphite_transform_loops): Provide opt-info for optimized nests.
38030 * tree-parloop.c (parallelize_loops): Provide opt-info for
38031 parallelized loops.
38033 2017-02-01 Richard Biener <rguenther@suse.de>
38035 PR middle-end/79315
38036 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38037 was not set before.
38039 2017-02-01 Richard Biener <rguenther@suse.de>
38041 PR tree-optimization/71824
38042 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38043 Verify the loops are valid in the merged SESE region.
38044 (scop_detection::can_represent_loop_1): Check analyzing the
38045 evolution of the number of iterations in the region succeeds.
38047 2017-01-31 Ian Lance Taylor <iant@golang.org>
38049 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38050 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38052 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38054 PR preprocessor/79210
38055 * input.c (get_substring_ranges_for_loc): Replace line_width
38056 assertion with error-handling.
38058 2017-01-31 Richard Biener <rguenther@suse.de>
38060 PR tree-optimization/77318
38061 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38062 (create_pw_aff_from_tree): Take loop parameter.
38063 (add_condition_to_pbb): Pass loop of the condition to
38064 create_pw_aff_from_tree.
38066 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38068 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38069 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38071 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38075 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38077 (convert_int_to_float128): Likewise.
38078 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38079 (convert_int_to_float128): Likewise.
38080 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38081 (UNSPEC_IEEE128_CONVERT): Likewise.
38082 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38083 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38084 Use local variables for IBM extended format.
38085 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38086 (fix_trunc<mode>si2_fprs): Likewise.
38087 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38088 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38089 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38090 to know that we can now have integers of all sizes in vector
38092 (fix<uns>_<mode>di2_hw): Likewise.
38093 (float<uns>_<mode>si2_hw): Likewise.
38094 (fix_<mode>si2_hw): Likewise.
38095 (fixuns_<mode>si2_hw): Likewise.
38096 (float<uns>_<mode>di2_hw): Likewise.
38097 (float_<mode>di2_hw): Likewise.
38098 (float_<mode>si2_hw): Likewise.
38099 (floatuns_<mode>di2_hw): Likewise.
38100 (floatuns_<mode>si2_hw): Likewise.
38101 (xscvqp<su>wz_<mode>): Delete, no longer used.
38102 (xscvqp<su>dz_<mode>): Likewise.
38103 (xscv<su>dqp_<mode>): Likewise.
38104 (ieee128_mfvsrd_64bit): Likewise.
38105 (ieee128_mfvsrd_32bit): Likewise.
38106 (ieee128_mfvsrwz): Likewise.
38107 (ieee128_mtvsrw): Likewise.
38108 (ieee128_mtvsrd_64bit): Likewise.
38109 (ieee128_mtvsrd_32bit): Likewise.
38111 2017-01-31 Martin Liska <mliska@suse.cz>
38114 * ipa-prop.c (ipa_free_all_node_params): Call release method
38115 instead of ~sumbol_summary to not to trigger double times
38118 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38120 PR tree-optimization/71691
38121 * bitmap.h (class auto_bitmap): New.
38122 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38123 is_maybe_undefined instead of ssa_undefined_value_p.
38125 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38127 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38128 __S390_ARCH_LEVEL__ to __ARCH__.
38130 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38132 PR tree-optimization/79267
38133 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38134 if should_remove_lhs_p is true.
38136 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38139 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38140 (add_alignment_attribute): New.
38141 (base_type_die): Add alignment attribute.
38142 (subrange_type_die): Likewise.
38143 (modified_type_die): Likewise.
38144 (gen_array_type_die): Likewise.
38145 (gen_descr_array_type_die: Likewise.
38146 (gen_enumeration_type_die): Likewise.
38147 (gen_subprogram_die): Likewise.
38148 (gen_variable_die): Likewise.
38149 (gen_field_die): Likewise.
38150 (gen_ptr_to_mbr_type_die): Likewise.
38151 (gen_struct_or_union_type_die): Likewise.
38152 (gen_subroutine_type_die): Likewise.
38153 (gen_typedef_die): Likewise.
38154 (base_type_cmp): Compare alignment attribute.
38156 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38159 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38160 (setb_unsigned) New pattern for setb with CCUNS.
38161 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38162 subfc./subfe sequence to avoid overflow problems. Generate a
38163 shorter sequence with cmpld/setb for power9.
38164 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38165 for generating subfc. instruction.
38166 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38167 now uses this instruction.
38169 2017-01-30 Ian Lance Taylor <iant@google.com>
38172 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38173 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38175 2017-01-30 Martin Sebor <msebor@redhat.com>
38177 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38178 Move constant to the right of a relational operator.
38179 (get_mpfr_format_length, format_character, format_string): Ditto.
38180 (should_warn_p, maybe_warn): Same.
38182 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38184 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38187 * asan.c (get_translation_unit_decl): Remove function.
38188 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38190 2017-01-30 Martin Liska <mliska@suse.cz>
38192 PR gcov-profile/79259
38193 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38194 -fprofile-generate.
38196 2017-01-30 Martin Liska <mliska@suse.cz>
38199 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38200 Initialize variables with NULL value.
38202 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38205 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38207 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38209 2017-01-30 Richard Biener <rguenther@suse.de>
38211 PR tree-optimization/79276
38212 * tree-vrp.c (process_assert_insertions): Properly adjust common
38213 when removing a duplicate.
38215 2017-01-30 Richard Biener <rguenther@suse.de>
38217 PR tree-optimization/79256
38218 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38219 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38221 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38223 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38226 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38227 ("*r<noxa>sbg_<mode>_sll_bitmask")
38228 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38229 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38230 Use contiguous_bitmask_nowrap_operand.
38232 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38235 * config/rs6000/altivec.h (vec_xl): Revise #define.
38236 (vec_xst): Likewise.
38238 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38240 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38242 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38244 PR rtl-optimization/79194
38245 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38246 traps before call to bypass_conditional_jumps.
38248 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38250 PR tree-optimization/71374
38251 * lra-constraints.c (check_conflict_input_operands): New.
38252 (match_reload): Use it.
38254 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38257 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38258 account to calculate conflict_set.
38260 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38262 PR rtl-optimization/78559
38263 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38264 other_insn in combine.
38266 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38268 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38269 uint16_type_node for BT_UINT16.
38271 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38273 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38274 "RTL Tests" to menu.
38275 (GIMPLE Tests): New node.
38276 (RTL Tests): New node.
38278 2017-01-27 Richard Biener <rguenther@suse.de>
38280 PR tree-optimization/79245
38281 * tree-loop-distribution.c (distribute_loop): Apply cost
38282 modeling also to detected patterns.
38284 2017-01-27 Richard Biener <rguenther@suse.de>
38286 PR tree-optimization/71433
38287 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38288 (compare_assert_loc): New function.
38289 (process_assert_insertions): Sort and optimize assert locations
38290 to remove duplicates and push down identical assertions on
38291 edges to their destination block.
38293 2017-01-27 Richard Biener <rguenther@suse.de>
38295 PR tree-optimization/79244
38296 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38297 out SSA names even if abnormal.
38299 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38301 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38302 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38303 instead of MPFR_RNDN.
38305 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38308 * arm.c (arm_option_override): Don't call build_target_option_node
38309 until after doing all option overrides.
38310 (arm_valid_target_attribute_tree): Likewise.
38312 2017-01-27 Martin Liska <mliska@suse.cz>
38314 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38315 for {cd}tors and C++ {cd}tors.
38317 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38319 * config/s390/s390.md ("*setmem_long_and")
38320 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38322 2017-01-26 Martin Sebor <msebor@redhat.com>
38324 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38327 2017-01-26 Martin Sebor <msebor@redhat.com>
38329 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38330 HAVE_DFmode before using XFmode or DFmode.
38331 (parse_directive): Avoid using the z length modifier to avoid
38332 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38334 PR middle-end/78703
38335 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38336 to accept adjustment as an array.
38337 (get_int_range): New function.
38338 (struct directive): Make width and prec arrays.
38339 (directive::set_width, directive::set_precision): Call get_int_range.
38340 (format_integer, format_floating): Handle width and precision ranges.
38341 (format_string, parse_directive): Same.
38343 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38346 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38347 comdat_type_p set, just clone them, but keep the children in the
38351 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38352 which have direct callers with -fvar-tracking-assignments enabled
38354 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38355 inside of type units.
38357 2017-01-26 Martin Sebor <msebor@redhat.com>
38359 PR middle-end/78703
38360 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38362 (struct format_result): Replace number_chars, number_chars_min,
38363 and number_chars_max with a single member of struct result_range.
38365 (format_result::operator+=): Adjust.
38366 (struct fmtresult): Remove bounded. Handle likely and unlikely
38368 (fmtresult::adjust_for_width_or_precision): New function.
38369 (fmtresult:type_max_digits): New function.
38370 (bytes_remaining): Handle likely and unlikely counters.
38371 (min_bytes_remaining): Remove.
38372 (format_percent): Simplify.
38373 (format_integer, format_floating): Set likely and unlikely counters.
38374 (get_string_length, format_character, format_string): Same.
38375 (format_plain, should_warn_p): New function.
38376 (maybe_warn): Call should_warn_p. Update diagnostic messages
38377 and handle those for all directives, including plain strings.
38378 (format_directive): Handle likely and unlikely counters.
38379 Remove unnecessary quoting from diagnostics. Add an informational
38381 (add_bytes): Remove.
38382 (pass_sprintf_length::compute_format_length): Simplify.
38383 (try_substitute_return_value): Handle likely and unlikely counters.
38385 2017-01-26 Carl Love <cel@us.ibm.com>
38387 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38388 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38390 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38393 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38394 endianess for subregs into account.
38395 * lra-constraints.c (lra_constraints): Do risky transformations
38396 always on the first iteration.
38397 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38398 last_call_used_reg_set.
38399 (process_bb_lives): Define and use last_call_used_reg_set.
38400 * lra.c (lra): Always continue after lra_constraints on the first
38403 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38405 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38407 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38409 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38411 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38412 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38413 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38414 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38415 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38416 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38417 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38418 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38419 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38421 2017-01-26 Marek Polacek <polacek@redhat.com>
38424 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38425 for the third operand.
38427 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38429 PR middle-end/79236
38430 * omp-low.c (struct omp_context): Add simt_stmt field.
38431 (scan_omp_for): Return omp_context *.
38432 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38433 context to the _simt_ SIMD stmt.
38434 (lower_omp_for): For combined SIMD with sibling _simt_
38435 SIMD, make sure to use the same decls in _looptemp_
38436 clauses as in the sibling.
38438 2017-01-26 David Sherwood <david.sherwood@arm.com>
38440 PR middle-end/79212
38441 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38444 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38447 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38448 emit fld b; fld a; if possible.
38450 * brig-builtins.def: Update copyright years.
38451 * config/arm/arm_acle_builtins.def: Update copyright years.
38453 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38456 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38457 constraint instead of o for the stxsd instruction.
38459 2017-01-25 Carl Love <cel@us.ibm.com>
38461 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38462 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38464 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38466 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38468 2017-01-25 Richard Biener <rguenther@suse.de>
38470 PR tree-optimization/69264
38471 * target.def (vector_alignment_reachable): Improve documentation.
38472 * doc/tm.texi: Regenerate.
38473 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38475 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38476 earlier changes with respect to TYPE_USER_ALIGN.
38477 (vector_alignment_reachable_p): Likewise. Improve dumping.
38479 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38482 * config/arm/arm.md (xordi3): Force constant operand into a register
38485 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38487 * doc/invoke.texi (-fstore-merging): Correct default optimization
38488 levels at which it is enabled.
38489 (-O): Move -fstore-merging from list to...
38492 2017-01-25 Richard Biener <rguenther@suse.de>
38495 * omp-expand.c: Include debug.h.
38496 (expand_omp_taskreg): Make sure to generate early debug before
38497 outlining anything from a function.
38498 (expand_omp_target): Likewise.
38499 (grid_expand_target_grid_body): Likewise.
38501 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38504 * asan.c (get_translation_unit_decl): New function.
38505 (asan_add_global): Extract modules file name from globals
38506 TRANSLATION_UNIT_DECL name.
38508 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38511 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38512 for long calls with APCS frame and VFP.
38514 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38516 * cfg.c (original_copy_tables_initialized_p): New function.
38517 * cfg.h (original_copy_tables_initialized_p): New decl.
38518 * cfgrtl.c (relink_block_chain): Guard the call to
38519 free_original_copy_tables with a call to
38520 original_copy_tables_initialized_p.
38521 * cgraph.h (symtab_node::native_rtl_p): New decl.
38522 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38523 (symtab_node::needed_p): Don't assert for early assembly output
38524 for __RTL functions.
38525 (cgraph_node::finalize_function): Set "force_output" for __RTL
38527 (cgraph_node::analyze): Bail out early for __RTL functions.
38528 (analyze_functions): Update assertion to support __RTL functions.
38529 (cgraph_node::expand): Bail out early for __RTL functions.
38530 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38532 * function.h (struct function): Update comment for field
38534 * gimple-expr.c: Include "tree-pass.h".
38535 (gimple_has_body_p): Return false for __RTL functions.
38536 * Makefile.in (OBJS): Add run-rtl-passes.o.
38537 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38539 (gcc::pass_manager::get_clean_slate): New accessor.
38540 * passes.c: Include "insn-addr.h".
38541 (should_skip_pass_p): Add logging. Update logic for running
38542 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38543 property-provider override so it is only done for gimple passes.
38545 (skip_pass): New function.
38546 (execute_one_pass): Call skip_pass when skipping passes.
38547 * read-md.c (md_reader::read_char): Support filtering
38548 the input to a subset of line numbers.
38549 (md_reader::md_reader): Initialize fields
38550 m_first_line and m_last_line.
38551 (md_reader::read_file_fragment): New function.
38552 * read-md.h (md_reader::read_file_fragment): New decl.
38553 (md_reader::m_first_line): New field.
38554 (md_reader::m_last_line): New field.
38555 * read-rtl-function.c (function_reader::create_function): Only
38556 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38557 curr_properties. Set DECL_INITIAL to a dummy block.
38558 (read_rtl_function_body_from_file_range): New function.
38559 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38561 * run-rtl-passes.c: New file.
38562 * run-rtl-passes.h: New file.
38564 2017-01-24 Jeff Law <law@redhat.com>
38566 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38569 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38571 PR tree-optimization/79159
38572 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38573 (record_nonwrapping_iv): Improve boundary using above function if no
38574 value range information.
38576 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38577 Martin Jambor <mjambor@suse.cz>
38579 * brig-builtins.def: New file.
38580 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38581 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38582 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38583 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38584 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38585 * builtin-types.def (BT_INT8): New.
38586 (BT_INT16): Likewise.
38587 (BT_UINT8): Likewise.
38588 (BT_UINT16): Likewise.
38589 (BT_FN_ULONG): Likewise.
38590 (BT_FN_UINT_INT): Likewise.
38591 (BT_FN_UINT_ULONG): Likewise.
38592 (BT_FN_UINT_LONG): Likewise.
38593 (BT_FN_UINT_PTR): Likewise.
38594 (BT_FN_ULONG_PTR): Likewise.
38595 (BT_FN_INT8_FLOAT): Likewise.
38596 (BT_FN_INT16_FLOAT): Likewise.
38597 (BT_FN_UINT32_FLOAT): Likewise.
38598 (BT_FN_UINT16_FLOAT): Likewise.
38599 (BT_FN_UINT8_FLOAT): Likewise.
38600 (BT_FN_UINT64_FLOAT): Likewise.
38601 (BT_FN_UINT16_UINT32): Likewise.
38602 (BT_FN_UINT32_UINT16): Likewise.
38603 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38604 (BT_FN_INT_PTR_INT): Likewise.
38605 (BT_FN_UINT_PTR_UINT): Likewise.
38606 (BT_FN_LONG_PTR_LONG): Likewise.
38607 (BT_FN_ULONG_PTR_ULONG): Likewise.
38608 (BT_FN_VOID_UINT64_UINT64): Likewise.
38609 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38610 (BT_FN_INT8_INT8_INT8): Likewise.
38611 (BT_FN_INT16_INT16_INT16): Likewise.
38612 (BT_FN_INT_INT_INT): Likewise.
38613 (BT_FN_UINT_FLOAT_UINT): Likewise.
38614 (BT_FN_FLOAT_UINT_UINT): Likewise.
38615 (BT_FN_ULONG_UINT_UINT): Likewise.
38616 (BT_FN_ULONG_UINT_PTR): Likewise.
38617 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38618 (BT_FN_UINT_UINT_UINT): Likewise.
38619 (BT_FN_VOID_UINT_PTR): Likewise.
38620 (BT_FN_UINT_UINT_PTR: Likewise.
38621 (BT_FN_UINT32_UINT64_PTR): Likewise.
38622 (BT_FN_INT_INT_UINT_UINT): Likewise.
38623 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38624 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38625 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38626 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38627 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38628 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38629 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38630 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38631 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38632 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38633 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38634 * doc/frontends.texi: List BRIG FE.
38635 * doc/install.texi (Testing): Add BRIG tesring requirements.
38636 * doc/invoke.texi (Overall Options): Mention BRIG.
38637 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38639 2017-01-24 Richard Biener <rguenther@suse.de>
38641 PR translation/79208
38642 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38644 2017-01-24 Martin Jambor <mjambor@suse.cz>
38647 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38648 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38649 and known_contexts.
38651 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38653 PR middle-end/79123
38654 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38655 casts from signed to unsigned really don't have a range.
38657 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38659 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38660 GMP_RNDx for compatiblity.
38662 2017-01-24 Martin Liska <mliska@suse.cz>
38665 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38666 that would prevent us to call alloca with -1 as argument.
38668 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38670 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38671 Avoid -Wformat-security warning.
38673 2017-01-23 Andrew Pinski <apinski@cavium.com>
38675 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38678 2017-01-23 Martin Sebor <msebor@redhat.com>
38680 PR middle-end/78703
38681 * gimple-ssa-sprintf.c (warn_level): New global.
38682 (format_integer): Use it here and throughout the rest of the file.
38683 Use the same switch to compute sign as base.
38684 (maybe_warn): New function.
38685 (format_directive): Factor out warnings into maybe_warn.
38686 Add debugging output. Use warn_level.
38687 (add_bytes): Use warn_level.
38688 (pass_sprintf_length::compute_format_length): Add debugging output.
38689 (try_substitute_return_value): Same.
38690 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38692 PR middle-end/78703
38693 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38694 (struct fmtresult, format_integer, format_floating): Adjust.
38695 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38696 (get_string_length, format_string,format_directive): Same.
38697 (pass_sprintf_length::compute_format_length): Same.
38698 (try_substitute_return_value): Simplify slightly.
38700 PR middle-end/78703
38701 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38702 (fmtresult::operator+=): Outlined.
38703 (struct fmtresult): Add ctors.
38704 (struct conversion_spec): Rename...
38705 (struct directive): ...to this. Add and remove data members.
38706 (directive::set_width, directive::set_precision): New functions.
38707 (format_percent): Use fmtresult ctor.
38708 (get_width_and_precision): Remove.
38709 (format_integer): Make naming changes. Avoid computing width and
38711 (format_floating): Same. Adjust indentation.
38712 (format_character, format_none): New functions.
38713 (format_string): Moved character handling to format_character.
38714 (format_directive): Remove arguments, change return type.
38715 (parse_directive): New function.
38716 (pass_sprintf_length::compute_format_length): Move directive
38717 parsing to parse_directive.
38719 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38721 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38722 (assign_assembler_name_if_needed): ... this.
38723 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38724 (assign_assembler_name_if_needed): ... this.
38725 (free_lang_data_in_cgraph): Adjust callers.
38726 * cgraphunit.c (cgraph_node::analyze): Likewise.
38727 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38730 2017-01-23 Richard Biener <rguenther@suse.de>
38732 PR tree-optimization/79088
38733 PR tree-optimization/79188
38734 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38735 resetting loop bounds after last path deletion. Reset loop
38736 bounds of the target loop, make code match the comments.
38737 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38738 Make sure loops need no fixups.
38740 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38742 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38743 exponent support with double type for first argument.
38744 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38745 type returned by __builtin_vec_extract_sig,
38746 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38747 functions from "vector int" to "vector unsigned int" or from
38748 "vector long long int" to "vector unsigned long long int".
38749 Changed type returned by __builtin_vec_extract_exp,
38750 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38751 functions from "vector int" to "vector unsigned int" or from
38752 "vector long long int" to "vector unsigned long long int".
38753 Changed return type of __builtin_vec_test_data_class,
38754 __builtin_vec_test_data_class_sp, and
38755 __builtin_vec_test_data_class_dp from "vector int" to
38756 "vector bool int" or from "vector long long int" to "vector bool
38757 long long int" and changed second argument type from "unsigned
38758 int" to "int". Added new overloaded function forms "vector float
38759 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38760 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38761 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38762 double, vector unsigned long long int)" and "vector double
38763 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38764 long int)". Changed return type of
38765 __builtin_scalar_test_data_class and
38766 __builtin_scalar_test_data_class_sp and
38767 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38768 int" and changed second argument from "unsigned int" to "int".
38769 Changed type returned by __builtin_scalar_test_neg,
38770 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38771 from "int" to "bool int". Added new overloaded function form
38772 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38773 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38774 exponent double-precision with floating point first argument.
38775 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38776 documentation of scalar_test_data_class, scalar_test_neg,
38777 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38778 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38779 vec_test_data_class built-in functions to reflect refinements in
38780 their type signatures.
38782 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38784 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38786 (aarch64_elf_asm_destructor): Likewise.
38788 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38790 PR rtl-optimization/78634
38791 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38792 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38793 * ifcvt.c (noce_try_cmove): Add missing cost check.
38795 PR rtl-optimization/71724
38796 * combine.c (if_then_else_cond): Look for situations where it is
38797 beneficial to undo the work of one of the recursive calls.
38799 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38801 PR tree-optimization/70754
38802 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38803 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38804 combined stmt before it if not NULL.
38805 (combine_chains): Process refs reversely and compute dominance point
38808 2017-01-23 Martin Liska <mliska@suse.cz>
38810 PR tree-optimization/79196
38811 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38812 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38814 (strlen_optimize_stmt): Call the renamed function.
38816 2017-01-23 Michael Matz <matz@suse.de>
38818 PR tree-optimization/78384
38819 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38821 2017-01-23 Richard Biener <rguenther@suse.de>
38823 PR tree-optimization/79186
38824 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38825 both incoming edges before moving an assert.
38827 2017-01-23 Martin Jambor <mjambor@suse.cz>
38829 * ipa-prop.c (load_from_param_1): Removed.
38830 (load_from_unmodified_param): Bits from load_from_param_1 put back
38832 (load_from_param): Removed.
38833 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38834 with stmt. Reverted back to use of load_from_unmodified_param.
38836 2017-01-23 Martin Jambor <mjambor@suse.cz>
38839 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38840 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38841 field a pointer to garbage collected vector, mark lattices and
38842 ipcp_orig_node with GTY((skip)).
38843 (ipa_get_param_count): Adjust to descriptors being a pointer.
38844 (ipa_get_param): Likewise.
38845 (ipa_get_type): Likewise.
38846 (ipa_get_param_move_cost): Likewise.
38847 (ipa_set_param_used): Likewise.
38848 (ipa_get_controlled_uses): Likewise.
38849 (ipa_set_controlled_uses): Likewise.
38850 (ipa_is_param_used): Likewise.
38851 (ipa_node_params_t): Move into garbage collector. New methods insert
38853 (ipa_node_params_sum): Annotate wth GTY(()).
38854 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38856 (ipa_load_from_parm_agg): Adjust declaration.
38857 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38858 * ipa-profile.c (ipa_profile): Likewise.
38859 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
38860 (ipa_populate_param_decls): Make descriptors parameter garbage
38862 (ipa_dump_param): Adjust to descriptors being a pointer.
38863 (ipa_alloc_node_params): Likewise.
38864 (ipa_initialize_node_params): Likewise.
38865 (load_from_param_1): Make descriptors parameter garbage collected.
38866 (load_from_unmodified_param): Likewise.
38867 (load_from_param): Likewise.
38868 (ipa_load_from_parm_agg): Likewise.
38869 (ipa_node_params::~ipa_node_params): Removed.
38870 (ipa_free_all_node_params): Remove call to delete operator.
38871 (ipa_node_params_t::insert): New.
38872 (ipa_node_params_t::remove): Likewise.
38873 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
38874 copy known_csts and known_contexts vectors.
38875 (ipa_read_node_info): Adjust to descriptors being a pointer.
38876 (ipcp_modif_dom_walker): Make m_descriptors field garbage
38878 (ipcp_transform_function): Make descriptors variable garbage
38881 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
38883 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
38884 * config/i386/avx512dqintrin.h: Ditto.
38885 * config/i386/avx512fintrin.h: Ditto.
38886 * config/i386/i386.c: Handle new builtins.
38887 * config/i386/i386-builtin.def: Add new builtins.
38888 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
38889 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
38891 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38892 Martin Liska <mliska@suse.cz>
38894 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
38895 * asan.c (asan_expand_poison_ifn): Support stores and use
38896 appropriate ASAN report function.
38897 * internal-fn.c (expand_ASAN_POISON_USE): New function.
38898 * internal-fn.def (ASAN_POISON_USE): Declare.
38899 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
38900 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
38901 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
38902 ASAN_POISON calls w/o LHS.
38903 * tree-ssa.c (execute_update_addresses_taken): Create clobber
38904 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
38905 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
38906 * gimplify.c (asan_poison_variables): Add attribute
38907 use_after_scope_memory to variables that really needs to live
38909 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
38910 having the attribute.
38912 2017-01-23 Martin Liska <mliska@suse.cz>
38914 * asan.c (create_asan_shadow_var): New function.
38915 (asan_expand_poison_ifn): Likewise.
38916 * asan.h (asan_expand_poison_ifn): New declaration.
38917 * internal-fn.c (expand_ASAN_POISON): Likewise.
38918 * internal-fn.def (ASAN_POISON): New builtin.
38919 * sanopt.c (pass_sanopt::execute): Expand
38920 asan_expand_poison_ifn.
38921 * tree-inline.c (copy_decl_for_dup_finish): Make function
38923 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
38924 * tree-ssa.c (is_asan_mark_p): New function.
38925 (execute_update_addresses_taken): Rewrite local variables
38926 (identified just by use-after-scope as addressable) into SSA.
38928 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38930 * doc/install.texi (Specific): opensource.apple.com uses https
38931 now. Remove trailing slash.
38933 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38935 * README.Portability: Remove note on an Irix compatibility issue.
38937 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
38939 * gcov.c (INCLUDE_ALGORITHM): Define.
38940 (INCLUDE_VECTOR): Define.
38941 No longer include <vector> and <algorithm> directly.
38943 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
38945 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
38947 * doc/invoke.texi (Code Gen Options): Ditto.
38949 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
38952 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
38954 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
38956 rtl-optimization/79125
38957 * cprop.c (local_cprop_pass): Handle cases where we make an
38958 unconditional trap.
38960 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
38964 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
38965 read from, for big endian.
38967 2017-01-20 Jiong Wang <jiong.wang@arm.com>
38969 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
38970 register pauth builtins for LP64 only.
38972 2017-01-20 Marek Polacek <polacek@redhat.com>
38975 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
38978 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38980 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
38982 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
38983 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
38984 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
38986 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38989 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
38990 in big-endian mode when they are not a single duplicated value.
38992 2017-01-20 Richard Biener <rguenther@suse.de>
38994 * BASE-VER: Bump to 7.0.1.
38996 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38998 * omp-low.c (omplow_simd_context): New struct. Use it...
38999 (lower_rec_simd_input_clauses): ...here and...
39000 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39001 references to idx, lane, max_vf, is_simt.
39003 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39005 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39008 2017-01-20 Martin Jambor <mjambor@suse.cz>
39010 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39011 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39013 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39014 (GTFILES): Rename hsa.c to hsa-common.c.
39015 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39016 * hsa-dump.c: Likewise.
39017 * hsa-gen.c: Likewise.
39018 * hsa-regalloc.c: Likewise.
39019 * ipa-hsa.c: Likewise.
39020 * omp-expand.c: Likewise.
39021 * omp-low.c: Likewise.
39022 * toplev.c: Likewise.
39024 2017-01-20 Marek Polacek <polacek@redhat.com>
39027 * doc/invoke.texi: Document -Wduplicated-branches.
39028 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39029 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39030 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39031 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39032 return 0 only when not OEP_LEXICOGRAPHIC.
39033 (fold_build_cleanup_point_expr): Use the expression
39034 location when building CLEANUP_POINT_EXPR.
39035 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39036 * tree.c (add_expr): Handle error_mark_node.
39038 2017-01-20 Martin Liska <mliska@suse.cz>
39041 * tree-profile.c (init_ic_make_global_vars): Do not call
39043 (gimple_init_gcov_profiler): Likewise.
39045 2017-01-20 Martin Liska <mliska@suse.cz>
39048 * cgraph.h (maybe_create_reference): Remove argument and
39050 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39052 * ipa-cp.c (create_specialized_node): Likewise.
39053 * symtab.c (symtab_node::maybe_create_reference): Handle
39054 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39056 2017-01-20 Martin Liska <mliska@suse.cz>
39058 * read-rtl-function.c (function_reader::create_function): Use
39059 build_decl instread of build_decl_stat.
39061 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39063 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39064 * config/i386/avx512dqintrin.h: Ditto.
39065 * config/i386/avx512fintrin.h: Ditto.
39066 * config/i386/i386-builtin-types.def: Add new types.
39067 * config/i386/i386.c: Handle new types.
39068 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39069 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39070 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39071 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39072 (__builtin_ia32_kshiftridi): New.
39073 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39075 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39079 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39080 define to rs6000_init_stack_protect_guard.
39081 (rs6000_init_stack_protect_guard): New function.
39083 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39084 Yunqiang Su <yunqiang.su@imgtec.com>
39086 * config.gcc (supported_defaults): Add madd4.
39087 (with_madd4): Add validation.
39088 (all_defaults): Add madd4.
39089 * config/mips/mips.opt (mmadd4): New option.
39090 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39092 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39094 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39095 (ISA_HAS_FUSED_MADD4): Likewise.
39096 * doc/invoke.texi (-mmadd4): Document the new option.
39097 * doc/install.texi (--with-madd4): Document the new option.
39099 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39101 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39102 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39103 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39104 (aarch64_init_pauth_hint_builtins): New.
39105 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39106 (aarch64_expand_builtin): Expand new builtins.
39108 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39110 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39111 * combine-stack-adj.c (no_unhandled_cfa): Handle
39112 REG_CFA_TOGGLE_RA_MANGLE.
39113 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39114 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39115 info for return address signing.
39116 (aarch64_expand_epilogue): Likewise.
39118 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39120 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39121 * config/aarch64/aarch64-protos.h
39122 (aarch64_return_address_signing_enabled): New declaration.
39123 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39125 (aarch64_expand_prologue): Sign return address before it's pushed onto
39127 (aarch64_expand_epilogue): Authenticate return address fetched from
39129 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39130 (aarch64_attributes): New function attributes for "sign-return-address".
39131 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39132 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39133 ("*do_return"): Generate combined instructions according to key index.
39134 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39135 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39137 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39138 * config/aarch64/aarch64.opt (msign-return-address=): New.
39139 * doc/extend.texi (AArch64 Function Attributes): Documents
39140 "sign-return-address=".
39141 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39143 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39145 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39146 overall option summary.
39148 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39150 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39151 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39152 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39153 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39155 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39157 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39158 -mpower9-minmax by default for -mcpu=power9.
39159 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39160 128-bit floating point.
39162 2017-01-20 Alan Modra <amodra@gmail.com>
39164 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39165 optimizing for size.
39167 2017-01-20 Alan Modra <amodra@gmail.com>
39170 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39171 for strcmp and strncmp from corresponding builtin decl.
39173 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39175 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39176 instead of i386/rtems-64.h.
39177 * config/i386/rtems-64.h: Remove.
39179 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39183 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39185 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39187 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39189 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39190 Change int to HOST_WIDE_INT.
39191 * config/aarch64/aarch64-protos.h
39192 (aarch64_simd_gen_const_vector_dup): Likewise.
39193 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39195 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39197 * langhooks-def.h (lhd_type_for_size): New decl.
39198 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39199 * langhooks.c (lhd_type_for_size): New function, taken from
39202 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39204 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39205 define_bypass for CR latency.
39206 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39207 (power9-alu2): Add define_bypass for CR latency.
39209 (power9-mul): Update insn latency.
39210 (power9-mul-compare): Update insn latency, bypass latency and remove
39213 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39215 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39217 * config/aarch64/aarch64.md
39218 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39219 aarch64_nopcrelative_literal_loads.
39220 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39222 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39224 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39225 TARGET_LOONGSON_3A.
39226 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39228 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39231 * config.gcc (supported_defaults): Add lxc1-sxc1.
39232 (with_lxc1_sxc1): Add validation.
39233 (all_defaults): Add lxc1-sxc1.
39234 * config/mips/mips.opt (mlxc1-sxc1): New option.
39235 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39237 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39238 __mips_no_lxc1_sxc1.
39239 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39240 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39241 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39243 2017-01-19 Richard Biener <rguenther@suse.de>
39245 PR tree-optimization/72488
39246 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39247 sure to restore SSA info.
39248 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39250 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39252 PR rtl-optimization/79121
39253 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39254 of the inner type when shifting an extended value.
39256 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39259 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39260 interposable aliases.
39262 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39265 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39266 Use the evmergelohi instruction.
39267 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39268 (mov_sitf_e500_subreg8_2_be): Likewise.
39269 (mov_sitf_e500_subreg12_2_le): Likewise.
39270 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39271 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39272 (mov_sitf_e500_subreg8_2_le): Likewise.
39273 (mov_sitf_e500_subreg12_2_be): Likewise.
39275 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39277 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39278 attribute from vecsimple to vecperm.
39279 (altivec_vbpermq2): Likewise.
39281 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39284 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39286 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39287 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39288 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39289 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39290 case where N arg is SIZE_MAX.
39291 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39292 (cmpstrsi): Add pattern.
39294 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39296 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39297 __builtin_vec_revb builtins.
39298 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39299 built-in functions to support generation of the ISA 3.0 XXBR<x>
39300 vector byte reverse instructions.
39301 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39302 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39303 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39304 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39305 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39306 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39307 (P9V_BUILTIN_VEC_REVB): Likewise.
39308 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39309 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39310 (p9_xxbrq_v16qi): Likewise.
39311 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39312 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39313 (p9_xxbrh_v8hi): Likewise.
39314 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39315 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39316 vec_revb built-in functions.
39318 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39320 PR rtl-optimization/78952
39321 * config/i386/i386.md (any_extract): New code iterator.
39322 (*insvqi_2): Use any_extract for source operand.
39323 (*insvqi_3): Use any_shiftrt for source operand.
39325 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39327 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39329 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39331 2017-01-18 Matthias Klose <doko@ubuntu.com>
39333 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39335 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39337 * config/rs6000/altivec.h (vec_bperm): Change #define.
39338 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39339 (altivec_vbpermq2): New define_insn.
39340 (altivec_vbpermd): Likewise.
39341 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39342 function interface.
39343 (VBPERMD): Likewise.
39344 (VBPERM): New polymorphic function interface.
39345 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39346 Add entries for P9V_BUILTIN_VEC_VBPERM.
39347 * doc/extend.texi: Add interfaces for vec_bperm.
39349 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39351 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39352 first letter of error messages.
39353 (s390_resolve_overloaded_builtin): Likewise.
39354 * config/s390/s390.c (s390_expand_builtin): Likewise.
39355 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39356 (s390_valid_target_attribute_inner_p): Likewise.
39357 * config/s390/s390.md ("tabort"): Likewise.
39359 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39361 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39362 (ISA_AVOID_DIV_HILO): New macro.
39363 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39364 (ISA_HAS_DDIV): Likewise.
39366 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39368 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39370 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39372 * doc/invoke.texi (fabi-version): Spelling fix.
39374 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39377 * doc/invoke.texi (fabi-version): Mention mangling fix for
39380 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39383 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39385 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39388 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39389 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39391 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39393 (stack_protect_test): Ditto.
39394 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39395 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39397 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39398 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39399 -mstack-protector-guard-offset=.
39400 (RS/6000 and PowerPC Options): Ditto.
39402 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39404 * config/i386/i386.h (MASK_CLASS_P): New define.
39405 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39406 there are no registers from different register sets also when
39407 mask registers are used. Update function comment.
39408 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39409 to (*k/*r) and (*k/*km) alternatives.
39411 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39413 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39414 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39415 (EH_RETURN_HANDLER_RTX): New define.
39416 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39417 Force frame pointer in EH return functions.
39418 (aarch64_expand_epilogue): Add barrier for eh_return.
39419 (aarch64_final_eh_return_addr): Remove.
39420 (aarch64_eh_return_handler_rtx): New function.
39421 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39423 (aarch64_eh_return_handler_rtx): New prototype.
39425 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39427 * config/rs6000/altivec.h (vec_rlmi): New #define.
39428 (vec_vrlnm): Likewise.
39429 (vec_rlnm): Likewise.
39430 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39431 (UNSPEC_VRLNM): Likewise.
39432 (VIlong): New mode iterator.
39433 (altivec_vrl<VI_char>mi): New define_insn.
39434 (altivec_vrl<VI_char>nm): Likewise.
39435 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39437 (VRLDNM): Likewise.
39438 (RLNM): New polymorphic function entry.
39439 (VRLWMI): New monomorphic function entry.
39440 (VRLDMI): Likewise.
39441 (RLMI): New polymorphic function entry.
39442 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39443 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39444 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39447 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39450 * dwarf2out.c (field_byte_offset): Restore the
39451 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39452 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39453 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39456 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39459 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39461 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39464 * dwarf2out.c (add_data_member_location_attribute): For constant
39465 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39466 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39467 DW_AT_byte_size attributes.
39469 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39471 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39472 after forcing to constant memory when the code model is medium.
39474 2017-01-17 Julia Koval <julia.koval@intel.com>
39477 * config/i386/avx512fintrin.h
39478 (_mm512_i32gather_ps): Change __addr type to void const*.
39479 (_mm512_mask_i32gather_ps): Ditto.
39480 (_mm512_i32gather_pd): Ditto.
39481 (_mm512_mask_i32gather_pd): Ditto.
39482 (_mm512_i64gather_ps): Ditto.
39483 (_mm512_mask_i64gather_ps): Ditto.
39484 (_mm512_i64gather_pd): Ditto.
39485 (_mm512_mask_i64gather_pd): Ditto.
39486 (_mm512_i32gather_epi32): Ditto.
39487 (_mm512_mask_i32gather_epi32): Ditto.
39488 (_mm512_i32gather_epi64): Ditto.
39489 (_mm512_mask_i32gather_epi64): Ditto.
39490 (_mm512_i64gather_epi32): Ditto.
39491 (_mm512_mask_i64gather_epi32): Ditto.
39492 (_mm512_i64gather_epi64): Ditto.
39493 (_mm512_mask_i64gather_epi64): Ditto.
39494 (_mm512_i32scatter_ps): Change __addr type to void*.
39495 (_mm512_mask_i32scatter_ps): Ditto.
39496 (_mm512_i32scatter_pd): Ditto.
39497 (_mm512_mask_i32scatter_pd): Ditto.
39498 (_mm512_i64scatter_ps): Ditto.
39499 (_mm512_mask_i64scatter_ps): Ditto.
39500 (_mm512_i64scatter_pd): Ditto.
39501 (_mm512_mask_i64scatter_pd): Ditto.
39502 (_mm512_i32scatter_epi32): Ditto.
39503 (_mm512_mask_i32scatter_epi32): Ditto.
39504 (_mm512_i32scatter_epi64): Ditto.
39505 (_mm512_mask_i32scatter_epi64): Ditto.
39506 (_mm512_i64scatter_epi32): Ditto.
39507 (_mm512_mask_i64scatter_epi32): Ditto.
39508 (_mm512_i64scatter_epi64): Ditto.
39509 (_mm512_mask_i64scatter_epi64): Ditto.
39510 * config/i386/avx512pfintrin.h
39511 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39512 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39513 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39514 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39515 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39516 (_mm512_prefetch_i32scatter_ps): Ditto.
39517 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39518 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39519 (_mm512_prefetch_i64scatter_pd): Ditto.
39520 (_mm512_prefetch_i64scatter_ps): Ditto.
39521 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39522 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39523 * config/i386/avx512vlintrin.h
39524 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39525 (_mm_mmask_i32gather_ps): Ditto.
39526 (_mm256_mmask_i32gather_pd): Ditto.
39527 (_mm_mmask_i32gather_pd): Ditto.
39528 (_mm256_mmask_i64gather_ps): Ditto.
39529 (_mm_mmask_i64gather_ps): Ditto.
39530 (_mm256_mmask_i64gather_pd): Ditto.
39531 (_mm_mmask_i64gather_pd): Ditto.
39532 (_mm256_mmask_i32gather_epi32): Ditto.
39533 (_mm_mmask_i32gather_epi32): Ditto.
39534 (_mm256_mmask_i32gather_epi64): Ditto.
39535 (_mm_mmask_i32gather_epi64): Ditto.
39536 (_mm256_mmask_i64gather_epi32): Ditto.
39537 (_mm_mmask_i64gather_epi32): Ditto.
39538 (_mm256_mmask_i64gather_epi64): Ditto.
39539 (_mm_mmask_i64gather_epi64): Ditto.
39540 (_mm256_i32scatter_ps): Change __addr type to void*.
39541 (_mm256_mask_i32scatter_ps): Ditto.
39542 (_mm_i32scatter_ps): Ditto.
39543 (_mm_mask_i32scatter_ps): Ditto.
39544 (_mm256_i32scatter_pd): Ditto.
39545 (_mm256_mask_i32scatter_pd): Ditto.
39546 (_mm_i32scatter_pd): Ditto.
39547 (_mm_mask_i32scatter_pd): Ditto.
39548 (_mm256_i64scatter_ps): Ditto.
39549 (_mm256_mask_i64scatter_ps): Ditto.
39550 (_mm_i64scatter_ps): Ditto.
39551 (_mm_mask_i64scatter_ps): Ditto.
39552 (_mm256_i64scatter_pd): Ditto.
39553 (_mm256_mask_i64scatter_pd): Ditto.
39554 (_mm_i64scatter_pd): Ditto.
39555 (_mm_mask_i64scatter_pd): Ditto.
39556 (_mm256_i32scatter_epi32): Ditto.
39557 (_mm256_mask_i32scatter_epi32): Ditto.
39558 (_mm_i32scatter_epi32): Ditto.
39559 (_mm_mask_i32scatter_epi32): Ditto.
39560 (_mm256_i32scatter_epi64): Ditto.
39561 (_mm256_mask_i32scatter_epi64): Ditto.
39562 (_mm_i32scatter_epi64): Ditto.
39563 (_mm_mask_i32scatter_epi64): Ditto.
39564 (_mm256_i64scatter_epi32): Ditto.
39565 (_mm256_mask_i64scatter_epi32): Ditto.
39566 (_mm_i64scatter_epi32): Ditto.
39567 (_mm_mask_i64scatter_epi32): Ditto.
39568 (_mm256_i64scatter_epi64): Ditto.
39569 (_mm256_mask_i64scatter_epi64): Ditto.
39570 (_mm_i64scatter_epi64): Ditto.
39571 (_mm_mask_i64scatter_epi64): Ditto.
39572 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39573 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39574 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39575 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39576 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39577 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39578 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39579 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39580 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39581 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39582 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39583 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39584 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39585 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39586 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39587 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39588 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39589 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39590 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39591 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39592 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39593 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39594 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39595 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39596 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39597 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39598 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39599 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39600 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39601 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39602 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39603 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39604 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39605 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39606 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39607 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39608 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39609 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39610 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39611 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39612 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39613 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39614 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39615 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39616 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39617 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39618 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39619 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39620 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39621 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39622 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39623 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39624 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39625 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39626 definitions accordingly.
39628 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39629 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39632 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39635 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39638 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39639 hard regs for inner regno.
39641 2017-01-17 Martin Liska <mliska@suse.cz>
39644 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39645 assumption and add comment.
39647 2017-01-17 Nathan Sidwell <nathan@acm.org>
39649 * ipa-visibility.c (localize_node): New function, broken out of ...
39650 (function_and_variable_visibility): ... here. Call it.
39652 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39654 PR middle-end/77445
39655 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39656 correctly set frequency of oudgoing edge.
39657 (duplicate_thread_path): Fix profile updating.
39659 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39662 * configure.ac: Add GCC_BASE_VER.
39663 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39664 version from BASE-VER file.
39665 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39666 (gcc.o): Depend on $(BASEVER).
39667 * common.opt (dumpfullversion): New option.
39668 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39669 * doc/invoke.texi: Document -dumpfullversion.
39670 * doc/install.texi: Document --with-gcc-major-version-only.
39671 * configure: Regenerated.
39673 2017-01-17 Richard Biener <rguenther@suse.de>
39675 PR tree-optimization/71433
39676 * tree-vrp.c (register_new_assert_for): Merge same asserts
39677 on all incoming edges.
39678 (process_assert_insertions_for): Handle insertions at the
39681 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39683 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39684 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39686 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39689 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39692 2017-01-17 Alan Modra <amodra@gmail.com>
39695 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39696 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39697 symbolic stack limit when pic.
39699 2017-01-16 Martin Sebor <msebor@redhat.com>
39701 PR tree-optimization/78608
39702 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39704 2017-01-16 Jeff Law <law@redhat.com>
39707 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39708 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39709 for several include directories that may be relative to sysroot.
39710 * config/i386/x-mingw32 (gplus_includedir): Define.
39711 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39712 (native_system_includedir): Likewise.
39713 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39714 override if TARGET_SYSTEM_ROOT is defined.
39715 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39717 PR tree-optimization/79090
39718 PR tree-optimization/33562
39719 PR tree-optimization/61912
39720 PR tree-optimization/77485
39721 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39722 and computed trims into the dump file.
39724 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39726 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39728 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39731 * gimplify.c (gimplify_init_constructor): If want_value and
39732 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39736 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39737 sequence. Formatting fixes.
39738 (doloop_optimize): Formatting fixes.
39741 * gcc.c (debug_level_greater_than_spec_func): New function.
39742 (static_spec_functions): Add debug-level-gt spec function.
39743 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39745 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39746 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39747 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39748 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39749 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39750 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39752 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39754 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39755 QImode fixups to general and mask registers only.
39757 2017-01-16 Carl Love <cel@us.ibm.com>
39759 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39760 for built-in functions
39761 vector signed char vec_nabs (vector signed char)
39762 vector signed short vec_nabs (vector signed short)
39763 vector signed int vec_nabs (vector signed int)
39764 vector signed long long vec_nabs (vector signed long long)
39765 vector float vec_nabs (vector float)
39766 vector double vec_nabs (vector double)
39767 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39769 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39770 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39771 * doc/extend.texi: Update the documentation file for the new built-in
39774 2017-01-16 Martin Sebor <msebor@redhat.com>
39776 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39779 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39781 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39782 UNSPEC_VSX__XXSPLTD to require special splat handling.
39784 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39787 * system.h: Poison strndup.
39789 2017-01-16 Alan Modra <amodra@gmail.com>
39792 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39795 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39797 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39799 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39801 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39802 call recog here. Assert that INSN_CODE (insn) is non-negative.
39804 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39807 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39809 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39810 in the currently scheduled RTL fragment.
39812 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39814 PR rtl-optimization/78751
39815 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39818 2017-01-14 Jeff Law <law@redhat.com>
39820 PR tree-optimization/79090
39821 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39822 variable length stores.
39823 (compute_trims): Delete dead assignment to *trim_tail.
39824 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39827 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39829 PR rtl-optimization/78626
39830 PR rtl-optimization/78727
39831 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39832 of a block, and split such blocks after everything else is finished.
39834 2017-01-14 Alan Modra <amodra@gmail.com>
39837 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39838 target legitimate_combined_insn.
39839 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39840 (rs6000_legitimate_combined_insn): New function.
39841 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39843 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39844 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39845 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39847 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39849 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39851 2017-01-13 Jeff Law <law@redhat.com>
39853 PR tree-optimization/33562
39854 PR tree-optimization/61912
39855 PR tree-optimization/77485
39856 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39858 (delete_dead_assignment): Likewise.
39859 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
39860 statement to delete_dead_call and delete_dead_assignment.
39862 2017-01-13 David Malcolm <dmalcolm@redhat.com>
39865 * substring-locations.c (format_warning_va): Strengthen case 1 so
39866 that both endpoints of the substring must be within the format
39867 range for just the substring to be printed.
39869 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
39871 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
39872 * config/i386/i386.c (ix86_target_string): Add missing options
39873 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
39874 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
39875 flags_other and ix86_target_other to flags2_other. Display unknown
39877 (ix86_valid_target_attribute_inner_p): Add missing options and
39878 reorder options by implied ISAs, as in ix86_target_string.
39880 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39882 * hash-table.h (hash_table::too_empty_p): New function.
39883 (hash_table::expand): Use it.
39884 (hash_table::traverse): Likewise.
39885 (hash_table::empty_slot): Use sizeof (value_type) instead of
39886 sizeof (PTR) to convert bytes to elements. Shrink the table
39887 if the current size is excessive for the current number of
39890 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39892 * ira-costs.c (record_reg_classes): Break from the inner loop
39893 early once alt_fail is known to be true. Update outer loop
39894 handling accordingly.
39896 2017-01-13 Jeff Law <law@redhat.com>
39898 * tree-ssa-dse.c (decrement_count): New function.
39899 (increment_start_addr, maybe_trim_memstar_call): Likewise.
39900 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
39901 when we know the partially dead statement is a mem* function.
39903 PR tree-optimization/61912
39904 PR tree-optimization/77485
39905 * tree-ssa-dse.c: Include expr.h.
39906 (maybe_trim_constructor_store): New function.
39907 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
39909 PR tree-optimization/33562
39910 PR tree-optimization/61912
39911 PR tree-optimization/77485
39912 * doc/invoke.texi: Document new dse-max-object-size param.
39913 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
39914 * tree-ssa-dse.c: Include params.h.
39915 (dse_store_status): New enum.
39916 (initialize_ao_ref_for_dse): New, partially extracted from
39918 (valid_ao_ref_for_dse, normalize_ref): New.
39919 (setup_live_bytes_from_ref, compute_trims): Likewise.
39920 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
39921 (maybe_trim_partially_dead_store): Likewise.
39922 (maybe_trim_complex_store): Likewise.
39923 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
39924 Track what bytes live from the original store. Return tri-state
39925 for dead, partially dead or live.
39926 (dse_dom_walker): Add constructor, destructor and new private members.
39927 (delete_dead_call, delete_dead_assignment): New extracted from
39929 (dse_optimize_stmt): Make a member of dse_dom_walker.
39930 Use initialize_ao_ref_for_dse.
39932 PR tree-optimization/33562
39933 PR tree-optimization/61912
39934 PR tree-optimization/77485
39935 * sbitmap.h (bitmap_count_bits): Prototype.
39936 (bitmap_clear_range, bitmap_set_range): Likewise.
39937 * sbitmap.c (bitmap_clear_range): New function.
39938 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
39940 2017-01-13 Martin Liska <mliska@suse.cz>
39943 * function.c (set_cfun): Add new argument force.
39944 * function.h (set_cfun): Likewise.
39945 * ipa-inline-transform.c (inline_call): Use the function when
39946 strict alising from is dropped for function we inline to.
39948 2017-01-13 Richard Biener <rguenther@suse.de>
39950 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
39951 for dumping GIMPLE INTEGER_CSTs.
39953 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
39955 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
39956 to 201112L since C++17.
39958 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
39961 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
39962 if -fsanitize=kernel-address is present.
39964 2017-01-13 Richard Biener <rguenther@suse.de>
39966 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
39967 as _Literal ( type ) number in case usual suffixes do not
39968 preserve all information.
39970 2017-01-13 Richard Biener <rguenther@suse.de>
39972 PR tree-optimization/77283
39973 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
39974 and ssa-iterators.h.
39975 (is_feasible_trace): Implement a cost model based on joiner
39978 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
39981 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
39982 char or short to __float128/_Float128 directly.
39984 2017-01-12 Martin Sebor <msebor@redhat.com>
39986 to -Wformat-overflow.
39987 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
39988 (min_bytes_remaining): Same.
39989 (get_string_length): Same.
39990 (format_string): Same.
39991 (format_directive): Same.
39993 (pass_sprintf_length::handle_gimple_call): Same.
39995 2017-01-12 Jakub Jelinek <jakub@redhat.com>
39997 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
39998 info.nowrite calls with no lhs that can't throw. Return bool
39999 whether gsi_remove has been called or not.
40000 (pass_sprintf_length::handle_gimple_call): Return bool whether
40001 try_substitute_return_value called gsi_remove. Formatting fix.
40002 (pass_sprintf_length::execute): Don't use gsi_remove if
40003 handle_gimple_call returned true.
40006 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40007 be removed due to side-effects, don't remove following barrier nor
40008 turn the successor edge into fallthru edge.
40010 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40013 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40014 element-reversing loads and stores as not swappable.
40016 2017-01-12 Nathan Sidwell <nathan@acm.org>
40017 Nicolai Stange <nicstange@gmail.com>
40019 * combine.c (try_combine): Don't ignore result of overlap checking
40020 loop. Combine overlap & asm check into single loop.
40022 2017-01-12 Richard Biener <rguenther@suse.de>
40024 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40025 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40027 2017-01-12 Richard Biener <rguenther@suse.de>
40029 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40030 and TS_TARGET_OPTION directly derive from TS_BASE.
40031 * tree-core.h (tree_optimization_option): Derive from tree_base.
40032 (tree_target_option): Likewise.
40034 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40036 * config/i386/i386.c (memory_address_length): Increase len
40037 only when rip_relative_addr_p returns false.
40039 2017-01-11 Julia Koval <julia.koval@intel.com>
40041 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40042 (OPTION_MASK_ISA_SGX_SET): New.
40043 (ix86_handle_option): Handle OPT_msgx.
40044 * config.gcc: Added sgxintrin.h.
40045 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40046 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40047 * config/i386/i386.c (ix86_target_string): Add -msgx.
40049 (ix86_option_override_internal): Handle new options.
40050 (ix86_valid_target_attribute_inner_p): Add sgx.
40051 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40052 * config/i386/i386.opt: Add msgx.
40053 * config/i386/sgxintrin.h: New file.
40054 * config/i386/x86intrin.h: Add sgxintrin.h.
40056 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40059 * fold-const.c (maybe_nonzero_address): Return 1 for function
40061 (tree_single_nonzero_warnv_p): Don't handle function local objects
40065 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40068 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40071 * opts.c: Include "spellcheck.h"
40072 (struct string_fragment): New struct.
40073 (struct edit_distance_traits<const string_fragment &>): New
40075 (get_closest_sanitizer_option): New function.
40076 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40078 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40080 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40082 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40083 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40084 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40085 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40086 for initial die_offset if dwarf_split_debug_info.
40087 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40088 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40090 (output_skeleton_debug_sections): Formatting fix. Use
40091 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40092 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40094 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40096 * config/arm/cortex-a53.md: Add bypasses for
40097 cortex_a53_r2f_cvt.
40098 (cortex_a53_r2f): Only use for transfers.
40099 (cortex_a53_f2r): Likewise.
40100 (cortex_a53_r2f_cvt): Add reservation for conversions.
40101 (cortex_a53_f2r_cvt): Likewise.
40103 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40105 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40106 to all inlined functions, change static to extern.
40108 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40111 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40113 (arm_assemble_integer): Likewise.
40115 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40117 * config.gcc: Use new awk script to check CPU, FPU and architecture
40118 parameters for --with-... options.
40119 * config/arm/parsecpu.awk: New file
40120 * config/arm/arm-cpus.in: New file.
40121 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40123 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40125 * config/arm/t-arm: Update dependency rules.
40126 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40127 of processing .def files.
40128 * config/arm/genopt.sh: Deleted.
40129 * config/arm/gentune.sh: Deleted.
40130 * config/arm/arm-cores.def: Deleted.
40131 * config/arm/arm-arches.def: Deleted.
40132 * config/arm/arm-fpus.def: Deleted.
40133 * config/arm/arm-tune.md: Regenerated.
40134 * config/arm/arm-tables.opt: Regenerated.
40135 * config/arm/arm-cpu.h: New generated file.
40136 * config/arm/arm-cpu-data.h: New generated file.
40137 * config/arm/arm-cpu-cdata.h: New generated file.
40139 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40142 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40144 (input_varpool_node): Unpack dynamically_initialized bit.
40146 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40148 PR rtl-optimization/79032
40149 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40150 the alignment of the adjusted memory reference against that of MODE,
40151 instead of the alignment of the original memory reference.
40153 2017-01-11 Martin Jambor <mjambor@suse.cz>
40155 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40157 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40158 decorated functions.
40160 2017-01-11 Richard Biener <rguenther@suse.de>
40162 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40163 set range/nonnull info for PHI results. Do not set it on
40164 stmts marked for removal.
40166 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40168 * expr.c (store_field): In the bitfield case, fetch the return value
40169 from the registers before applying a single big-endian adjustment.
40170 Always do a final load for a BLKmode value not larger than a word.
40172 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40175 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40176 that we correctly handle column numbers greater than
40177 LINE_MAP_MAX_COLUMN_NUMBER.
40179 2017-01-10 Martin Sebor <msebor@redhat.com>
40181 PR middle-end/78245
40182 * gimple-ssa-sprintf.c (get_destination_size): Call
40183 {init,fini}object_sizes.
40184 * tree-object-size.c (addr_object_size): Adjust.
40185 (pass_through_call): Adjust.
40186 (pass_object_sizes::execute): Adjust.
40187 * tree-object-size.h (fini_object_sizes): Declare.
40189 2017-01-10 Martin Sebor <msebor@redhat.com>
40191 PR tree-optimization/78775
40192 * builtins.c (get_size_range): Move...
40193 * calls.c: ...to here.
40194 (alloc_max_size): Accept zero argument.
40195 (operand_signed_p): Remove.
40196 (maybe_warn_alloc_args_overflow): Call get_size_range.
40197 * calls.h (get_size_range): Declare.
40199 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40201 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40202 from TI's devices.csv file as of September 2016.
40203 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40205 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40207 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40208 * doc/invoke.texi: Likewise.
40209 * doc/md.texi: Likewise.
40210 * doc/objc.texi: Likewise.
40212 2017-01-10 Joshua Conner <joshconner@google.com>
40214 * config/arm/fuchsia-elf.h: New file.
40215 * config/fuchsia.h: New file.
40216 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40217 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40219 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40221 2016-01-10 Richard Biener <rguenther@suse.de>
40223 PR tree-optimization/79034
40224 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40225 Propagate out degenerate PHIs in the joiner.
40227 2017-01-10 Martin Liska <mliska@suse.cz>
40229 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40230 (sort_congruence_classes_by_decl_uid): Likewise.
40231 (sort_congruence_class_groups_by_decl_uid): Likewise.
40232 (sem_item_optimizer::merge_classes): Sort class, groups in these
40233 classes and members in the groups by DECL_UID of declarations.
40234 This would make merge operations stable.
40236 2017-01-10 Martin Liska <mliska@suse.cz>
40238 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40239 usage of m_classes_vec.
40240 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40241 (sem_item_optimizer::get_group_by_hash): Likewise.
40242 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40243 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40244 (sem_item_optimizer::verify_classes): Likewise.
40245 (sem_item_optimizer::process_cong_reduction): Likewise.
40246 (sem_item_optimizer::dump_cong_classes): Likewise.
40247 (sem_item_optimizer::merge_classes): Likewise.
40248 * ipa-icf.h (congruence_class_hash): Rename from
40249 congruence_class_group_hash. Remove declaration of m_classes_vec.
40251 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40253 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40254 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40255 * config.gcc: Add avx512vpopcntdqintrin.h.
40256 * config/i386/avx512vpopcntdqintrin.h: New.
40257 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40258 * config/i386/i386-builtin-types.def: Add new types.
40259 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40260 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40261 __builtin_ia32_vpopcountq_v8di_mask): New.
40262 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40263 __AVX512VPOPCNTDQ__.
40264 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40265 (PTA_AVX512VPOPCNTDQ): Define.
40266 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40267 TARGET_AVX512VPOPCNTDQ_P): Define.
40268 * config/i386/i386.opt: Add mavx512vpopcntdq.
40269 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40270 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40272 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40274 PR middle-end/77484
40275 * predict.def (PRED_CALL): Set to 67.
40277 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40279 * expr.c (store_field): In the bitfield case, if the value comes from
40280 a function call and is of an aggregate type returned in registers, do
40281 not modify the field mode; extract the value in all cases if the mode
40282 is BLKmode and the size is not larger than a word.
40284 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40287 * config/i386/cpuid.h: Fix undefined behavior.
40289 2017-01-04 Jeff Law <law@redhat.com>
40291 PR tree-optimization/79007
40292 PR tree-optimization/67955
40293 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40294 conservative for pt.null when flag_non_call_exceptions is on.
40296 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40298 PR translation/79019
40299 PR translation/79020
40300 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40301 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40302 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40304 * config/avr/avr.opt (maccumulate-args): Likewise.
40305 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40306 * common.opt (freport-bug): Likewise.
40307 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40308 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40309 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40310 translatable string.
40311 * config/i386/i386.c (function_value_32): Likewise.
40312 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40313 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40315 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40316 * common/config/msp430/msp430-common.c (msp430_handle_option):
40318 * symtab.c (symtab_node::verify_base): Likewise.
40319 * opts.c (set_debug_level): Likewise.
40320 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40321 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40322 missing whitespace to translatable strings.
40323 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40324 * config/sh/superh.h: Likewise.
40325 * config/i386/xopintrin.h: Likewise.
40326 * config/i386/znver1.md: Likewise.
40327 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40328 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40329 * double-int.h (struct double_int): Likewise.
40330 * double-int.c (div_and_round_double): Likewise.
40331 * wide-int.cc: Likewise.
40332 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40333 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40334 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40335 (crossjumps_occurred): ... this.
40336 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40339 PR tree-optimization/78899
40340 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40341 returning bool return struct loop *, NULL for failure and the new
40343 (versionable_outer_loop_p): Don't version outer loop if it has
40344 dont_vectorized bit set.
40345 (tree_if_conversion): When versioning outer loop, ensure
40346 tree_if_conversion is performed also on the inner loop of the
40347 non-vectorizable outer loop copy.
40348 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40349 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40350 prevent vectorization of it.
40351 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40352 the outer loop vectorization of the non-scalar version is attempted
40353 before vectorization of the inner loop in scalar version. If
40354 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40355 vectorization of its inner loop.
40356 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40357 has 2 inner loops, rename also on edges from bb whose single pred
40358 is outer_loop->header. Fix typo in function comment.
40360 2017-01-09 Martin Sebor <msebor@redhat.com>
40363 * asan.c (asan_emit_stack_protection): Increase local buffer size
40364 to avoid snprintf truncation warning.
40366 2017-01-09 Andrew Pinski <apinski@cavium.com>
40368 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40369 to reference thunderx2t99 for the tuning structure
40370 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40372 (thunderx2t99_extra_costs): This.
40373 * config/aarch64/aarch64-tune.md: Regenerate.
40374 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40375 (vulcan_addrcost_table): This.
40376 (vulcan_regmove_cost): Rename to ...
40377 (thunderx2t99_regmove_cost): This.
40378 (vulcan_vector_cost): Rename to ...
40379 (thunderx2t99_vector_cost): this.
40380 (vulcan_branch_cost): Rename to ...
40381 (thunderx2t99_branch_cost): This.
40382 (vulcan_tunings): Rename to ...
40383 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40384 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40386 2017-01-09 Martin Jambor <mjambor@suse.cz>
40390 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40391 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40392 (propagate_vr_accross_jump_function): Use the above function for all
40393 value range computations for pass-through jump functions and type
40394 converasion from explicit value range values.
40395 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40396 parameters from TYPE_ARG_TYPES.
40397 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40398 (ipa_write_node_info): Stream type of the actual argument.
40399 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40401 2017-01-09 Martin Liska <mliska@suse.cz>
40404 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40405 (lookup_compiler): Do not show error message with have_E.
40407 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40409 PR tree-optimization/78938
40410 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40411 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40412 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40413 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40416 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40418 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40421 2017-01-09 Richard Biener <rguenther@suse.de>
40423 PR tree-optimization/78997
40424 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40425 name condition properly.
40427 2017-01-09 Richard Biener <rguenther@suse.de>
40430 * dwarf2out.c (is_cxx): New overload with context.
40431 (is_naming_typedef_decl): Use it.
40433 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40435 * invoke.texi (Option Summary): Correct spacing in option lists
40436 and add line breaks to fix over-long lines.
40438 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40440 PR middle-end/17660
40442 * extend.texi (Common Variable Attributes): Add xref to GCC
40443 Internals manual to explain mode attribute keywords.
40445 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40448 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40449 and Preprocessor Options.
40450 (Options for Linking): Document -pthread here....
40451 (RS/6000 and PowerPC Options): ...not here.
40452 (Solaris 2 Options): ...or here.
40453 * doc/cppopts.texi: Document -pthread.
40455 2017-01-08 Martin Sebor <msebor@redhat.com>
40457 PR middle-end/77708
40458 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40459 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40460 New member functions.
40461 (format_directive): Used them.
40463 (pass_sprintf_length::handle_gimple_call): Same.
40464 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40465 to avoid truncation for any argument.
40466 (extract_affine_mul): Same.
40467 * tree.c (get_file_function_name): Same.
40469 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40471 PR middle-end/77484
40472 * predict.def (PRED_INDIR_CALL): Set to 86.
40474 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40476 PR preprocessor/54124
40477 * doc/cppopts.texi: Reformat -d subtable to list the full name
40478 of the options. Add cross-reference to the docs for the general
40479 compiler -d options.
40480 * doc/invoke.texi (Developer Options): Add cross-reference to the
40481 preprocessor-specific -d option documentation.
40483 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40485 PR preprocessor/13498
40486 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40487 redudant material, and reflect new command-line options.
40488 (System Headers): Likewise.
40490 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40492 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40493 -isystem, and -idirafter. Copy-edit.
40494 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40495 default for -ftrack-macro-expansion. Delete obsolete and
40496 badly-formatted implementation details about -fdebug-cpp output.
40497 * doc/cppwarnopts.texi: Copy-edit.
40499 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40502 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40503 that the transition from a max line width >= 1<<10 to narrower
40504 lines works correctly.
40506 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40508 * doc/options.texi (PerFunction): New.
40509 * opt-functions.awk (switch_flags): Map both Optimization and
40510 PerFunction to CL_OPTIMIZATION.
40511 * opth-gen.awk: Test for PerFunction flag along with
40513 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40514 it only when the latter is present. Skip those that don't in
40515 the hash function generator.
40516 * common.opt (fvar-tracking): Mark as PerFunction instead of
40518 (fvar-tracking-assignments): Likewise.
40519 (fvar-tracking-assignments-toggle): Likewise.
40520 (fvar-tracking-uninit): Likewise.
40522 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40524 PR translation/79018
40525 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40528 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40531 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40532 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40533 TARGET_LONG_JUMP_TABLE_OFFSETS.
40534 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40535 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40536 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40537 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40538 * config/m68k/m68k.md (tablejump expander): Likewise.
40539 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40540 TARGET_LONG_JUMP_TABLE_OFFSETS.
40541 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40542 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40544 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40545 David Holsgrove <david.holsgrove@xilinx.com>
40547 * common/config/microblaze/microblaze-common.c
40548 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40549 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40551 * config/microblaze/microblaze.c (microblaze_must_save_register)
40552 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40554 (microblaze_eh_return): New function.
40555 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40556 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40557 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40558 * config/microblaze/microblaze.md (eh_return): New pattern.
40560 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40562 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40563 GCC_DIAGNOSTIC_STRINGIFY): Define.
40565 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40567 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40569 * config/arm/arm.md (<mcrr>): New.
40571 * config/arm/arm.c (arm_arch5te): New.
40572 (arm_option_override): Set arm_arch5te.
40573 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40575 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40576 (arm_mcrr_qualifiers): ... this. New.
40577 (MRRC_QUALIFIERS): Define to...
40578 (arm_mrrc_qualifiers): ... this. New.
40579 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40581 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40582 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40583 (MRRCI, mrrc, MRRC): New.
40584 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40585 VUNSPEC_MRRC2): New.
40587 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40589 * config/arm/arm.md (<mcr>): New.
40591 * config/arm/arm.c (arm_coproc_builtin_available): Add
40592 support for mcr, mrc, mcr2 and mrc2.
40593 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40594 (arm_mcr_qualifiers): ... this. New.
40595 (MRC_QUALIFIERS): Define to ...
40596 (arm_mrc_qualifiers): ... this. New.
40597 (MCR_QUALIFIERS): Define to ...
40598 (arm_mcr_qualifiers): ... this. New.
40599 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40601 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40602 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40603 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40604 VUNSPEC_MRC2): New.
40606 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40608 * config/arm/arm.md (*ldc): New.
40612 * config/arm/arm.c (arm_coproc_builtin_available): Add
40613 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40614 (arm_coproc_ldc_stc_legitimate_address): New.
40615 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40616 'qualifier_const_pointer'.
40617 (LDC_QUALIFIERS): Define to...
40618 (arm_ldc_qualifiers): ... this. New.
40619 (STC_QUALIFIERS): Define to...
40620 (arm_stc_qualifiers): ... this. New.
40621 * config/arm/arm-protos.h
40622 (arm_coproc_ldc_stc_legitimate_address): New.
40623 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40624 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40625 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40626 stc2, stcl, stc2l): New.
40627 * config/arm/constraints.md (Uz): New.
40628 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40629 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40630 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40631 VUNSPEC_STC2L): New.
40633 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40635 * config/arm/arm.md (<cdp>): New.
40636 * config/arm/arm.c (neon_const_bounds): Rename this ...
40637 (arm_const_bounds): ... this.
40638 (arm_coproc_builtin_available): New.
40639 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40640 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40641 (CDP_QUALIFIERS): Define to...
40642 (arm_cdp_qualifiers): ... this. New.
40644 (arm_expand_builtin_args): Add case for 6 arguments.
40645 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40646 (arm_const_bounds): ... this.
40647 (arm_coproc_builtin_available): New.
40648 * config/arm/arm_acle.h (__arm_cdp): New.
40650 * config/arm/arm_acle_builtins.def (cdp): New.
40652 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40653 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40654 'arm_const_bounds'.
40655 * config/arm/types.md (coproc): New.
40656 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40657 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40658 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40659 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40661 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40663 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40664 (UBINOP_QUALIFIERS): New.
40666 (acle_builtin_data): New. Change comment.
40667 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40668 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40669 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40670 arm_acle_builtins.def.
40671 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40672 (arm_init_acle_builtins): New.
40673 (CRC32_BUILTIN): Remove.
40674 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40675 crc32cb, crc32ch and crc32cw.
40676 (arm_init_crc32_builtins): Remove.
40677 (arm_init_builtins): Use arm_init_acle_builtins rather
40678 than arm_init_crc32_builtins.
40679 (arm_expand_acle_builtin): New.
40680 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40681 * config/arm/arm_acle_builtins.def: New.
40683 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40685 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40686 (arm_builtin_datum): ... this.
40687 (arm_init_neon_builtin): Rename to ...
40688 (arm_init_builtin): ... this. Add a new parameters PREFIX
40689 and USE_SIG_IN_NAME.
40690 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40691 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40692 'arm_builtin_datum'.
40693 (arm_init_vfp_builtins): Likewise.
40694 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40695 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40696 (arm_expand_neon_args): Rename to ...
40697 (arm_expand_builtin_args): ... this. Rename builtin_arg
40698 enum values and differentiate between ARG_BUILTIN_MEMORY
40699 and ARG_BUILTIN_NEON_MEMORY.
40700 (arm_expand_neon_builtin_1): Rename to ...
40701 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40702 values, arm_expand_builtin_args and add bool parameter NEON.
40703 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40704 (arm_expand_vfp_builtin): Likewise.
40705 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40707 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40709 PR middle-end/77484
40710 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40711 * predict.c (tree_estimate_probability_bb): Reverse direction of
40712 polymorphic call predictor.
40714 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40716 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40717 (determine_pass_name_match): ...this new function and...
40718 (should_skip_pass_p): ...this new function.
40720 2017-01-06 Nathan Sidwell <nathan@acm.org>
40722 * ipa-visibility.c (function_and_variable_visibility): Reformat
40723 comments and long lines. Remove extrneous if.
40724 * symtab.c (symtab_node::make_decl_local): Fix code format.
40725 (symtab_node::set_section_for_node): Fix comment typo.
40727 2017-01-06 Martin Liska <mliska@suse.cz>
40730 * lra-constraints.c: Rename invariant to lra_invariant.
40731 * predict.c (set_even_probabilities): Initialize e to NULL.
40733 2017-01-05 Martin Sebor <msebor@redhat.com>
40735 PR tree-optimization/78910
40736 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40737 (format_integer): Correct off-by-one error in the handling
40738 of precision with negative numbers in signed conversions..
40740 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40742 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40744 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40746 PR tree-optimization/71016
40747 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40748 factor_out_conditional_conversion. Formatting fix.
40749 (factor_out_conditional_conversion): Add cond_stmt argument.
40750 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40751 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40754 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40756 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40757 read-rtl-function.o, and selftest-rtl.o.
40758 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40759 (selftest::aarch64_test_loading_full_dump): New function.
40760 (selftest::aarch64_run_selftests): New function.
40761 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40762 selftest::aarch64_run_selftests.
40763 * config/i386/i386.c
40764 (selftest::ix86_test_loading_dump_fragment_1): New function.
40765 (selftest::ix86_test_loading_call_insn): New function.
40766 (selftest::ix86_test_loading_full_dump): New function.
40767 (selftest::ix86_test_loading_unspec): New function.
40768 (selftest::ix86_run_selftests): Call the new functions.
40769 * emit-rtl.c (maybe_set_max_label_num): New function.
40770 * emit-rtl.h (maybe_set_max_label_num): New decl.
40771 * function.c (instantiate_decls): Guard call to
40772 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40773 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40775 * gensupport.c (gen_reader::gen_reader): Pass "false"
40776 for new "compact" param of rtx_reader.
40777 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40778 rather than an empty string for NULL strings.
40779 * read-md.c: Potentially include config.h rather than bconfig.h.
40780 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40781 (have_error): New global, copied from errors.c.
40782 (md_reader::read_name): Rename to...
40783 (md_reader::read_name_1): ...this, adding "out_loc" param,
40784 and converting "missing name or number" to returning false, rather
40786 (md_reader::read_name): Reimplement in terms of read_name_1.
40787 (md_reader::read_name_or_nil): New function.
40788 (md_reader::read_string): Handle "(nil)" by returning NULL.
40789 (md_reader::md_reader): Add new param "compact".
40790 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40791 (md_reader::read_file): New method.
40792 * read-md.h (md_reader::md_reader): Add new param "compact".
40793 (md_reader::read_file): New method.
40794 (md_reader::is_compact): New accessor.
40795 (md_reader::read_name): Convert return type from void to file_location.
40796 (md_reader::read_name_or_nil): New decl.
40797 (md_reader::read_name_1): New decl.
40798 (md_reader::m_compact): New field.
40799 (noop_reader::noop_reader): Pass "false" for new "compact" param
40801 (rtx_reader::rtx_reader): Add new "compact" param.
40802 (rtx_reader::read_rtx_operand): Make virtual and convert return
40803 type from void to rtx.
40804 (rtx_reader::read_until): New decl.
40805 (rtx_reader::handle_any_trailing_information): New virtual function.
40806 (rtx_reader::postprocess): New virtual function.
40807 (rtx_reader::finalize_string): New virtual function.
40808 (rtx_reader::m_in_call_function_usage): New field.
40809 (rtx_reader::m_reuse_rtx_by_id): New field.
40810 * read-rtl-function.c: New file.
40811 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40812 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40813 (selftest::verify_three_block_rtl_cfg): New decl.
40814 * read-rtl-function.h: New file.
40815 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40816 For host, include function.h, memmodel.h, and emit-rtl.h.
40817 (one_time_initialization): New function.
40818 (struct compact_insn_name): New struct.
40819 (compact_insn_names): New array.
40820 (find_code): Handle insn codes in compact dumps.
40821 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40822 (bind_subst_iter_and_attr): Likewise.
40823 (add_condition_to_string): Likewise.
40824 (add_condition_to_rtx): Likewise.
40825 (apply_attribute_uses): Likewise.
40826 (add_current_iterators): Likewise.
40827 (apply_iterators): Likewise.
40828 (initialize_iterators): Guard usage of apply_subst_iterator with
40829 #ifdef GENERATOR_FILE.
40830 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40831 (md_reader::read_mapping): Likewise.
40832 (add_define_attr_for_define_subst): Likewise.
40833 (add_define_subst_attr): Likewise.
40834 (read_subst_mapping): Likewise.
40835 (check_code_iterator): Likewise.
40836 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40838 (one_time_initialization): New function.
40839 (rtx_reader::read_until): New method.
40840 (read_flags): New function.
40841 (parse_reg_note_name): New function.
40842 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40843 Handle reuse_rtx ids.
40844 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40845 Add parsing support for RTL dumps, mirroring the special-cases in
40846 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40847 values, and calling handle_any_trailing_information.
40848 (rtx_reader::read_rtx_operand): Convert return type from void
40849 to rtx, returning return_rtx. Handle case 'e'. Call
40850 finalize_string on XSTR and XTMPL fields.
40851 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40852 "(nil)" values were omitted. Call the postprocess vfunc on the
40854 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40855 class ctor. Initialize m_in_call_function_usage. Call
40856 one_time_initialization.
40857 * rtl-tests.c (selftest::test_uncond_jump): Call
40858 set_new_first_and_last_insn.
40859 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
40860 * selftest-rtl.c: New file.
40861 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
40862 (selftest::get_insn_by_uid): New decl.
40863 * selftest-run-tests.c (selftest::run_tests): Call
40864 read_rtl_function_c_tests.
40865 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
40866 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
40869 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
40871 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
40872 operands in a special way. Assert that pos+len <= mode precision.
40874 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40876 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
40877 3 argument Alias with unlimited for the negative form.
40878 (fno-vect-cost-model): Removed.
40880 2017-01-05 Martin Liska <mliska@suse.cz>
40882 * hsa-gen.c (gen_hsa_divmod): New function.
40883 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
40885 2017-01-05 Martin Liska <mliska@suse.cz>
40888 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
40891 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40893 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
40894 small constant length operands.
40896 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40898 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
40899 between loop iterations.
40901 2017-01-05 Martin Liska <mliska@suse.cz>
40904 * gimplify.c (gimplify_decl_expr): Compare to
40905 asan_poisoned_variables instread of checking flags.
40906 (gimplify_target_expr): Likewise.
40907 (gimplify_expr): Likewise.
40908 (gimplify_function_tree): Conditionally initialize
40909 asan_poisoned_variables.
40911 2017-01-04 Jeff Law <law@redhat.com>
40913 PR tree-optimizatin/78812
40914 * rtl.h (contains_mem_rtx_p): Prototype.
40915 * ifcvt.c (containts_mem_rtx_p): Move from here to...
40916 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
40917 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
40918 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
40919 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
40921 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40923 * input.c (assert_char_at_range): Default-initialize actual_range.
40925 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40927 * df-scan.c (df_ref_create_structure): Make regno unsigned,
40928 to match the caller.
40930 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40932 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
40933 insns after final jump in test to emit dummy move.
40935 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40937 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
40938 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
40940 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40942 * multiple_target.c (create_dispatcher_calls): Init e_next.
40943 * tree-ssa-loop-split.c (split_loop): Init border.
40944 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
40947 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
40952 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
40953 (altivec_register_operand): Do not return true if the operand
40954 contains a SUBREG mixing SImode and SFmode.
40955 (vsx_register_operand): Likewise.
40956 (vsx_reg_sfsubreg_ok): New predicate.
40957 (vfloat_operand): Do not return true if the operand contains a
40958 SUBREG mixing SImode and SFmode.
40959 (vint_operand): Likewise.
40960 (vlogical_operand): Likewise.
40961 (gpc_reg_operand): Likewise.
40962 (int_reg_operand): Likewise.
40963 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
40964 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
40965 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
40967 (rs6000_emit_move_si_sf_subreg): New helper function.
40968 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
40969 fixup SUBREGs involving SImode and SFmode.
40970 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
40971 numbers for the new peephole2 optimization.
40972 (peephole2 for SFmode unions): New peephole2 to optimize cases in
40973 the GLIBC math library that do AND/IOR/XOR operations on single
40974 precision floating point.
40975 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
40976 target macros to say whether we need to avoid SUBREGs mixing
40978 (TARGET_ALLOW_SF_SUBREG): Likewise.
40979 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
40980 (UNSPEC_SI_FROM_SF): Likewise.
40981 (iorxor): Change spacing.
40982 (and_ior_xor): New iterator for AND, IOR, and XOR.
40983 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
40984 (movdi_from_sf_zero_ext): Likewise.
40985 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
40986 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
40987 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
40988 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
40989 (fms<mode>4): Likewise.
40990 (fnma<mode>4): Likewise.
40991 (fnms<mode>4): Likewise.
40992 (nfma<mode>4): Likewise.
40993 (nfms<mode>4): Likewise.
40995 2017-01-04 Marek Polacek <polacek@redhat.com>
40998 * doc/invoke.texi: Document -Wpointer-compare.
41000 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41002 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41005 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41006 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41009 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41012 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41013 documentation of the powerpc_popcntb_ok attribute.
41014 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41015 code to issue warning messages if a requested CPU configuration is
41016 not supported by the binary (assembler and loader) toolchain.
41017 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41018 made to define a built-in function that has been disabled.
41019 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41020 made to define a built-in function that has been disabled.
41021 (altivec_init_builtins): Add comment explaining why definition
41022 of the DST built-in functions is not preceded by an assertion
41023 check. Add assertions to prevent ICE if attempts are made to
41024 define an altivec predicate or an abs* built-in function that has
41026 (htm_init_builtins): Add comment explaining why definition of the
41027 htm built-in functions is not preceded by an assertion check.
41029 2017-01-04 Jeff Law <law@redhat.com>
41031 PR tree-optimizatin/67955
41032 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41033 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41034 the points-to solution does not include pt_null. Use DECL_PT_UID
41037 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41039 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41040 Use gen_int_mode instead of gen_lopwart for const_int operands.
41042 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41044 PR tree-optimization/71563
41045 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41046 out of range value - has low bits known to be zero.
41048 2017-01-04 Alan Modra <amodra@gmail.com>
41050 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41051 * configure: Regenerate.
41052 * config.in: Regenerate.
41054 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41057 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41058 a substring of the message, but strcmp with the whole message. Ifdef
41059 ENABLE_NLS, translate the message first using dgettext.
41061 2017-01-03 Jeff Law <law@redhat.com>
41063 PR tree-optimizatin/78856
41064 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41065 (mark_threaded_blocks): Remove code to truncate thread paths that
41066 cross multiple loop headers. Instead invalidate the cached loop
41067 iteration information and handle case of a thread path walking
41068 into an irreducible region.
41070 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41073 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41074 assertions. Add support for doing the signbit if the IEEE 128-bit
41075 floating point value is in a GPR.
41076 * config/rs6000/rs6000.md (Fsignbit): Delete.
41077 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41078 Update the length attribute if the value is in a GPR.
41079 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41080 the sign or zero extension instruction, since the value is always 0/1.
41081 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41084 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41085 extracting SImode to a GPR register so that we can generate a
41086 store, limit the vector to be in a traditional Altivec register
41087 for the vextuwrx instruction.
41089 2017-01-03 Ian Lance Taylor <iant@google.com>
41091 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41093 2017-01-03 Martin Sebor <msebor@redhat.com>
41095 PR tree-optimization/78696
41096 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41097 precision. Use MPFR for %f for greater fidelity. Correct handling
41099 (pass_sprintf_length::compute_format_length): Set width and precision
41100 specified by asrerisk to void_node for vararg functions.
41101 (try_substitute_return_value): Adjust dump output.
41103 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41105 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41107 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41109 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41110 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41111 -mlra/-mno-lra was passed to the compiler.
41113 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41115 PR rtl-optimization/65618
41116 * emit-rtl.c (try_split): Move initialization of "before" and
41117 "after" to just before the call to emit_insn_after_setloc.
41119 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41121 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41123 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41125 * dwarf2out.c (gen_enumeration_type_die): When
41126 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41128 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41130 PR tree-optimization/78965
41131 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41132 Change first argument from const call_info & to call_info &. For %n
41133 set info.nowrite to false.
41135 PR middle-end/78901
41136 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41137 possibly throwing calls.
41139 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41140 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41141 and fns handling, rather than in a separate case SSA_NAME.
41143 2017-01-02 Jeff Law <law@redhat.com>
41145 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41146 fixes for first_period and second_period variables.
41148 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41151 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41152 (*insvqi_1): New insn pattern.
41153 (*insvqi_1_mem_rex64): Ditto.
41154 (*insvqi_2): Ditto.
41155 (*insvqi_3): Rename from *insvqi.
41157 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41159 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41161 * doc/cfg.texi (Edges): Remove reference to Java.
41162 (Maintaining the CFG): Ditto.
41164 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41166 PR middle-end/77674
41167 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41168 transparent aliases.
41170 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41172 PR middle-end/77484
41173 * predict.def (PRED_CALL): Update hitrate.
41174 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41175 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41176 into direct/indirect/polymorphic variants.
41178 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41180 Update copyright years.
41182 * gcc.c (process_command): Update copyright notice dates.
41183 * gcov-dump.c (print_version): Ditto.
41184 * gcov.c (print_version): Ditto.
41185 * gcov-tool.c (print_version): Ditto.
41186 * gengtype.c (create_file): Ditto.
41187 * doc/cpp.texi: Bump @copying's copyright year.
41188 * doc/cppinternals.texi: Ditto.
41189 * doc/gcc.texi: Ditto.
41190 * doc/gccint.texi: Ditto.
41191 * doc/gcov.texi: Ditto.
41192 * doc/install.texi: Ditto.
41193 * doc/invoke.texi: Ditto.
41195 Copyright (C) 2017 Free Software Foundation, Inc.
41197 Copying and distribution of this file, with or without modification,
41198 are permitted in any medium without royalty provided the copyright
41199 notice and this notice are preserved.